Adopt shared history ownership in the interface roadmap

README.md

@@ -9,7 +9,7 @@ ConfigKit's simple `Load` accepts one strict YAML source (or explicit
 defaults-only input). `LoadLayered` additionally combines explicit lower-priority
 source groups with an independently selected main input. Sources can be host
 files, native `fs.FS` files, labeled bytes or readers, or snapshotted mappings.
-Enabled path and glob imports compose strictly within their group; precedence
+Enabled explicit imports compose strictly within their group; precedence
 between groups combines mappings recursively. Both entry points return a complete,
 validated configuration with detached metadata and safe diagnostics.
 Applications may perform audited defaulting and normalization through a resolver,
@@ -17,10 +17,38 @@ then report domain errors and warnings through read-only semantic validation. Se
 tags and a path predicate drive deterministic redacted effective YAML. Typed secret
 references resolve environment names before validation, and explicit
 secrets-directory bootstrap is available as a separate startup operation. Explicit
-CLI override records apply above overlays. Value explanation is available through
+CLI override records apply after inheritance and typed decoding, before application
+resolution. Value explanation is available through
 detached structured records and deterministic redacted output. Built-in
 `extends` relationships let a mapping inherit from a named sibling during the
-same load, retaining explicit presence and source/derivation history.
+same load, retaining explicit presence and source/derivation history. The root
+`Diff` API compares frozen effective values from two successful same-type loads;
+the [configuration contract](docs/config.md#effective-configuration-diffs) and
+[consumer guide](docs/consumers/README.md#effective-configuration-diffs) describe
+its safe changed-region records and maintained workflows.
+
+Sources can also be explicitly placed as sequence items. Configuration
+declarations and application registrations can give selected sequences scalar
+identities and choose how they combine across precedence and inheritance.
+Strict peers combine distinct records, while higher-priority collections can
+replace, append, or upsert whole records. See the
+[sequence contract](docs/config.md#strict-document-composition),
+[maintained catalog examples](docs/consumers/README.md#identified-prompt-sequences),
+and [ADR 0017](docs/adr/0017-collection-sequence-policy.md).
+
+Imports are also explicit: a source may name files, search declared directories,
+or select files with path globs. Applications can set load-wide import defaults
+while keeping declaration-specific exceptions. Files are acquired in authored
+declaration order; dependency-ready composition uses a shared stable scheduler.
+ConfigKit does not search conventional application paths implicitly. See the
+[import contract](docs/config.md#explicit-imports), [consumer examples](docs/consumers/README.md),
+and [ADR 0015](docs/adr/0015-shared-dependency-ready-ordering.md) and
+[ADR 0016](docs/adr/0016-independent-import-controls.md).
+
+Main-file discovery is explicit too: applications choose and order `Search`
+descriptions and pass `SearchConfig` to the existing load APIs. See the
+[discovery contract](docs/config.md#configuration-discovery), [consumer guide](docs/consumers/README.md#explicit-main-file-discovery),
+and [ADR 0019](docs/adr/0019-explicit-configuration-discovery.md).
 
 ```go
 type Config struct {
@@ -56,20 +84,29 @@ mappings combine, while overlapping non-mapping definitions are errors. Between
 precedence tiers, mappings combine recursively, retaining omitted lower-priority
 children; other values replace lower-priority values. The main document and its
 imports form one strictly composed tier. Explicit underlay groups sit below it
-and above typed defaults; a selected overlay and explicit CLI overrides sit
-above it. The [configuration contract](docs/config.md#strict-document-composition)
-owns the detailed combination rules and input-specific constraints.
+and above typed defaults; a selected overlay applies before inheritance, while
+explicit CLI overrides apply after decoding and before application resolution.
+The [configuration contract](docs/config.md#strict-document-composition) owns
+the detailed combination rules and input-specific constraints. Declared or
+registered collections combine distinct identified records within strict tiers
+and use their selected operation at precedence boundaries; other sequences keep
+their strict-conflict and whole-value-replacement behavior.
 
 The [architecture goals](docs/policy/architecture.md#success-criteria) explain
 these priorities and how genuine new needs can justify extending the shared
-model. The [deferred scope](docs/roadmap/deferred.md) records deliberate exclusions.
+model. The [future feature list](docs/roadmap/future.md) preserves candidates for
+consideration in future releases.
 
-The [configuration contract](docs/config.md) describes implemented behavior. The
-[consumer adoption guide](docs/consumers/README.md) provides maintained simple
-native-filesystem, layered, and sibling-inheritance examples alongside the migration shapes.
-See the [development guide](docs/development.md) for repository orientation and the
-[release procedure](docs/release.md) for maintainer-only candidate and publication
-steps.
+The [configuration contract](docs/config.md) describes implemented behavior,
+including [input destinations](docs/config.md#input-destinations). The
+[consumer adoption guide](docs/consumers/README.md) provides maintained simple,
+layered, mounted-collection, and identified-prompt examples alongside the
+migration shapes. Input placement decisions are recorded in
+[ADR 0011](docs/adr/0011-input-destinations.md) and sequence placement and
+composition in [ADR 0014](docs/adr/0014-sequence-placement-and-composition.md).
+See the [development guide](docs/development.md) for repository orientation and
+the [release procedure](docs/release.md) for maintainer-only candidate and
+publication steps.
 
 ConfigKit code is licensed under the [LGPL v3.0 or later](COPYRIGHT).
 

conversion.go

@@ -43,10 +43,61 @@ func toEngineLocations(values []Location) []engine.Location {
 	return result
 }
 func fromEngineSourceInfo(value engine.SourceInfo) SourceInfo {
-	return SourceInfo{ID: value.ID, Label: value.Label, Path: value.Path, Layer: Layer(value.Layer), Kind: SourceKind(value.Kind), UnderlayIndex: value.UnderlayIndex, ImportedFrom: fromEngineLocationPointer(value.ImportedFrom)}
+	return SourceInfo{
+		ID: value.ID, Label: value.Label, Path: value.Path, Destination: fromEnginePath(value.Destination),
+		PlacementMode:         PlacementMode(engine.SourceInfoPlacementModeForFacade(value)),
+		DestinationUnresolved: engine.SourceInfoDestinationUnresolvedForFacade(value),
+		Layer:                 Layer(value.Layer), Kind: SourceKind(value.Kind), UnderlayIndex: value.UnderlayIndex,
+		ImportedFrom: fromEngineLocationPointer(value.ImportedFrom), VersionCheck: fromEngineSourceVersionCheck(value.VersionCheck),
+		ImportSelections: fromEngineImportSelections(engine.SourceInfoImportSelectionsForFacade(value)),
+	}
 }
-func toEngineSourceInfo(value SourceInfo) engine.SourceInfo {
-	return engine.SourceInfo{ID: value.ID, Label: value.Label, Path: value.Path, Layer: engine.Layer(value.Layer), Kind: engine.SourceKind(value.Kind), UnderlayIndex: value.UnderlayIndex, ImportedFrom: toEngineLocationPointer(value.ImportedFrom)}
+func fromEngineImportSelections(values []engine.ImportSelectionForFacade) []ImportSelection {
+	if values == nil {
+		return nil
+	}
+	result := make([]ImportSelection, len(values))
+	for index, value := range values {
+		result[index] = ImportSelection{
+			Path: fromEnginePath(value.Path), Location: fromEngineLocation(value.Location),
+			Selector: ImportSelector(value.Selector), Outcome: ImportSelectionOutcome(value.Outcome),
+		}
+	}
+	return result
+}
+
+func toEngineOverrides(values []Override) []engine.Override {
+	if values == nil {
+		return nil
+	}
+	result := make([]engine.Override, len(values))
+	for index, value := range values {
+		result[index] = engine.Override{
+			Path: toEnginePath(value.Path), Value: value.Value, Flag: value.Flag,
+		}
+	}
+	return result
+}
+
+func toEngineImportDefaults(value ImportDefaults) engine.ImportDefaults {
+	return engine.ImportDefaults{
+		Required:  value.Required,
+		Follow:    value.Follow,
+		Hidden:    value.Hidden,
+		Recursive: value.Recursive,
+		Pattern:   value.Pattern,
+	}
+}
+
+func fromEngineSourceVersionCheck(value *engine.SourceVersionCheck) *SourceVersionCheck {
+	if value == nil {
+		return nil
+	}
+	return &SourceVersionCheck{
+		Path:     fromEnginePath(value.Path),
+		Version:  Version{Present: value.Version.Present, Null: value.Version.Null, Text: value.Version.Text},
+		Location: fromEngineLocationPointer(value.Location), Redacted: value.Redacted,
+	}
 }
 func fromEngineOrigin(value engine.Origin) Origin {
 	result := Origin{Layer: Layer(value.Layer), Locations: fromEngineLocations(value.Locations), Operation: ChangeKind(value.Operation)}
@@ -61,19 +112,6 @@ func fromEngineOrigin(value engine.Origin) Origin {
 	}
 	return result
 }
-func toEngineOrigin(value Origin) engine.Origin {
-	result := engine.Origin{Layer: engine.Layer(value.Layer), Locations: toEngineLocations(value.Locations), Operation: engine.ChangeKind(value.Operation)}
-	if value.Inheritance != nil {
-		result.Inheritance = make([]engine.InheritanceStep, len(value.Inheritance))
-		for index, step := range value.Inheritance {
-			result.Inheritance[index] = engine.InheritanceStep{
-				From: toEnginePath(step.From), To: toEnginePath(step.To), Reference: toEnginePath(step.Reference),
-				Locations: toEngineLocations(step.Locations),
-			}
-		}
-	}
-	return result
-}
 func fromEngineDiagnostic(value engine.Diagnostic) Diagnostic {
 	return Diagnostic{Code: Code(value.Code), Severity: Severity(value.Severity), Path: fromEnginePath(value.Path), Message: value.Message, Primary: fromEngineLocationPointer(value.Primary), Related: fromEngineLocations(value.Related)}
 }

development_check_test.go

@@ -66,7 +66,7 @@ func TestDevelopmentCheckHandlesFilenamesAndStopsOnFailure(t *testing.T) {
 						t.Fatal(err)
 					}
 					cmd := exec.Command("/bin/bash", filepath.Join(root, "scripts", scriptName))
-					cmd.Env = append(os.Environ(), "PATH="+bin+string(os.PathListSeparator)+os.Getenv("PATH"), "CONFIGKIT_BASELINE_TOOLCHAIN=test-baseline", "CONFIGKIT_NEWER_TOOLCHAIN=", "CONFIGKIT_TEST_FMT_LOG="+log, "CONFIGKIT_TEST_GO_LOG="+calls)
+					cmd.Env = append(os.Environ(), "PATH="+bin+string(os.PathListSeparator)+os.Getenv("PATH"), "CONFIGKIT_TEST_FMT_LOG="+log, "CONFIGKIT_TEST_GO_LOG="+calls)
 					// Log only targets selected by the script, not inherited host settings.
 					cmd.Env = append(cmd.Env, "GOOS=", "GOARCH=")
 					if name == "failed-go-check" {
@@ -108,90 +108,3 @@ func TestDevelopmentCheckHandlesFilenamesAndStopsOnFailure(t *testing.T) {
 		})
 	}
 }
-
-func TestCheckScriptsReleaseGates(t *testing.T) {
-	script, err := os.ReadFile("scripts/release-check.sh")
-	if err != nil {
-		t.Fatal(err)
-	}
-	for _, scenario := range []string{"success", "failed-test", "missing-toolchain"} {
-		t.Run(scenario, func(t *testing.T) {
-			root := t.TempDir()
-			bin := t.TempDir()
-			for _, directory := range []string{"scripts", "nested"} {
-				if err := os.Mkdir(filepath.Join(root, directory), 0700); err != nil {
-					t.Fatal(err)
-				}
-			}
-			checker := filepath.Join(root, "scripts", "release-check.sh")
-			if err := os.WriteFile(checker, script, 0700); err != nil {
-				t.Fatal(err)
-			}
-			calls := filepath.Join(bin, "go.calls")
-			goStub := `#!/bin/sh
-printf '%s|%s|%s|%s|%s|%s\n' "$GOTOOLCHAIN" "$GOPROXY" "${CGO_ENABLED-}" "${GOOS-}" "${GOARCH-}" "$*" >> "$CONFIGKIT_TEST_GO_LOG"
-if [ "${CONFIGKIT_TEST_FAIL-}" = toolchain ] && [ "${1-}" = version ]; then exit 37; fi
-if [ "${CONFIGKIT_TEST_FAIL-}" = test ] && [ "${1-}" = test ]; then exit 37; fi
-`
-			if err := os.WriteFile(filepath.Join(bin, "go"), []byte(goStub), 0700); err != nil {
-				t.Fatal(err)
-			}
-			cmd := exec.Command("/bin/bash", checker)
-			cmd.Dir = filepath.Join(root, "nested")
-			cmd.Env = append(os.Environ(),
-				"PATH="+bin+string(os.PathListSeparator)+os.Getenv("PATH"),
-				"CONFIGKIT_BASELINE_TOOLCHAIN=test-baseline",
-				"CONFIGKIT_TEST_GO_LOG="+calls,
-				"GOOS=", "GOARCH=", "CGO_ENABLED=",
-			)
-			switch scenario {
-			case "failed-test":
-				cmd.Env = append(cmd.Env, "CONFIGKIT_TEST_FAIL=test")
-			case "missing-toolchain":
-				cmd.Env = append(cmd.Env, "CONFIGKIT_TEST_FAIL=toolchain")
-			}
-			output, err := cmd.CombinedOutput()
-			if scenario == "success" && err != nil {
-				t.Fatalf("checker: %v\n%s", err, output)
-			}
-			if scenario != "success" && err == nil {
-				t.Fatalf("checker accepted %s", scenario)
-			}
-			data, err := os.ReadFile(calls)
-			if err != nil {
-				t.Fatal(err)
-			}
-			lines := strings.Split(strings.TrimSpace(string(data)), "\n")
-			for _, line := range lines {
-				if !strings.HasPrefix(line, "test-baseline|off|") {
-					t.Fatalf("toolchain or offline setting lost: %q", line)
-				}
-			}
-			if scenario == "missing-toolchain" {
-				if len(lines) != 1 || !strings.HasSuffix(lines[0], "|version") {
-					t.Fatalf("checker ran after unavailable toolchain: %q", lines)
-				}
-				return
-			}
-			if len(lines) < 2 || lines[1] != "test-baseline|off||||test -mod=readonly -vet=off -race -v ./..." {
-				t.Fatalf("release suite command changed: %q", lines)
-			}
-			if scenario == "failed-test" {
-				if len(lines) != 2 {
-					t.Fatalf("checker continued after failure: %q", lines)
-				}
-				return
-			}
-			want := []string{
-				"test-baseline|off||||version",
-				"test-baseline|off||||test -mod=readonly -vet=off -race -v ./...",
-				"test-baseline|off||||vet -mod=readonly ./...",
-				"test-baseline|off|0|linux|amd64|build -mod=readonly ./...",
-				"test-baseline|off|0|linux|arm64|build -mod=readonly ./...",
-			}
-			if strings.Join(lines, "\n") != strings.Join(want, "\n") {
-				t.Fatalf("unexpected release gates:\n%s", data)
-			}
-		})
-	}
-}

diff.go

@@ -0,0 +1,53 @@
+package configkit
+
+import "gitea.maximumdirect.net/eric/ConfigKit/internal/engine"
+
+// Difference describes one canonical changed region between two successful
+// loads. Left is the earlier value and Right is the later value. Each snapshot
+// is detached; Present distinguishes absence from an explicit null, and
+// Redacted reports whether combined sensitivity affected that side. Mappings
+// are compared recursively, while sequences are reported as atomic values.
+// Records are ordered by canonical JSON pointer. HiddenChanged indicates that
+// a value concealed by combined sensitivity changed, without revealing which
+// concealed member changed.
+type Difference struct {
+	Path          Path
+	Left          ValueSnapshot
+	Right         ValueSnapshot
+	HiddenChanged bool
+}
+
+// Diff compares the frozen effective values of two successful loads of the
+// same configuration type. It does not reread sources, environment variables,
+// or rerun defaults or resolvers. Equal values produce no records even when
+// their sources, defaults, or resolution histories differ. Credential contents
+// are not compared: typed secret references compare by reference name only.
+//
+// Both results must be non-nil successful loads with inspection state; nil or
+// zero results return an error and no records. Load errors are returned by
+// Load, so callers must handle those before calling Diff. Snapshots are safe
+// for presentation but are not a complete reloadable configuration.
+func Diff[T any](left, right *Result[T]) ([]Difference, error) {
+	var leftInspection, rightInspection *engine.Inspection
+	if left != nil {
+		leftInspection = left.inspection
+	}
+	if right != nil {
+		rightInspection = right.inspection
+	}
+
+	values, err := engine.Diff(leftInspection, rightInspection)
+	if err != nil {
+		return nil, err
+	}
+	differences := make([]Difference, len(values))
+	for index, value := range values {
+		differences[index] = Difference{
+			Path:          fromEnginePath(value.Path),
+			Left:          fromEngineValueSnapshot(value.Left),
+			Right:         fromEngineValueSnapshot(value.Right),
+			HiddenChanged: value.HiddenChanged,
+		}
+	}
+	return differences, nil
+}

discovery.go

@@ -0,0 +1,90 @@
+package configkit
+
+import "gitea.maximumdirect.net/eric/ConfigKit/internal/engine"
+
+// Search is an immutable ordered description of possible host-file inputs.
+// It is consumed only when passed to SearchConfig and used by a load.
+type Search struct{ value engine.Search }
+
+// FilenameGroup selects filenames in order for a directory search.
+// Construct groups with NamedFilenames, GenericFilenames, or CustomFilenames.
+type FilenameGroup struct{ value engine.FilenameGroup }
+
+// SearchOptions controls whether a search must select a file.
+type SearchOptions struct {
+	// Required reports a missing-source error if the complete search is exhausted.
+	Required bool
+}
+
+// DirectorySearchOptions selects the filename groups used by SearchDirectory.
+type DirectorySearchOptions struct {
+	// Filenames concatenates groups in order. A nil slice uses Named then Generic;
+	// a non-nil empty slice selects no filenames.
+	Filenames []FilenameGroup
+}
+
+// NamedFilenames selects <application>.yml followed by <application>.yaml.
+func NamedFilenames() FilenameGroup { return FilenameGroup{value: engine.NamedFilenames()} }
+
+// GenericFilenames selects config.yml, config.yaml, main.yml, and main.yaml.
+func GenericFilenames() FilenameGroup { return FilenameGroup{value: engine.GenericFilenames()} }
+
+// CustomFilenames selects literal filename components in the supplied order.
+// Names cannot be empty or contain a directory component.
+func CustomFilenames(names ...string) FilenameGroup {
+	return FilenameGroup{value: engine.CustomFilenames(names...)}
+}
+
+// SearchDirectory describes a search within directory. Omitted options or a nil
+// Filenames slice uses Named then Generic groups. Explicitly empty Filenames
+// selects no names and does not resolve the directory base.
+func SearchDirectory(directory string, supplied ...DirectorySearchOptions) Search {
+	options := make([]engine.DirectorySearchOptions, len(supplied))
+	for index, option := range supplied {
+		if option.Filenames == nil {
+			continue
+		}
+		options[index].Filenames = make([]engine.FilenameGroup, len(option.Filenames))
+		for groupIndex, group := range option.Filenames {
+			options[index].Filenames[groupIndex] = group.value
+		}
+	}
+	return Search{value: engine.SearchDirectory(directory, options...)}
+}
+
+// Paths describes literal host-file candidates in the supplied order.
+func Paths(paths ...string) Search { return Search{value: engine.Paths(paths...)} }
+
+// Searches concatenates descriptions in order and snapshots their entries.
+func Searches(searches ...Search) Search {
+	values := make([]engine.Search, len(searches))
+	for index, search := range searches {
+		values[index] = search.value
+	}
+	return Search{value: engine.Searches(values...)}
+}
+
+// UnixSystemDirectories describes the conventional Unix system directories.
+// Loading on an unsupported platform fails if this preset is reached.
+func UnixSystemDirectories() Search { return Search{value: engine.UnixSystemDirectories()} }
+
+// UserConfigDirectory describes searches rooted at the Go user-config directory.
+func UserConfigDirectory() Search { return Search{value: engine.UserConfigDirectory()} }
+
+// HomeDirectory describes searches rooted at the Go home directory.
+func HomeDirectory() Search { return Search{value: engine.HomeDirectory()} }
+
+// WorkingDirectory describes application-named files in the current directory.
+func WorkingDirectory() Search { return Search{value: engine.WorkingDirectory()} }
+
+// SearchConfig creates a source that selects the first matching host file when
+// consumed by Load or LoadLayered. Without Required, exhausting the search uses
+// the absent-input lifecycle. Description and option errors are reported while
+// loading.
+func SearchConfig(application string, search Search, supplied ...SearchOptions) Source {
+	options := make([]engine.SearchOptions, len(supplied))
+	for index, option := range supplied {
+		options[index] = engine.SearchOptions{Required: option.Required}
+	}
+	return Source{value: engine.SearchConfig(application, search.value, options...)}
+}

docs/adr/0005-explicit-overrides-and-safe-explanation.md

@@ -1,6 +1,8 @@
 # ADR 0005: Record explicit CLI overrides and safe value explanation
 
-Status: accepted
+Status: Partially superseded by
+[ADR 0018](0018-cli-after-structural-resolution.md) for override-before-decoding
+ordering. The remaining ownership and safe explanation decisions remain accepted.
 
 Date: 2026-09-06
 

docs/adr/0007-source-backed-underlays.md

@@ -2,7 +2,9 @@
 
 ## Status
 
-Proposed draft for review. Completed planning and remediation records are
+Proposed draft for review; its whole-sequence precedence replacement rule is
+superseded by [ADR 0017](0017-collection-sequence-policy.md). The remainder
+retains its proposed status. Completed planning and remediation records are
 preserved in Git history at `b604aa1` under `docs/roadmap/`; the
 [configuration contract](../config.md#layered-loading) owns current behavior.
 

docs/adr/0009-sibling-inheritance.md

@@ -2,8 +2,14 @@
 
 ## Status
 
-Accepted and implemented. The [configuration contract](../config.md#sibling-inheritance)
-owns current behavior. The completed implementation plan is preserved in Git
+Partially superseded by [ADR 0017](0017-collection-sequence-policy.md) for
+sequence combination and the addition of identified-sequence sibling scopes,
+and [ADR 0018](0018-cli-after-structural-resolution.md) for CLI-before-inheritance
+ordering. Existing mapping-sibling scopes and the remaining ownership and
+inspection decisions remain accepted.
+
+The [configuration contract](../config.md#sibling-inheritance) owns implemented
+behavior. The completed implementation plan is preserved in Git
 history at `83c59ea`.
 
 ## Date

docs/adr/0010-effective-configuration-diff.md

@@ -0,0 +1,99 @@
+# ADR 0010: Compare Frozen Effective Configuration Snapshots
+
+## Status
+
+Accepted and implemented. The
+[configuration contract](../config.md#effective-configuration-diffs) owns current
+behavior. Completed planning and remediation records are preserved in Git
+history at `1c2610c` under `docs/roadmap/`.
+
+## Date
+
+2026-10-09
+
+## Context
+
+Applications need to inspect how two completed configurations differ, including
+the effective impact of an overlay after defaults and application resolution.
+ConfigKit already retains canonical typed inspection snapshots, semantic paths,
+presence information, and sensitivity classification for successful loads.
+
+Extending this inspection model lets applications share configuration comparison
+and safe disclosure while keeping their presentation and domain interpretation.
+The operation should follow ConfigKit's existing equality, ownership, and secret
+boundaries so consumers can reason about loading and inspection consistently.
+
+## Decision
+
+Provide structured effective-value comparison between successful results of the
+same configuration type. Compare their retained canonical inspection snapshots
+through the root-owned public facade. Each comparison uses the completed load
+state and returns detached records with a semantic path and safe left and right
+value snapshots.
+
+Use canonical value equality and the existing structural comparison rules.
+Mappings compare recursively, sequences report atomic changes, and scalar or
+kind replacements report the affected region. Presence distinguishes additions
+and removals from explicit null. Equal effective values produce an empty diff;
+source attribution, explicit-input status, and history remain separately
+inspectable through each result. Canonical pointer ordering makes reporting
+deterministic.
+
+Detect changes before masking values. Apply sensitivity from both loads to both
+returned sides, including historical and dependency-derived classification.
+Preserve public content where the shapes permit corresponding sensitive regions
+to be identified. When a structural replacement prevents safe correspondence,
+mask the complete reported replacement region on both present sides. Absence
+retains its own representation.
+
+Keep masking and hidden-change information distinct. Each record indicates
+whether an actual change is withheld by redaction, allowing applications to
+display a changed-sensitive-value indicator even when safe snapshots render
+identically. Public-only changes alongside unchanged masked content retain that
+distinction.
+
+Typed secret references participate through their configured environment names.
+Resolved credential bytes remain behind the existing explicit secret accessor.
+The comparison uses the same frozen inspection boundary as effective rendering
+and explanation.
+
+ConfigKit owns comparison and paired safe disclosure. Applications own labels,
+layout, and rendering of the returned changed regions. Reusing the existing
+snapshot representation and engine boundary keeps the extension consistent with
+the library's established inspection model.
+
+## Alternatives considered
+
+- **Compare input text or separately rendered YAML.** Supports familiar textual
+  displays, but input text precedes composition and resolution, and independently
+  redacted output can hide changes or apply different disclosure rules. Canonical
+  comparison with paired masking represents the completed configuration.
+- **Leave comparison to applications using returned Go values.** Offers direct
+  access to application types, but duplicates structural and redaction machinery
+  and observes caller mutation. Retained inspection snapshots preserve the
+  completed load's values and established ownership boundary.
+- **Align sequence elements and return context-rich rendering data.** Enables
+  finer-grained displays, but requires additional correspondence and disclosure
+  rules. Atomic sequence records and existing value snapshots satisfy the
+  selected changed-region workflows with a smaller public model.
+- **Compare resolved credentials or retain credential-derived tokens.** Could
+  detect rotation behind an unchanged reference, but expands retained sensitive
+  state and introduces a separate trusted comparison mechanism. Reference-name
+  comparison preserves the existing inspection boundary.
+
+## Consequences
+
+Applications can compare independently loaded configurations using ConfigKit's
+shared equality, presence, and disclosure rules. Existing composition and load
+semantics continue to determine the values being compared. Caller mutation of a
+returned configuration leaves its retained comparison snapshot unchanged.
+
+Paired redaction can conceal public content within a structurally incompatible
+replacement. The changed region and hidden-change indicator still communicate
+its effective impact. Credential rotation behind an unchanged environment name
+produces equal reference values for comparison.
+
+The implementation requires operation-local disclosure preparation and focused
+coverage for presence, shape changes, historical sensitivity, and detached
+results. Applications retain control of presentation; changed-region records
+supply the safe data needed for that responsibility.

docs/adr/0011-input-destinations.md

@@ -0,0 +1,74 @@
+# ADR 0011: Place Configuration Inputs at Mapping Destinations
+
+## Status
+
+Accepted and implemented. Current placement behavior is owned by the
+[configuration contract](../config.md#input-destinations). The accepted
+relative-import extension is recorded in
+[ADR 0012](0012-relative-import-destinations.md). The completed roadmap, consumer
+request, and review records are preserved at `e4270c9` under `docs/roadmap/` in
+Git history; the original input-destination implementation plan remains at
+`315f464:docs/roadmap/implementation.md`.
+
+## Date
+
+2026-10-14
+
+## Context
+
+Consumers assembling collections of independently authored configuration
+fragments need each source to contribute beneath a selected application path.
+Requiring consumers to rewrite source documents or merge completed values would
+separate placement from ConfigKit's existing composition, precedence, source
+identity, and inspection behavior. Placement should remain part of the ordinary
+loading lifecycle and preserve the root-owned public facade.
+
+## Decision
+
+Add an optional `Destination Path` to root-owned `Input` and expose the effective
+destination on existing `SourceInfo` records. The empty path means the existing
+application root. A non-empty destination is a sequence of literal mapping keys;
+explicit sequence-index segments are invalid. Existing reserved structural keys
+remain reserved, while keys that merely contain their spelling remain ordinary
+keys.
+
+Acquire and interpret each source using its original physical identity and
+source-local directives, then place its application body before strict
+composition and the existing final lifecycle. Imported application bodies
+inherit the input destination. Placement applies to main and underlay inputs;
+ordinary root loading, selected overlays, and CLI inputs retain root placement.
+The main-owned version rule continues to check its effective application path
+at its existing lifecycle point.
+
+Placement preserves authored node locations and source identity. Introduced
+mapping structure belongs to the contributing source and uses that document's
+root location as its source-coordinate fallback. A present empty mapping
+materializes its destination; an absent `NoSource` contributes nothing. The
+result continues through the existing strict composition, overlay, inheritance,
+typed loading, diagnostics, provenance, history, sensitivity, and resource
+context mechanisms.
+
+The added exported fields preserve keyed and zero-value struct use. Callers
+using unkeyed composite literals for `Input` or `SourceInfo` must migrate to
+keyed literals. Treat this feature as minor-release work; select the version and
+document the compatibility impact during separately authorized release
+preparation, consistent with the release policy.
+
+## Alternatives considered
+
+- **Rewrite or merge sources in consumers.** This makes each consumer reproduce
+  placement, strict composition, source attribution, and inspection semantics.
+- **Add a separate loading entry point or public wrapper type.** This creates a
+  parallel API for a behavior that belongs to existing per-input loading.
+- **Load each fragment as a complete application before placement.** This runs
+  defaults, decoding, resolution, and validation on incomplete fragments rather
+  than once on the assembled application.
+
+## Consequences
+
+Consumers can assemble contributions at effective mapping paths in one ordinary
+layered load and inspect each source's placement through existing source
+records. Paths, conflicts, and later application operations use the placed
+structure while source coordinates and relative-resource contexts continue to
+identify their original sources. Public field additions require release-note
+guidance for callers using unkeyed literals.

docs/adr/0012-relative-import-destinations.md

@@ -0,0 +1,78 @@
+# ADR 0012: Place Imported Configuration at Relative Destinations
+
+## Status
+
+Accepted and implemented. The
+[configuration contract](../config.md#input-destinations) owns current behavior.
+The completed roadmap and review records are preserved at `e4270c9` under
+`docs/roadmap/` in Git history; the relative-import implementation plan remains
+at `d9f08ff:docs/roadmap/implementation.md`.
+
+## Date
+
+2026-10-14
+
+## Context
+
+ConfigKit can place an input document at a literal mapping path while retaining
+source identity, composition, and inspection in its normal loading lifecycle.
+An importing document can currently bring in other documents, but those
+contributions inherit only the importing input's destination. Consumers that
+assemble a collection therefore cannot declare the layout of imported
+fragments alongside the import itself without rewriting source documents or
+performing another merge outside ConfigKit.
+
+The extension should reuse input placement and the existing import lifecycle.
+Logical application placement must remain independent from physical path
+resolution, source identity, precedence, and resource ownership.
+
+## Decision
+
+Add an optional `destination` field to each YAML import declaration. Its value
+is a sequence of strings representing literal mapping-key segments. Omission
+and an empty sequence add no segments. A non-empty destination appends to the
+effective destination of the immediately importing document; nested imports
+continue from the resulting child destination. Every match of one glob
+declaration receives the same computed destination.
+
+Use the existing input-destination validation and reserved-key rules. Do not
+interpret destination segments as filesystem paths, separators, sequence
+indices, or root-reset instructions. Physical import paths continue to resolve
+from their declaring source under existing containment rules.
+
+Imports remain in their existing strict-composition group. A destination does
+not grant precedence, bypass cycle or repeated-source rules, or change source
+identity. Existing placement, version ownership, source records, diagnostics,
+resource contexts, inheritance, history, and inspection mechanisms continue to
+govern the placed contributions. No public Go entry point or independent
+specification-header behavior is introduced.
+
+This is additive YAML functionality and is minor-release work. Version
+selection, release notes, roadmap retirement, and publication remain separate
+release-preparation work.
+
+## Alternatives Considered
+
+- **Require complete application-root destinations on every import.** This
+  avoids inherited context but repeats collection prefixes in nested source
+  declarations and does not express placement relative to the importing
+  document.
+- **Encode destinations as dotted or slash-delimited strings.** This introduces
+  separator and escaping rules even though existing paths already represent
+  literal mapping keys as segments.
+- **Rewrite or merge imported documents in consumers.** This duplicates source
+  placement, strict composition, attribution, and inspection behavior outside
+  ConfigKit's loading lifecycle.
+- **Add independent specification headers or collection-specific validation.**
+  These are separate design questions and are not required to declare relative
+  placement.
+
+## Consequences
+
+Import declarations can describe the logical placement of their contributions
+while preserving the physical source and existing composition lifecycle.
+Placement is immediate-parent-relative, so nested imports accumulate their
+declared segments and sibling declarations remain independent. The feature is
+implemented behind the existing public loading surface; detailed validation,
+diagnostic, accounting, and inspection requirements are tracked by the
+implementation roadmap.

docs/adr/0013-shared-source-version-checks.md

@@ -0,0 +1,92 @@
+# ADR 0013: Check Source Versions with One Shared Policy
+
+## Status
+
+Accepted and implemented. The
+[configuration contract](../config.md#shared-source-version-checks) owns current
+behavior. The completed roadmap, implementation plan, consumer request, and
+resolved review findings are preserved at `e4270c9` under `docs/roadmap/` in
+Git history; the [development guide](../development.md#baseline-validation)
+provides retrieval examples.
+
+## Date
+
+2026-10-15
+
+## Context
+
+Applications that assemble one configuration from independently authored
+documents need a way to require that each contributing document declares a
+version accepted by the application's existing `Definition.Version` checker.
+Checking only the eventual composed value lets another source hide a rejected
+declaration, while separate consumer-side parsing duplicates ConfigKit's source
+handling and diagnostics.
+
+The extension must preserve one application-owned acceptance policy and the
+existing loading lifecycle. Source declarations are compatibility evidence, not
+independent fragment schemas or values to compare for equality. Existing
+composition, placement, precedence, source identity, diagnostics, and
+sensitivity rules remain authoritative.
+
+## Decision
+
+Provide a load-wide source-version mode for ordinary and layered loading. The
+zero-value mode preserves current behavior. The enabled modes check declarations
+when present or require them, using the existing `Definition.Version` path and
+checker. Interpret the configured path relative to each participating source
+before placement or composition. Different declarations may be accepted by the
+same checker; ConfigKit adds no equality, SemVer, or expected-version rule.
+
+Apply the selected policy to the actual main document, direct underlays,
+imports, and selected overlays. Missing non-main declarations are skipped in
+if-present mode and rejected in required mode. Existing main absence behavior
+remains, with required mode additionally rejecting an absent declaration on an
+actual main document. `NoSource`, unmatched optional inputs, synthetic field
+operations, and empty groups do not become documents that need declarations.
+
+Reuse the existing version vocabulary and validation boundary: absence, null,
+exact strings, and strict decimal integers retain their established meanings;
+unsupported present values fail version-type validation. Warnings remain
+nonfatal, and source declarations do not replace final application decoding or
+domain validation.
+
+After an accepted non-main declaration is checked, consume exactly that node
+before its remaining application content is composed or overlaid. Preserve the
+main declaration and its value semantics. Keep the declaration, source-local
+path, physical location, and checked absence as detached source evidence,
+separate from effective application values and history. Apply established
+sensitivity rules to that evidence, including original mounted paths and
+ancestors; retain existing safe failure disclosure.
+
+In enabled modes, the main declaration remains source-local while its protected
+application path is derived from the main input's destination. This distinction
+does not change default-mode path interpretation or weaken protection of
+remaining payload and synthetic operations.
+
+Public option-field additions and the enabled-mode interpretation for mounted
+main inputs are compatibility considerations for separately authorized minor-
+release preparation. This decision does not select a release version or
+authorize release notes or publication.
+
+## Alternatives Considered
+
+- **Check only the final composed value.** Later sources can hide an earlier
+  rejected declaration, so this does not validate every contributing document.
+- **Let consumers parse and check each source.** This duplicates acquisition,
+  source attribution, safe diagnostics, and lifecycle coordination in every
+  application.
+- **Require declarations to be equal or impose a universal version grammar.**
+  Consumers already own the acceptance policy; ConfigKit need not invent an
+  expected value or a domain-specific compatibility rule.
+- **Keep accepted non-main declarations in application payload.** A selected
+  source-local declaration is metadata for this policy and cannot also remain
+  an ordinary field at the same path without ambiguity.
+
+## Consequences
+
+Consumers can apply one existing version policy across contributing documents
+and select gradual or required adoption per load. Accepted non-main headers no
+longer participate in application composition; their inspectable evidence is
+kept separate and subject to sensitivity classification. Default callers retain
+current behavior. The implementation plan records the concrete API names,
+integration order, and validation evidence.

docs/adr/0014-sequence-placement-and-composition.md

@@ -0,0 +1,81 @@
+# ADR 0014: Place and Compose Sequence Items Explicitly
+
+## Status
+
+Partially superseded by [ADR 0017](0017-collection-sequence-policy.md) for
+identity ownership and sequence combination. Explicit source placement remains
+accepted.
+
+The [configuration contract](../config.md#strict-document-composition)
+defines implemented behavior, and the [consumer guide](../consumers/README.md#identified-prompt-sequences)
+shows the maintained example. The completed roadmap, implementation plan, and
+resolved remediation record are preserved in Git history at `008d2ba` under
+`docs/roadmap/`; see the [development guide](../development.md) for retrieval
+commands.
+
+## Date
+
+2026-10-16
+
+## Context
+
+ConfigKit strictly composes independent documents by recursively combining
+mappings and rejecting other overlapping values. This makes two independently
+authored sequence fragments conflict, even when their records are distinct.
+Precedence assembly has a different purpose: a higher-priority present sequence
+replaces the lower-priority sequence as one value. Those existing rules are
+predictable, but they do not express the separate operation of collecting
+independent records while rejecting duplicate identities.
+
+Likewise, mapping destinations place a source beneath literal mapping keys, but
+there is no explicit way to state that a source mapping represents one item in a
+sequence. Inferring that intent from document shape or fields would make source
+meaning implicit and change existing placement behavior.
+
+## Decision
+
+Keep mapping placement as the zero-value behavior and add an explicit item
+placement choice for a source whose mapping should become one element at a
+non-root sequence destination. Item placement provides structure only; it does
+not infer or synthesize identity values.
+
+Allow a definition to register literal mapping paths and one literal identity
+field for sequences at those paths. Within each strict composition tier, a
+registered sequence accepts mapping records with a present string identity,
+appends distinct records in contribution order, and rejects repeated exact
+identity strings. Identical registrations coalesce; conflicting identity fields
+for the same path are invalid. Registration does not recursively select paths
+inside sequence records.
+
+Keep precedence assembly unchanged. A present higher-priority sequence replaces
+the lower-priority value wholesale, including at a registered path. Registration
+changes strict composition only; it does not turn precedence into append or
+identity-based merging.
+
+This decision preserves ConfigKit's distinction between independent
+same-priority contributions and intentional precedence. Explicit source
+placement and definition-owned composition policy make the two required choices
+visible while keeping the existing zero-value load behavior.
+
+## Alternatives Considered
+
+- **Infer item placement from a source mapping or identity field.** Shape-based
+  inference would make ordinary mapping placement depend on document contents
+  and would leave the intended destination ambiguous.
+- **Append all overlapping sequences during strict composition.** This would
+  change existing conflict semantics at unregistered paths and would not detect
+  duplicate records.
+- **Merge matching records recursively by identity.** This would add a second
+  recursive merge rule with different conflict and precedence behavior from
+  ConfigKit's existing mapping composition and overlay operations.
+- **Use identity to merge sequences across precedence tiers.** This would
+  change established whole-value replacement semantics and conflate strict
+  composition with precedence.
+
+## Consequences
+
+Consumers can explicitly place mappings as sequence items and opt selected
+sequences into duplicate-rejecting strict composition. Unregistered sequences
+retain their current strict-conflict behavior, and higher-priority sequences
+continue to replace lower-priority ones. The implementation plan owns the exact
+API, YAML syntax, diagnostics, and implementation status.

docs/adr/0015-shared-dependency-ready-ordering.md

@@ -0,0 +1,68 @@
+# ADR 0015: Share Dependency-Ready Ordering Across Imports and Inheritance
+
+## Status
+
+Accepted. Implemented behavior is described in the
+[configuration contract](../config.md#explicit-imports) and
+[internal import overview](../internal/imports.md).
+
+## Date
+
+2026-10-19
+
+## Context
+
+Imports and sibling inheritance both resolve relationships between configuration
+contributions. Their existing resolvers use different traversal rules, making
+ordering depend on which mechanism expresses a dependency. Ordering affects
+sequence contributions, provenance, and diagnostics, so a consistent rule is
+part of ConfigKit's configuration model.
+
+The resolvers also serve distinct purposes. Imports strictly compose sources;
+inheritance overlays a child on its resolved parent after source assembly.
+Item-local contributions and nested inheritance introduce scope boundaries that
+resolution must preserve.
+
+## Decision
+
+ConfigKit owns one shared dependency-ready scheduler for imports and sibling
+inheritance. Both resolvers use Kahn's algorithm with a stable ordering key,
+selecting the smallest currently eligible node after its dependencies have
+completed. The shared scheduler owns eligibility, deterministic tie-breaking,
+and dependency-cycle identification. Each resolver owns construction of its
+dependency relationships and application of its domain operations.
+
+Apply the common scheduling rule within the scopes where dependencies are
+available. Complete item-local contributions before item placement and resolve
+enclosing inheritance before relationships in the resulting nested scopes.
+Source discovery supplies the import graph; the scheduler determines resolution
+order.
+
+Strict composition, inheritance overlay, and precedence tiers retain their
+existing ownership and meanings under [ADR 0001](0001-separate-composition-and-overlay.md).
+The lifecycle and scope decisions in
+[ADR 0009](0009-sibling-inheritance.md) continue to govern inheritance. Import
+admission retains its cycle and repeated-source rules.
+
+## Alternatives Considered
+
+- **Deterministic depth-first postorder.** Keeps dependency branches together,
+  but gives traversal branch selection an additional role in ordering.
+- **Dependency rounds.** Groups nodes by readiness rounds, but delays newly
+  eligible nodes until the current round finishes.
+- **Descending longest-path depth.** Prioritizes the furthest dependencies,
+  but adds distance from the root as an ordering policy.
+- **Separate resolver algorithms.** Allows local specialization, but leaves
+  consumers and maintainers with different rules for analogous dependencies.
+
+## Consequences
+
+Both resolvers follow one deterministic ordering model and share responsibility
+for maintaining that model. Resolver-specific scopes, combination semantics,
+and source-admission rules remain explicit at their integration boundaries.
+
+Independent branches may interleave as nodes become eligible. Adopting this
+rule changes existing traversal-dependent sequence positions and reporting
+order, requiring coordinated review of source assembly, provenance, and
+inheritance behavior. The configuration contract owns the implemented ordering
+details.

docs/adr/0016-independent-import-controls.md

@@ -0,0 +1,83 @@
+# ADR 0016: Separate Import Selection, Traversal, and Presence Requirements
+
+## Status
+
+Accepted. Implemented behavior is described in the
+[configuration contract](../config.md#explicit-imports), the
+[public API declarations](../../lifecycle.go), and the
+[internal import overview](../internal/imports.md).
+
+## Date
+
+2026-10-19
+
+## Context
+
+Consumers assemble configurations from named files and collections whose layout
+varies between installations. Simple imports should require little setup, while
+larger collections need control over discovery, recursion, filtering, and
+required contributions. Combining those choices implicitly makes declarations
+harder to interpret and pushes source-selection work into applications.
+
+Discovery can also reach a candidate through overlapping searches or encounter
+different names for one canonical source. Selection and source admission need
+distinct identities to support useful search unions while preserving ConfigKit's
+existing source-integrity rules.
+
+## Decision
+
+ConfigKit owns import discovery and selection within its source-loading
+lifecycle. Support explicit files, directory searches, and path globs as
+distinct selection forms. Keep traversal, filename filtering, and presence
+requirements independently expressible. Discovery determines the selected
+contributions; the shared scheduler in
+[ADR 0015](0015-shared-dependency-ready-ordering.md) determines their resolution
+order.
+
+Use optional-by-default imports with explicit requiredness controls. Selected
+sources remain subject to ConfigKit's validity, containment, resource, and
+composition contracts. This applies the library's
+[simplicity and defaults principles](../policy/architecture.md#simplicity-defaults-and-explicit-controls)
+to import absence while retaining meaningful errors for invalid contributions.
+
+Applications own their selected import defaults. Authored declaration values
+take precedence over application defaults, which take precedence over library
+defaults. Each declaration resolves applicable controls independently,
+including declarations in imported sources. This gives applications a shared
+baseline while preserving explicit choices in individual declarations.
+
+Keep declaration-local candidate identity separate from canonical source
+identity. Search unions coalesce repeated candidate paths within a source
+namespace before symlink resolution; filename filtering uses the selected or
+discovered name. Source admission continues to use canonical identity and the
+existing repeated-source rules. Distinct aliases retain their separate
+inclusion routes.
+
+## Alternatives Considered
+
+- **Use path globs for all discovery.** Offers one selection form, but makes
+  collection traversal depend on path-pattern structure and complicates
+  independently controlling recursion and filename filtering.
+- **Offer only explicit files and directory searches.** Reduces the selector
+  vocabulary, but moves wildcard directory selection into repeated declarations
+  or application-side discovery.
+- **Require every import by default.** Detects absent contributions immediately,
+  but requires repeated opt-outs for installations with optional components.
+- **Make application settings override authored declarations.** Provides central
+  enforcement, but prevents declarations from expressing local exceptions.
+- **Deduplicate selections globally by canonical source identity.** Coalesces
+  aliases and separate inclusion routes, but changes the existing repeated-source
+  contract and obscures independently authored contributions.
+
+## Consequences
+
+Consumers can express simple imports and more selective collection searches
+through the same loading lifecycle. Applications choose defaults and validate
+their required effective configuration; ConfigKit owns consistent selection,
+source admission, composition, and inspection.
+
+Optional absence can conceal misspelled selectors, so requiredness and
+inspection provide explicit ways to detect and explain missing contributions.
+Declaration-local deduplication requires the loader to preserve candidate
+identity alongside canonical source identity. The configuration and public API
+contracts own the implemented behavior.

docs/adr/0017-collection-sequence-policy.md

@@ -0,0 +1,143 @@
+# ADR 0017: Define Identity and Merge Rules for Each Collection
+
+## Status
+
+Accepted.
+
+This supersedes the decisions about who declares identity and how
+sequences combine in [ADR 0014](0014-sequence-placement-and-composition.md),
+and the rule that inheritance always replaces sequences in
+[ADR 0009](0009-sibling-inheritance.md). It also extends that ADR's sibling
+scopes to identified sequences. Existing mapping-sibling scopes and source
+placement continue to apply. It supersedes the whole-sequence
+precedence replacement rule in the proposed
+[ADR 0007](0007-source-backed-underlays.md).
+
+## Date
+
+2026-10-22
+
+## Context
+
+ConfigKit can load separate files as records in a sequence. Applications can
+register an identity key so records with different identities combine and
+duplicates fail. However, a higher configuration layer or an inheriting mapping
+currently replaces the whole sequence. Applications that need to extend a
+catalog across layers must write their own merging code and keep source and
+history information consistent with its results.
+
+Configuration authors need to choose how records are identified and whether a
+layer adds records, updates records, or replaces the collection. Applications
+also need a way to require unique identities in the final loaded configuration.
+These rules must work consistently for imported records, inline sequences,
+overlays, and inheritance.
+
+Records in a sequence may also share settings. Authors need to reuse those
+settings through inheritance while keeping each record's own identity.
+
+## Decision
+
+ConfigKit manages identity and merge rules for each collection during loading.
+Configuration files can declare those rules for imported and inline sequences,
+including sequences in selected overlays. Applications can require a particular
+identity definition, and configuration declarations must agree with it. Source
+placement determines where each contribution enters the configuration.
+
+Keep the distinction between contributions at the same priority and overrides
+from a higher layer, as established in
+[ADR 0001](0001-separate-composition-and-overlay.md). Within one tier, records
+with distinct identities combine; duplicate identities and conflicting
+declarations fail. Determine the applicable rules before combining those
+contributions, including any identity established in lower tiers.
+
+Across layers, support three operations: replace the whole sequence, append
+records while rejecting duplicate identities, or replace whole records by
+identity (`upsert`). Source layering and sibling inheritance use the same
+operations. For inheritance, the rules at the inheriting destination determine
+which operation applies.
+
+An identified sequence is also a sibling scope for inheritance. A record may
+inherit from one other record in the same sequence, selected by its complete
+identity. Each record supplies its own identity before inheritance. The child
+keeps that identity and combines its settings with the parent's other fields,
+using the existing mapping and collection rules. Parents remain ordinary records
+subject to normal validation.
+
+Resolve these relationships after source assembly, using the same dependency
+ordering as mapping-sibling inheritance. Resolve nested relationships at each
+inheriting destination. Preserve the assembled sequence order. Whole-record
+upsert keeps its existing meaning; inheritance then uses the records that remain.
+Typed defaults, CLI overrides, and application resolution follow the lifecycle
+in [ADR 0018](0018-cli-after-structural-resolution.md).
+
+An explicitly selected operation remains in effect at its destination until a
+higher layer changes it. When no explicit operation applies, use upsert if an
+identity definition is available; otherwise replace the whole sequence. Authors
+can declare identity and an operation before supplying any records. Keep source
+acquisition order separate from layer priority and the dependency ordering
+established by [ADR 0015](0015-shared-dependency-ready-ordering.md).
+
+Rules declared inside a record belong to that record. Retaining the record
+retains its local rules; replacing it discards them. Replacing a whole collection
+can discard its previous YAML identity definition, while application requirements
+still apply. Inheritance can copy collection identity, but the inheriting
+destination chooses its operation. Application registrations continue to govern
+the paths where they were registered. The same ownership rules apply to
+declarations made before a sequence exists.
+
+Check YAML-declared identity when combining collections during source assembly
+and inheritance. Application registrations also require the final loaded
+sequences to satisfy their identity rules, including after typed defaults, CLI
+overrides, and application resolution. This guarantee checks sequences that
+exist; the application schema and validation decide whether a collection must
+exist and which value types are allowed. The guarantee applies when loading
+completes, rather than to every intermediate callback state or changes made by
+the caller afterward.
+
+Preserve explicit presence, source attribution, history, sensitivity, and
+resource context through these operations, following
+[ADR 0003](0003-preserve-presence-and-resolution-history.md). Typed defaults,
+CLI assignments, and application resolution keep their own rules for changing
+values. The public API provides registration and inspection; the implementation
+chooses how to store and track collection rules internally.
+
+## Alternatives Considered
+
+- **Keep identity application-owned and always replace sequences across layers.**
+  Keeps the current design small, but reusable catalogs require application
+  changes or custom merging code.
+- **Put merge rules on each import.** Keeps the rules beside the imported source,
+  but inline sequences and overlays need a separate way to declare them. It also
+  makes rules for a collection spread across several imports harder to reconcile.
+- **Merge individual fields of records with matching identities.** Allows partial
+  record updates, but adds rules for field conflicts and retained values.
+  Replacing the whole record makes the update's scope clear.
+- **Keep inheritance limited to mapping siblings.** Keeps parent lookup simpler,
+  but authors of sequence records must duplicate shared settings or applications
+  must implement inheritance themselves. Collection identity provides a way to
+  select a parent within the existing loading process.
+- **Check every YAML identity again at the end of loading.** Gives YAML declarations
+  authority over values changed by defaults, CLI assignments, and application
+  resolution. This extends the lifetime of those declarations; application
+  registration provides that final guarantee when needed.
+
+## Consequences
+
+Consumers can build and customize catalogs in one load while retaining useful
+inspection information. Configuration authors choose how collections combine
+within application requirements. Duplicate identities among contributions at
+the same priority remain errors.
+
+Existing registered collections use upsert across layers by default unless an
+explicit operation selects otherwise. Correct loading now requires tracking
+which declarations still apply as records are replaced, copied, or moved to new
+positions. Lower-tier identity must be available before combining higher-tier
+contributions.
+
+Record inheritance also requires parent lookup by identity and checking for
+missing parents and cycles. Copied settings retain the same source, history,
+and sensitivity information as mapping-sibling inheritance.
+
+The [configuration contract](../config.md#strict-document-composition) defines
+declaration syntax, identity comparison, omission and empty-value behavior, and
+diagnostics for the implemented feature.

docs/adr/0018-cli-after-structural-resolution.md

@@ -0,0 +1,84 @@
+# ADR 0018: Apply CLI Overrides After Inheritance and Decoding
+
+## Status
+
+Accepted.
+
+This supersedes the rule that CLI overrides apply before decoding in
+[ADR 0005](0005-explicit-overrides-and-safe-explanation.md) and the
+rule that they apply before inheritance in [ADR 0009](0009-sibling-inheritance.md).
+Their remaining ownership, inheritance scope, and safe inspection decisions
+continue to apply.
+
+## Date
+
+2026-10-22
+
+## Context
+
+Explicit command-line values are the operator's highest-priority input. If they
+apply before inheritance, a sequence supplied as a replacement could gain
+inherited records afterward. Its final contents would then differ from the
+operator's assignment.
+
+The application also needs these values when it fills missing settings,
+normalizes values, or computes derived settings. ConfigKit needs a clear loading
+order that preserves both CLI precedence and the application's role in resolving
+the final configuration.
+
+## Decision
+
+ConfigKit uses this loading order:
+
+1. Combine sources within tiers and apply precedence between layers.
+2. Resolve sibling inheritance, apply typed defaults, and decode the configuration
+   into the application's types.
+3. Apply explicit CLI overrides and check them against the application schema.
+4. Run application resolution using the overridden configuration.
+5. Look up effective secret references, validate the final configuration, and
+   return it.
+
+CLI overrides are the highest-priority input to application resolution. A CLI
+sequence replaces the whole sequence; mappings combine recursively. Collection
+merge rules apply during the earlier source and inheritance steps. Inheritance
+and typed defaulting do not run again after CLI overrides.
+
+Keep track of which values were explicitly supplied and which CLI input supplied
+them. Applications continue to parse command-line arguments; ConfigKit applies
+and validates the overrides and makes their effects inspectable. Reject
+structural directives in override paths and values, so instructions such as
+inheritance relationships remain part of source processing.
+
+Application resolution keeps its existing rules for defaults and recorded
+changes under [ADR 0003](0003-preserve-presence-and-resolution-history.md).
+It receives CLI values together with their presence and source information.
+Defaults that only fill missing settings preserve explicit CLI values.
+Application normalization and derived-value calculations can still change them,
+with those changes recorded. Secret lookup and final validation use the resulting
+configuration.
+
+## Alternatives Considered
+
+- **Apply CLI overrides before inheritance.** Preserves existing ordering, but
+  later merging needs special handling to preserve CLI replacement intent while
+  still combining ordinary source contributions.
+- **Apply CLI overrides after inheritance but before typed decoding.** Allows an
+  override to fix a source value that would otherwise fail decoding, but leaves
+  typed defaulting after the operator's assignment. Completing defaulting and
+  decoding first gives CLI overrides an already resolved configuration to modify.
+- **Apply CLI overrides after application resolution.** Applies assignments after
+  the application's changes, but derived settings still reflect earlier values
+  unless resolution runs again.
+
+## Consequences
+
+CLI overrides change a configuration whose inheritance and typed defaults are
+already resolved. The application can then derive settings from the operator's
+values. The final values, source information, history, and sensitivity handling
+must remain consistent throughout loading.
+
+CLI input can no longer add or cancel inheritance relationships or bypass an
+earlier error in source assembly or inheritance. It also cannot fix a value that
+fails the earlier decoding step. Semantic validation still runs after overrides
+and application resolution. The [configuration contract](../config.md#explicit-cli-overrides)
+and [consumer guidance](../consumers/README.md) describe this behavior.

docs/adr/0019-explicit-configuration-discovery.md

@@ -0,0 +1,68 @@
+# ADR 0019: Keep Configuration Discovery Explicit and Inside Loading
+
+## Status
+
+Accepted.
+
+## Date
+
+2026-10-24
+
+## Context
+
+Applications often need a small ordered convention for choosing a main
+configuration file. Keeping that choice entirely in each application repeats
+path selection and file-loading plumbing. Moving the convention into ConfigKit
+without an explicit declaration would make ordinary loads search paths that a
+consumer did not choose, obscuring both precedence and filesystem access.
+
+ConfigKit already owns synchronous source acquisition, parsing, composition,
+decoding, validation, and inspection. A selected host file can therefore enter
+that lifecycle as the existing file source, preserving its source identity,
+resource context, and diagnostics. Optional exhaustion can use the same absent
+main lifecycle as an explicitly absent source.
+
+## Decision
+
+Consumers choose and order the main-file conventions they want to use. They may
+describe literal paths, directory filename groups, and supported directory
+presets, and may mark the complete search as required. Search descriptions are
+explicit inputs to loading; no preset or convention is activated implicitly.
+
+During synchronous loading, ConfigKit resolves the description to one real host
+file or to absence before parsing. The selected file follows the existing source
+acquisition and processing rules. A selected-file access, parse, version,
+composition, decode, or validation failure is terminal and does not resume the
+search. Optional exhaustion shares the absent-main lifecycle; required
+exhaustion reports a missing-source diagnostic. No candidate-attempt history is
+added to public metadata.
+
+This keeps convention choice with applications while moving repeated selection
+and loading plumbing into the library's established lifecycle. It preserves
+explicit control, ordinary source behavior, and complete-on-return ownership.
+
+## Alternatives Considered
+
+- **Keep all search and fallback loading in each application.** This preserves
+  consumer ownership but repeats the file-selection and load coordination that
+  ConfigKit already owns, and risks divergent handling of selected-file errors.
+- **Search conventional locations implicitly during every load.** This reduces
+  setup, but activates filesystem access and precedence the consumer did not
+  select and makes ordinary source behavior less predictable.
+- **Return candidate paths for the application to load one by one.** This keeps
+  selection outside the load lifecycle and encourages repeated complete loads,
+  potentially rerunning defaults, callbacks, and validation during fallback.
+
+## Consequences
+
+The public search description must be immutable and root-owned, with private
+engine conversion and selection. The contract documents candidate ordering,
+selection, exhaustion, context, and provenance; the API declaration documents
+the public symbols. Dynamic bases and file availability are observed when each
+load consumes the description. Applications remain responsible for selecting
+conventions and for their schema and domain requirements.
+
+The [configuration contract](../config.md) owns implemented behavior, and the
+[component overview](../internal/overview.md) describes implementation ownership.
+The completed implementation plan and remediation record are preserved in Git
+history at `0369111` under `docs/roadmap/`.

docs/config.md

@@ -12,6 +12,7 @@ ConfigKit exposes opaque sources through these constructors:
 | Constructor | Current contract |
 | --- | --- |
 | `File(path)` | Requires a nonempty path, opens a regular file for each consumption, and closes only the library-owned file handle. Missing, inaccessible, unreadable, directory, and other nonregular inputs fail. |
+| `SearchConfig(application, search, options...)` | Creates an explicit, reusable search source. Loading selects one host file or follows the absent-input lifecycle when an optional search is exhausted. |
 | `NewFilesystem(label, filesystem).File(name)` | Borrows a caller-owned `fs.FS` under a safe nonempty display label. `name` must be a non-root slash-relative `fs.ValidPath`. Each load reopens the selected regular file; copies of the `Filesystem` preserve namespace identity. Separate constructors create separate namespaces even for the same label or filesystem. |
 | `Bytes(label, data)` | Requires a nonempty, non-secret diagnostic label and snapshots bytes during construction. Later caller mutation cannot affect the source. |
 | `Reader(label, input)` | Requires a nonempty, non-secret label and non-nil reader. The reader is claimed once across all copies of the source, consumed but never closed, and cannot be reused. |
@@ -28,6 +29,79 @@ Keys are sorted for stable provenance, integer decimal spelling stays exact, and
 cycles or unsupported types are rejected without exposing values. Mapping input
 uses the same directive and version rules as YAML, not CLI override restrictions.
 
+### Configuration discovery
+
+Discovery is opt-in. Construct `Search` from literal `Paths`, directory searches,
+or the supported directory presets, then pass it to `SearchConfig`. Search
+descriptions are immutable and reusable; each load resolves dynamic directories
+and checks current candidate files again. `Searches` is ordered concatenation.
+It preserves the supplied description order and does not deduplicate candidates.
+
+`NamedFilenames` selects `<application>.yml`, then `<application>.yaml`.
+`GenericFilenames` selects `config.yml`, `config.yaml`, `main.yml`, then
+`main.yaml`. `SearchDirectory` defaults to `[Named, Generic]` when options are
+omitted or `Filenames` is nil. A non-nil empty filename list contributes no
+candidates and does not resolve its directory. `CustomFilenames` supplies
+literal single filename components; it does not expand application-name
+placeholders. An application slug is required only when a selected directory or
+filename group uses it. Literal custom paths are not expanded for `~`, variables,
+or wildcards; relative paths are resolved from the working directory at load time.
+The zero `Search`, `Searches()`, `Paths()`, and an empty custom filename group
+contribute no candidates. An explicitly supplied empty path or filename is
+invalid. Slugs and custom filenames must be usable single host filename
+components; ConfigKit preserves their case and spelling.
+
+Search expansion is ordered concatenation. For each directory search, ConfigKit
+tries its filename groups and names before advancing to the next directory.
+When multiple directories use the same filename list, candidates follow the
+directory-first Cartesian product: every filename in one directory is tried
+before the next directory. Explicit file candidates stay at their declared
+positions among directory searches.
+
+The presets expand to these ordered directories and filename groups:
+
+| Preset | Directory, in order | Filename groups |
+| --- | --- | --- |
+| `UnixSystemDirectories` | `/usr/local/etc/<application>`, `/etc/<application>` | Named, then Generic |
+| `UnixSystemDirectories` | `/etc` | Named only |
+| `UserConfigDirectory` | Go user-config directory / `<application>` | Named, then Generic |
+| `UserConfigDirectory` | Go user-config directory | Named only |
+| `HomeDirectory` | Go home directory / `.<application>` | Named, then Generic |
+| `HomeDirectory` | Go home directory | Named only |
+| `WorkingDirectory` | Current working directory | Named only |
+
+The Unix system preset is supported on aix, android, darwin, dragonfly, freebsd,
+hurd, illumos, ios, linux, netbsd, openbsd, and solaris; reaching it on another
+platform is an option error. Other presets use Go's user-config, home, and
+working-directory helpers. A reached
+directory-helper error terminates selection. Bases for later searches are not
+resolved after an earlier candidate is selected. Missing candidates continue;
+the first existing candidate is selected and then follows ordinary file loading
+rules. Errors while acquiring or processing a selected file are terminal and do
+not resume the search.
+
+Searches are optional by default. Exhaustion uses the same absent-main lifecycle
+as `NoSource()`: there is no main document, source record, source ID, or explicit
+presence contribution. Imports may be enabled but there are no main declarations
+to traverse. Defaults, version checks, overlays, CLI overrides, resolution, and
+validation retain their normal lifecycle. `SearchOptions{Required: true}` makes
+exhaustion an error; it does not change candidate or selected-file error handling.
+
+For an ordinary active overlay after optional exhaustion, provide an explicit
+filesystem context. A host context with only `Root` is sufficient; ConfigKit
+loads the overlay at its normal root-relative location and does not infer a root
+from attempted search directories. The same root-only context can accompany a
+search that selects a file: the selected file determines its base and must be
+inside `Root`. A selected host file rejects an explicit `BaseDir` or native
+filesystem context. Invalid context fails without resuming selection. Layered
+overlays keep their independent `OverlayFilesystem` requirement.
+
+When a search selects a file, inspection reports that actual file with ordinary
+source IDs, paths, placement, metadata origins, and resource context. Missing
+candidates create no records. `SearchConfig` can be used as the main source of
+ordinary or layered loading, including an underlay input, and does not add a new
+precedence tier.
+
 Each source is limited to 8 MiB. A load using imports or an active overlay is limited
 to 64 MiB across all accepted sources, 1,024 accepted sources, 64 active import
 edges, and 1,000,000 input nodes in aggregate. Glob expansion separately permits
@@ -52,10 +126,11 @@ preserves already validated secret state without rereading the environment.
 The current lifecycle is:
 
 ```text
-validate options → parse the selected source → traverse/compose enabled imports
+validate options → parse the selected source → extract source controls and
+  traverse/strictly compose enabled imports
   → build schema → check the optional application version → apply a selected overlay
-  → apply explicit CLI overrides → expand sibling inheritance
-  → create and clone defaults → strictly decode
+  → expand sibling inheritance
+  → create and clone defaults → strictly decode → apply explicit CLI overrides
   → recorded application resolution
   → resolve typed secret references → semantic validation → classify secrets
   → snapshot configuration and metadata
@@ -67,13 +142,24 @@ ID `source-1`; enabled imports continue in depth-first order as `source-2` onwar
 contributes no source, and continues through defaults, decode, and projection. A
 missing file remains an error; it is never interpreted as no input.
 
-The current `Options` boundary accepts a defaults factory, `VersionRule`, resolver,
-semantic validator, explicit path and glob imports, one selected overlay, and
-filesystem context. It also accepts explicit CLI override records, static secret
-tags, and a dynamic path predicate. `NoSource` rejects imports, active overlays,
-and filesystem context as invalid combinations but accepts overrides. Input
-containing a reserved `configkit` mapping key is rejected while imports are
-disabled.
+`Options` accepts the application definition, ordinary-source `Destination` and
+`PlacementMode`, explicit imports, one selected overlay, filesystem
+context, explicit CLI overrides, and shared source-version controls. The definition
+may register sequence identities through `Definition.SequenceIdentities`. Its
+zero-valued placement mode places the source mapping
+at the destination; `sequence-item` explicitly wraps the mapping as one item at a
+non-root destination. Unknown modes, invalid destinations, and invalid or
+conflicting sequence registrations fail before source acquisition. Overlay and
+CLI coordinates remain rooted at the application root. Ordinary `Load` recognizes
+source-root collection declarations whether or not imports are enabled; an
+`imports` declaration still requires imports to be enabled. `NoSource` may enable
+imports, but it has no main declarations to traverse. It accepts an explicit
+filesystem context: a host context requires an absolute existing `Root` and may
+omit `BaseDir` to use that root as its internal import base; a native context
+requires its namespace, `BaseDir`, and `Root`. Supplied context is validated even
+when imports are disabled. An active ordinary overlay requires this explicit
+context. `NoSource` accepts overrides and remains absent rather than contributing
+an empty document.
 
 Every source, options, schema, version, defaults, decode, resolution, or validation
 failure returns a nil result and a `*LoadError`. Successful results contain caller-owned `Config`, detached
@@ -123,62 +209,148 @@ nothing. Underlay groups are listed from lowest to highest priority. Repeating
 a file within one group remains an import/repetition error; the same reusable
 source may be selected again in another tier.
 
-ConfigKit acquires and composes the main input first, constructs the schema,
-and checks the optional application version from the main tier only. Underlay
-roots cannot claim that version path (`underlay.version`); imported fragments
-retain `import.version` protection. It then acquires and composes underlay
-groups in order, recursively combines the groups, and applies the main tier,
-one selected overlay, and explicit CLI overrides. Scalar, sequence, and null
-definitions replace lower values between tiers; mappings combine recursively.
-Syntax and strict-composition errors inside a tier cannot be masked by a
-higher-priority tier. Defaults, typed decoding, resolution, secret lookup,
-validation, and publication run once on the final assembled input.
+ConfigKit acquires the main input first and constructs the schema. It then
+acquires and strictly composes underlay groups from lowest to highest priority,
+passing each group's accumulated lower collection policy into the next group.
+Each group is applied once as a precedence tier. The main tier is then composed
+with the accumulated underlay policy, and the optional application version is
+checked against that composed main tier only. Underlay roots cannot claim that
+version path (`underlay.version`); imported fragments retain `import.version`
+protection. ConfigKit then applies the selected overlay and resolves inheritance.
+Scalar, sequence, and null definitions replace lower values between tiers unless
+a collection policy defines the sequence operation; mappings combine recursively.
+After successful defaults and typed decoding, CLI overrides are applied to that
+typed configuration before application resolution. Syntax, strict-composition,
+inheritance, and decoding errors cannot be masked by a CLI override. Resolution,
+secret lookup, validation, and publication run once on the resulting values.
 
 An active layered overlay requires `LayeredOptions.OverlayFilesystem` explicitly,
 independent of the main source's namespace. Its name and root-relative lookup
 follow the selected-overlay rules below; inactive overlays ignore the context.
 The same path in a different filesystem namespace is not a repeated file.
-Ordinary `Load` continues to infer its selected-overlay context from its source.
+Ordinary `Load` infers overlay context from sources that provide one; with
+`NoSource`, it requires the explicit `Options.Filesystem` context.
 
 Source IDs follow acquisition order—main and its depth-first imports, then each
 underlay tier's inputs and depth-first imports, then overlay and CLI—even though
-history follows precedence order. Source `Layer` describes precedence while
+history follows precedence order. Default main-version checking happens after
+underlay composition because the main tier inherits the accumulated collection
+policy. Enabled per-source version checks retain acquisition order and source
+local headers. Source `Layer` describes precedence while
 `Kind` describes acquisition. Underlay records use `LayerUnderlay` and a
 one-based `UnderlayIndex` giving their original tier position; all other records
 have index zero. Explicit empty groups and `NoSource` consume no source ID.
 Byte, source, node, and glob-entry budgets are shared across the load, while
 the existing per-source and import-depth bounds still apply.
 
+When a source mapping is explicitly placed as an item, its effective destination
+is the sequence container. After assembly assigns an index, nested source
+destinations are projected through that item. Physical source locations and
+resource contexts continue to identify the declaring source.
+
+#### Input destinations
+
+`Input.Destination` places the source's application mapping beneath a path
+relative to the configuration root. Its zero or empty value means root, so
+existing inputs keep their behavior. Main inputs and underlay inputs use the
+same destination model. Destination segments are literal mapping keys: dots,
+slashes, and numeric-looking text are not split or interpreted as sequence
+indexes. Explicit index segments and paths that introduce the reserved
+`configkit` or mapping-valued `extends` structure are invalid input options and
+produce a structured `LoadError` before source acquisition.
+
+Source-root `configkit.imports` directives are processed before placement. An
+import without an explicit destination inherits the effective destination of
+its importing document, including through nested imports. An import may instead
+declare a relative destination as a sequence of literal mapping keys; its keys
+append to the importing document's effective destination. Omission and `[]`
+add no keys. Nested imports continue relative to their immediate importing
+document, each declaration is independent of its siblings, and all matches of
+one glob receive the same resulting destination. This logical placement does
+not change how the physical import path resolves from its declaring file.
+
+A valid destination on `NoSource()` contributes nothing; an actual empty
+mapping materializes its destination containers. Once placed, contributions
+follow the same strict composition within a tier and recursive precedence
+behavior described above. A selected overlay and CLI overrides remain rooted at
+the application root and are not moved beneath the main input's destination.
+
+`SourceInfo.Destination` reports each acquired source's effective mapping or
+sequence-container path, including imports. `PlacementMode` reports how that
+source was placed; mapping is the zero value, and CLI/overlay sources report
+mapping unless item placement was applied. `DestinationUnresolved` is true only
+when an enclosing item index needed to name the source's destination has not yet
+been assigned. In that case `Destination` reports the nearest resolved enclosing
+collection container and omits the unresolved local suffix. A source's own item
+placement still has an exact destination at its sequence container. Completed
+loads project nested source destinations through their owning items. Callback and
+failed-load snapshots preserve the evidence available at that point; a fallback
+snapshot remains unresolved, while a later exact snapshot is false. Returned
+source records and paths are detached. Placement does not change source IDs,
+acquisition identity, physical source locations, or passive resource contexts.
+Introduced ancestor mappings use the contributing document's root location as a
+source fallback; authored content retains its original source coordinates.
+
 The maintained [layered consumer example](consumers/README.md#native-filesystem-and-layered-input)
 shows a main host file, two underlay tiers, embedded defaults, a mapping source,
 and explicit resource access without a second configuration pass.
+The accepted rationale for input placement and its relative-import extension
+is recorded in [ADR 0011](adr/0011-input-destinations.md) and
+[ADR 0012](adr/0012-relative-import-destinations.md).
 
 ### Sibling inheritance
 
 Any input mapping may declare `extends: name` to inherit from the mapping named
 `name` among its immediate siblings. Names are exact keys, including the empty
-string; lookup does not search ancestor or unrelated scopes. The document root
-and a sequence-element mapping have no sibling namespace for their own string
-relationship, though mappings nested inside them can have one. An `extends`
+string; lookup does not search ancestor or unrelated scopes. A mapping inside a
+sequence can also participate in record inheritance when that sequence has an
+effective identity declaration from `configkit.sequences` or an application
+registration. In that case, `extends` is a mapping containing exactly the
+sequence's identity fields, with scalar values; lookup is limited to that final
+sequence. The document root and a sequence-element mapping without an effective
+identity declaration have no sibling namespace for non-null string
+relationships, though mappings nested inside them can have one. An `extends`
 value of null cancels a relationship without looking up a parent. Other value
-types, missing or non-mapping parents, and cycles fail the load with value-free
-diagnostics.
+types, incomplete or extra reference fields, missing parents, and cycles fail
+the load with value-free diagnostics.
+
+Within each mapping sibling scope or identified sequence, direct definitions
+resolve by dependency readiness. A parent is resolved before a child; among
+currently eligible definitions, the lexically smallest canonical semantic
+pointer is selected, including newly eligible definitions as soon as their
+parent completes. This ordering governs inheritance operations and their
+history, controls, and transfers. It does not reorder mappings or sequence
+records in the resulting configuration.
 
 `extends` is reserved structural input across ordinary and layered loads, every
 source kind, imports, selected overlays, and explicit CLI overrides. It is
 consumed before typed decoding and need not appear in the application's Go
 types. Applications previously using `extends` as an ordinary YAML field must
 rename that key and its field. No collection registration or keyword setup is
-required.
+required for mapping-sibling relationships; identified-sequence relationships
+require effective identity fields for that collection.
 
 ConfigKit completes strict composition within tiers and precedence assembly
-between tiers before resolving relationships. An omitted `extends` retains a
-lower-priority relationship; `extends: null` cancels it while preserving the
-child's own settings. Superseded references are historical evidence, not active
-relationships to validate. Each child recursively overlays its own settings
-over its resolved parent: explicit zero, false, empty string, sequence, and
-null values replace inherited values, while mappings combine recursively.
-The child retains its key. All effective relationships resolve, including
+between tiers before resolving relationships. Only records that survive that
+assembly participate in identified-sequence lookup: an upserted parent supplies
+its final record, a replaced parent is unavailable, and a higher replacement
+without `extends` supersedes a lower relationship. An omitted `extends` retains
+a lower-priority mapping-sibling relationship; an identified record's effective
+directive depends on the record that survives assembly. `extends: null` cancels
+a relationship while preserving the child's own settings. Superseded references
+are historical evidence, not active relationships to validate. Each child
+recursively overlays its own settings over its resolved parent: explicit zero,
+false, empty string, and
+null values replace inherited values, while mappings combine recursively. For
+identified records, the child's identity fields and their origins remain its
+own; parent identity fields do not contribute to the payload.
+Sequences at collection destinations use the shared collection operation with
+the parent as lower input and the child as higher input. The destination's
+explicit mode applies; otherwise a copied identity selects upsert and an
+unkeyed destination replaces. A parent's explicit mode does not move to a
+different destination. Sequences outside collection destinations remain atomic
+replacements.
+All effective relationships resolve, including
 entries the application may never select for execution. Partial parent
 definitions undergo normal typed decoding and application-owned validation,
 not a special structural exemption.
@@ -269,6 +441,67 @@ error without reporting one, ConfigKit emits a generic `version.hook_failed`
 diagnostic and suppresses the returned error text. Warnings alone are retained on a
 successful result. Callback panics remain programming failures.
 
+#### Shared source-version checks
+
+`Options.SourceVersionMode` and `LayeredOptions.SourceVersionMode` select how
+one `Definition.Version` policy applies across source documents. The zero value
+`SourceVersionModeDefault` preserves the existing main-only check at its
+effective path and existing protected-path behavior. `SourceVersionModeIfPresent`
+checks each participating document that declares a version and permits absent
+non-main declarations. `SourceVersionModeRequired` rejects a participating
+document without a declaration. Enabled modes require a usable `VersionRule`
+and checker; an unknown mode or unusable rule is an options error before source
+access or callbacks.
+
+In enabled modes, `Definition.Version.Path` is interpreted in each acquired
+document's own coordinates, before placement and composition. The main document,
+direct underlay documents, imported documents (including transitive and glob
+matches), and the selected overlay participate. `NoSource`, absent optional
+imports, empty underlay groups, defaults, resolution output, inheritance, and
+synthetic CLI overrides are not documents to check. An absent declaration in an
+actual main document still reaches the checker in if-present mode; required mode
+rejects it first. With `NoSource`, the existing single absent-main checker
+behavior remains in both enabled modes. If-present omissions in non-main
+documents skip the checker. Explicit null is present and reaches the checker;
+unsupported selected values produce `version.type`. No equality with the main
+declaration or non-null rule is imposed. Acceptance establishes only that the
+declaration is accepted by the consumer's checker; it does not establish that
+each fragment conforms to a complete application schema. Final assembled
+decoding and application validation still run once.
+
+A header that is absent below a structural control follows the same omission
+policy. Actual header/control overlap is rejected, and ordinary structural
+validation and main-version ownership protection continue independently.
+
+Accepted non-main declarations are consumed from their owned payload before
+composition; their authored ancestors remain, even if empty. The main
+declaration remains in the application. Ordinary release metadata should use a
+different path. Enabled main protection uses the main input's effective path:
+its `Destination` prefixed to the source-local `VersionRule.Path`. This protects
+the remaining payload against imports, underlays, inheritance, and CLI
+overrides. Thus a mounted main looks up a local header while ownership checks
+protect its destination-prefixed application path. Main checker diagnostics and
+declaration locations use source-local coordinates; ordinary payload
+diagnostics continue to use effective paths. Default-mode diagnostics are
+unchanged.
+
+`SourceInfo.VersionCheck` is nil when enabled declaration processing did not
+occur. Otherwise it records a source-local `Path`, the observed `Version` state
+(`Present`, `Null`, and, when disclosed, exact `Text`), an actual or nearest
+available `Location`, and whether text was redacted. A non-nil record with
+`Present == false` records observed absence; it does not say the checker ran or
+accepted a declaration. Absent evidence still identifies the full requested
+header path, while its location identifies the nearest available source node.
+During callbacks and in failed-load source records,
+text is conservatively withheld because final classification may not be
+available. Successful final source records reflect completed sensitivity
+classification at the declaration's mounted path and ancestors, including
+historical classification. Redaction empties `Text` while retaining presence
+and null state. Returned copies are detached; evidence does not reopen inputs.
+Consult [ADR 0013](adr/0013-shared-source-version-checks.md) for the accepted
+decision and the [development guide](development.md#baseline-validation) for
+the completed implementation and remediation records in Git history.
+
 ### Layered metadata
 
 `Result.Metadata` describes every final canonical path. Queries retain explicit
@@ -299,15 +532,22 @@ the descendant can therefore retain history after `Lookup` stops reporting it as
 effective. Equal-valued input is still an explicit base definition, while a derived
 value for absent input remains non-explicit.
 
+Strict composition assigns contributed records their assembled sequence positions
+in history. Across tiers, history follows the selected
+[collection operation](#strict-document-composition), including append and
+whole-record upsert. Incoming contributions use their assigned positions, and
+replaced payload remains available as historical evidence.
+
 A mapping assembled from the selected file and imports has one composite base origin
-whose locations retain source traversal order. A mapping also visited by an overlay
-has an overlay-layer composite origin with the lower contributing locations followed
-by the overlay location; an empty mapping is still an explicit container even though
-it contributes no children. Resolution origins are synthetic: they have layer
+whose locations follow [scoped assembly order](#explicit-imports), not source-record
+acquisition order. A mapping also visited by an overlay has an overlay-layer
+composite origin with the lower contributing locations followed by the overlay
+location; an empty mapping is still an explicit container even though it contributes
+no children. Resolution origins are synthetic: they have layer
 `resolution`, no fabricated file location, and a `derived_default` or `normalization`
 operation. Reasons and dependency paths retain first occurrence order with duplicates
-removed. Source records remain ordered as selected source, depth-first imports, the
-selected overlay, then synthetic CLI records in supplied order.
+removed. Source records and IDs retain [acquisition order](#explicit-imports),
+including the underlay ordering described in [layered loading](#layered-loading).
 
 Inheritance adds no acquired source or precedence layer. `Origin.Inheritance`
 lists concrete transfers from contributor paths through destinations, with
@@ -370,29 +610,34 @@ historical path. Its effective value and each chronological before/after value a
 `ValueSnapshot` records containing deterministic YAML text rather than mutable Go or
 parser values. `Present=false` with empty YAML means absence or deletion; explicit
 null is present and renders as `null`. The history starts with the canonical typed
-application-default or Go-zero baseline, then includes only relevant base sources in
-traversal order, the selected overlay, supplied CLI records, and audited net
-resolution operations. Structural inheritance events appear after source replay
-and before typed reconciliation, including consumed or canceled `extends` input
-and equal-valued child replacements. Overlapping parent/child resolution records
+application-default or Go-zero baseline, followed by relevant underlay and main-tier
+contributions in precedence order, the selected overlay, and structural inheritance
+events. Typed defaulting and decoding reconcile that source history before supplied
+CLI records and audited net resolution operations. Within each strict tier, replay follows
+[scoped assembly order](#explicit-imports), not source-record acquisition order.
+Inheritance events include consumed or canceled `extends` input and equal-valued
+child replacements. Overlapping parent/child resolution records
 are rendered
 as one net subtree event with combined reasons and dependencies, rather than
 pretending that the recorded assignments ran in lexical order. Explicit equal
-writes remain visible. When input replay
-differs from the final typed input, a `ChangeDecoding` event records the transition
-before domain resolution. Its origin retains the decoded path's source/default
-attribution; its reason identifies static-default retention and canonical decoding.
-This built-in inspection operation is not accepted by `Resolution.Record`.
+writes remain visible. When structural replay differs from the pre-CLI typed input,
+a `ChangeDecoding` event records retained defaults and canonical decoding. CLI
+contributions then retain their raw input representation while effective and
+resolution values use the typed projection. This built-in inspection operation is
+not accepted by `Resolution.Record`.
 
 Base, overlay, and CLI history preserves the policy-validated input representation,
 including a superseded value that could not have decoded into the final Go field.
 Baseline, effective, and resolution values use canonical typed projection. A
 decoding event bridges replayed input and canonical typed values, so intermediate
 null/scalar replacements cannot incorrectly imply permanent default removal. Mapping
-queries reconstruct each source or override contribution, while sequences remain
-atomic: a descendant history identifies its presence before and after the containing
-sequence replacement without implying element-wise merging. Removed paths remain
-explainable; unknown and malformed paths return false.
+queries reconstruct each source or override contribution. A sequence snapshot
+represents the collection at that point in replay, and sequence diffs remain
+atomic. Descendant queries describe their presence and values before and after
+containing sequence operations. Combination follows the
+[collection operation tables](#strict-document-composition); upsert replaces
+matching records in full. Removed paths remain explainable; unknown and malformed
+paths return false.
 
 Every snapshot applies the same historical tag, predicate, and recorded-dependency
 sensitivity closure as effective rendering before YAML is exposed. A redacted scalar
@@ -423,6 +668,46 @@ an explicit `LoadSecretsDir` bootstrap, then use `WriteDiagnostics`,
 ConfigKit does not infer bootstrap outcomes from a secret reference and does not
 bundle these independent reports into a command framework.
 
+## Effective configuration diffs
+
+`configkit.Diff(left, right)` compares two completed successful loads of the same
+configuration type. `left` is the earlier state and `right` is the later state.
+Applications load both sides with the sources and options they choose; to inspect
+an overlay's impact, load the same base once without the overlay and once with it
+selected. The [consumer guide](consumers/README.md#effective-configuration-diffs)
+links to maintained examples. The public API returns safe changed-region records;
+applications own their presentation.
+
+Comparison uses the frozen canonical effective inspection values, not mutable
+`Result.Config`, source text, provenance, or history. It does not reopen sources,
+reread environment variables, or rerun defaults or callbacks. Equal effective
+values produce no records even when sources, explicit-input status, defaults, or
+resolution histories differ; for example, explicit zero and defaulted zero are
+equal. Mappings compare recursively, sequences compare atomically, and scalar,
+kind, or null/container replacements report their affected region. Records use
+the existing comparator's canonical pointer ordering. `Present=false` with empty
+YAML means absence; explicit null is present and renders as `null`.
+
+Before either side is returned, sensitivity from both loads is applied to both
+snapshots, including historical and dependency-derived classification. A
+classification on either side protects the corresponding values on both sides.
+`Redacted` reports masking, while `HiddenChanged` reports an actual concealed
+value or presence change; it is not implied merely by redaction. A structural
+replacement that prevents safe correspondence masks the complete reported
+region, and an absent side remains absent. Record membership establishes that a
+change occurred even when safe snapshots render identically. Literal
+marker-looking input is ordinary input; consumers use the record fields rather
+than interpreting marker text.
+
+Typed secret references contribute their configured environment names to the
+frozen comparison, never resolved credential bytes. Credential rotation behind
+an unchanged reference is therefore not detected. The records contain changed
+regions only, not paired full-document context; independently calling
+`WriteEffective` on each result does not create cross-protected context. Nil or
+uninitialized results return an error with no records. A failed `Load` retains
+its normal nil-result/`LoadError` behavior, which callers handle before invoking
+`Diff`.
+
 ## Typed secret references
 
 Credential-bearing fields use the opaque `Secret` type and store an environment
@@ -439,13 +724,15 @@ must resolve to a nonempty value. Secret references work in typed structs, maps,
 pointers, slices, and arrays. A dynamic `any` value does not establish secret
 semantics and cannot contain `Secret`; applications must use a typed slot.
 
-ConfigKit resolves only the effective names after recorded application resolution
-and before semantic validation. Each unique name is read with `os.LookupEnv` once
-per load. Validation can inspect the credential through the deliberately named
-`Value()` accessor and distinguish success with `IsResolved()`. Later loads read
-the environment again, and cloned defaults are rebuilt from names so a previously
-resolved result cannot retain a stale credential. `Load` never writes to the
-environment.
+ConfigKit resolves only the effective names after CLI assignment and recorded
+application resolution, and before semantic validation. CLI overrides may supply
+or replace a typed secret reference; lookup uses that effective name and retains
+its effective source location for diagnostics. Each unique name is read with
+`os.LookupEnv` once per load. Validation can inspect the credential through the
+deliberately named `Value()` accessor and distinguish success with `IsResolved()`.
+Later loads read the environment again, and cloned defaults are rebuilt from names
+so a previously resolved result cannot retain a stale credential. `Load` never
+writes to the environment.
 
 Credential bytes are held only in the per-load lookup state and the trusted returned
 configuration, and are returned only by `Value`. Secret formatting uses a safe
@@ -509,18 +796,19 @@ files when appropriate; ConfigKit stores no implicit global report.
 
 ### Recorded application resolution
 
-An optional `Definition.Resolve` callback runs after successful typed decoding. It
-receives the working configuration and a per-load `Resolution` context. The context's
+An optional `Definition.Resolve` callback runs after successful defaults, typed
+decoding, and explicit CLI assignment. It receives the working configuration and a
+per-load `Resolution` context. The context's
 `IsExplicit` query uses surviving input presence, so explicit false, zero, empty, and
 null are never confused with absence or static defaults.
 
 `Resolution.InputInfo(path)` returns detached effective metadata from the fixed
-decoded entry state, before application resolution runs. It includes retained static
-defaults, preserves explicit null, and uses the same exact semantic path matching as
-`Metadata.Lookup`; missing, malformed, and key/index-shape-mismatched paths return
-false without diagnostics. Its `Explicit` value agrees with `IsExplicit` for every
-found entry path. Resolution changes, additions, removals, and moves do not alter
-this entry view.
+post-decode entry state, including explicit CLI overrides, before application
+resolution runs. It includes retained static defaults, preserves explicit null, and
+uses the same exact semantic path matching as `Metadata.Lookup`; missing,
+malformed, and key/index-shape-mismatched paths return false without diagnostics.
+Its `Explicit` value agrees with `IsExplicit` for every found entry path. Resolution
+changes, additions, removals, and moves do not alter this entry view.
 
 `Resolution.Source(id)` returns a detached record for an already acquired source in
 the same load: the main source, imports, selected overlay, and supplied CLI records
@@ -595,8 +883,11 @@ live services as a substitute for configuration validation.
 
 ## Strict document composition
 
-ConfigKit strictly composes parsed base documents when `Options.Imports` is true.
-Application defaults do not participate in this operation.
+Ordinary `Load` applies strict document composition to its parsed source scope;
+`Options.Imports` adds imported documents as peers in that scope. Application
+defaults do not participate. With imports disabled, the single source still has
+its controls extracted and any declared or registered collection identity is
+validated, but there are no imported peers to combine.
 
 | Existing node | Incoming node | Result |
 | --- | --- | --- |
@@ -606,20 +897,148 @@ Application defaults do not participate in this operation.
 | Mapping | Mapping | Recurse by mapping key |
 | Any other pair of present nodes | Any other pair of present nodes | `compose.conflict` |
 
-Every non-mapping overlap is a conflict, even when values are equal. Nulls and
-sequences are present atomic values, so they also conflict when repeated; sequences
-are never joined by index. Disjoint entries under shared mappings compose, making
-stable keyed mappings the natural representation for independently contributed
-entries.
+Every non-mapping overlap is a conflict, even when values are equal, except for
+sequence/sequence contributions at a destination with effective identity from an
+application registration, a current YAML `key` declaration, or retained lower
+policy. Nulls remain atomic conflicts.
+Sequences without effective identity conflict when repeated and are never joined
+by index. Disjoint entries under shared mappings compose, making stable keyed
+mappings the natural representation for independently contributed entries.
+
+`Definition.SequenceIdentities` contains registrations created by
+`SequenceIdentityAt` or `SequenceIdentityMatching`. The exact form accepts a
+mapping-only `Path`; the matching form accepts literal-key and any-mapping-key
+segments created by `SequenceLiteralKey` and `SequenceAnyMappingKey`. A literal
+`"*"` key is distinct from the wildcard constructor. Parsed numeric path tokens
+remain literal mapping keys; explicit index segments and reserved `configkit` or
+mapping-valued `extends` segments are invalid. The root selector is harmless under
+the mapping-root schema. Selector segments consume one mapping level and match a
+complete destination; an any-mapping-key segment matches exactly one key, not an
+arbitrary prefix or suffix. Each registration requires one or more literal,
+direct field names; compound field sets are normalized as sets, and field names
+are not otherwise normalized. The same field-set and scalar-equality rules apply
+to YAML `key` declarations.
+
+At each selected path, records must provide every effective identity field as a
+scalar string, number, boolean, or explicit null. Field names form a set, so
+order and repetition do not affect compound identity. The type-tagged tuple
+preserves category distinctions: strings compare exactly, including empty
+strings; numbers, booleans, and null remain distinct categories. Finite numbers
+compare by exact mathematical value, so `1`, `1.0`, and `1e0` are equal, as are
+`0` and `-0`, without floating-point rounding. Distinct tuples append in source
+and list order; duplicate tuples conflict even when records are otherwise equal
+or disjoint. Malformed records produce `compose.invalid`; duplicate identities
+produce `compose.conflict`, with paths and source locations identifying a
+duplicate without exposing identity values. Missing paths and non-sequence values
+do not fail by themselves. Equivalent overlapping registrations coalesce.
+Incompatible overlaps and invalid registrations produce `options.invalid` before
+source acquisition. Unused and root registrations are allowed and have no effect.
+
+Registrations provide identity for strict composition, sequence operations,
+inheritance lookup, and final validation. Dynamic selectors apply at concrete
+strict-composition, inheritance, and final-result destinations without requiring
+the application to enumerate mapping keys. Each strict tier has independent
+duplicate-identity state among its peers; layered precedence and inheritance
+then apply the collection operation described below.
+
+Final registration checks run in both ordinary and layered loads after effective
+secret lookup and before application semantic validation. They examine actual
+final typed sequences, including values introduced by defaults, CLI, or recorded
+resolution; they do not require path existence or sequence shape. Diagnostics
+use concrete collection and item paths and available source locations, without
+exposing identity or credential values. The guarantee ends when loading publishes
+successfully and does not cover later caller mutations.
+YAML-only identity is enforced during structural composition, precedence, and
+identified inheritance; it carries no final-result guarantee after typed
+defaults, CLI, or application resolution. Applications that need that guarantee
+register the identity constraint.
+The [maintained catalog examples](consumers/README.md#identified-prompt-sequences)
+show strict peers, precedence operations, inheritance, dynamic destinations, and
+the distinction between YAML-only declarations and application registrations.
+When an enabled source-version mode is selected, declarations are checked in each
+source's own coordinates before placement and composition. Accepted non-main
+headers are consumed before strict composition and cannot supply a registered
+record identity. Final typed decoding and application validation still run on the
+completed configuration.
+
+Ordinary sources, layered inputs, and selected overlays may declare collection
+behavior under `configkit.sequences` at the document root. Each entry has a
+required literal `destination` and may specify `key` as one string or a nonempty
+list of strings, plus a `mode` of `replace`, `append`, or `upsert`. These
+controls are removed before application decoding. A declaration destination is
+relative to that source's placement, so a mounted fragment's declaration is
+mounted with its values. Declarations from strict peers must agree on identity
+fields and on any explicitly supplied mode; omission expresses no mode
+preference. They apply even when the declaring source has no value at that
+destination: a key validates an existing retained sequence or remains pending
+until records arrive, while a mode-only declaration affects later combinations
+without recombining earlier tiers. A key supplies the identity used by `append`
+and `upsert` and is checked against any application registration.
+
+Strict peers at an identified destination first combine distinct records in
+source and list order. Duplicate identities among peers or within one
+contribution fail under every mode. After strict composition, precedence and
+inheritance use one operation for the assembled incoming collection:
+
+| Mode | Behavior at source-precedence and inheritance boundaries |
+| --- | --- |
+| `replace` | Replace the complete lower sequence with the assembled higher sequence. |
+| `append` | Retain lower records and append higher records; duplicate identities fail. |
+| `upsert` | Replace matching records in full at their lower positions, retain unmatched lower records, and append new identities. |
+
+Upsert replaces the complete matched record; it does not recursively merge
+record fields. Append and upsert require identity. With no explicit mode in the
+current tier, an explicit lower mode persists; if none exists, effective identity
+selects upsert and its absence selects replace. Inferred modes do not persist.
+Explicit modes persist at their destination until changed by a higher explicit
+mode, independently of YAML identity lifetime. A declaration alone does not
+clear or recombine a collection.
+
+The presence and value kind at a precedence or inheritance boundary determine
+whether a collection changes:
+
+| Higher contribution | `replace` | `append` / `upsert` |
+| --- | --- | --- |
+| Absent | Leave the lower value unchanged. | Leave the lower value unchanged. |
+| Explicit `[]` | Clear the sequence. | Contribute no records. |
+| Explicit null or another non-sequence value | Replace the lower value, subject to application schema. | Error; the operation requires a sequence. |
+
+An absent lower value can be initialized by an incoming sequence under any mode.
+A present lower null or other non-sequence value is not absence: append and
+upsert cannot combine with it. An empty sequence still requires a valid mode and
+identity where applicable. Null used as an identity component inside a record is
+distinct from null used as the collection value.
+
+Declarations govern ordinary strict source composition, including loads with
+imports disabled, and each strict scope in layered loading. Before composing a
+higher tier's peers, ConfigKit makes applicable lower identity and explicit mode
+available; after peer composition, the tier applies its resolved operation
+against the lower collection. Selected overlays use the same collection policy
+after the base source graph is composed. During mapping-sibling inheritance, a
+parent's YAML identity can follow its collection to the child destination, while
+the parent's explicit operation stays at the parent destination. The child
+destination selects its operation; absent one, effective identity implies
+upsert and its absence implies replace. Whole-sequence replacement discards
+lower YAML identity and uses the replacing scope's key declaration, if any;
+without one, the collection is unkeyed unless an application registration still
+supplies identity. Replacement discards policies owned by removed records, while
+an independently declared explicit mode and application registration remain
+applicable. Item-local
+policies follow their owning records: retaining a record retains its rules, while
+whole-record upsert replacement discards the displaced record's rules. Matching
+identity or reusing an index does not transfer them. These policies do not change
+CLI behavior: CLI sequence assignments replace whole sequences after decoding.
+A root `configkit.imports` declaration still requires `Options.Imports`;
+`NoSource` has no declaration scope.
 
 An empty mapping contributes no child definitions. Two empty mappings compose, and
 an empty mapping neither clears nor conflicts with another mapping's children.
 Shared mapping containers retain all contributing value locations as their composite
-base origin. A conflict reports the first claim as its primary location and every
-remaining claim as related locations, in deterministic source and declaration
-traversal order. Traversal order never authorizes replacement. Composition clones
-the resulting tree and its ordered claim records rather than mutating or aliasing an
-input document.
+base origin. A conflict reports the first claim in
+[scoped assembly order](#explicit-imports) as its primary location and every
+remaining claim as related locations in the same order. Assembly order never
+authorizes replacement. Composition clones the resulting tree and its ordered
+claim records rather than mutating or aliasing an input document.
 
 ## Selected overlays
 
@@ -630,13 +1049,21 @@ Every other name resolves only to `<configuration-root>/overlays/<name>.yml`, wi
 no extension fallback, case normalization, inheritance between overlay files,
 or overlay stack. A selected missing overlay is an error.
 
+With `NoSource`, the main input stays absent and the explicit filesystem context
+supplies the overlay root. The overlay is recorded as the first acquired source
+when no other source exists. Host contexts may use the root-only form described
+under [single-source loading](#single-source-loading); native contexts retain
+their complete namespace, base, and root requirements.
+
 The overlay is a separately policy-validated YAML mapping. It is loaded after the
 base import graph is composed and its schema version is checked, but before defaults
 and typed decoding. Overlay bytes, nodes, and source records count toward the same
 aggregate limits as base inputs. Its canonical path must remain within the configured
 root and must not duplicate a file already present in the base graph.
 
-Overlay is intentionally distinct from strict composition:
+Collection destinations follow the
+[collection operation and presence rules](#strict-document-composition) above.
+The following table describes ordinary overlay behavior elsewhere:
 
 | Existing lower node | Incoming overlay node | Result |
 | --- | --- | --- |
@@ -645,18 +1072,24 @@ Overlay is intentionally distinct from strict composition:
 | Mapping | Mapping | Recurse by mapping key |
 | Any other pair of present nodes | Any other pair of present nodes | Replace the entire lower node |
 
-Scalars, sequences, and null are atomic. Thus an overlay sequence replaces rather
-than extends a base sequence, `[]` clears it, null is an explicit replacement, and
-cross-kind replacement is allowed when the final application schema accepts it.
+Under these ordinary rules, scalars, sequences, and null are atomic. An overlay
+sequence replaces rather than extends a base sequence, `[]` clears it, null is an
+explicit replacement, and cross-kind replacement is allowed when the final
+application schema accepts it.
 Replacing a subtree removes all displaced descendants from effective presence and
 origin data; history retains the displaced definitions and identifies the parent
 replacement. Surviving overlay paths have overlay origins, while recursively shared
 mapping containers retain their contributing locations.
 
-An overlay may not contain `configkit` anywhere, contribute the configured schema
-version, or replace an ancestor of that version with a scalar or null. These rules
-apply even if the supplied version text would be unchanged. Overlays cannot mask a
-base parse, composition, or version failure because those checks run first.
+An overlay may contain only root `configkit.sequences` declarations; overlay
+imports and nested directives are rejected. Sequence declarations are removed
+before application decoding. At declared or registered sequence destinations,
+`replace`, `append`, and `upsert` use the same identity and policy rules as source
+composition. Other overlay sequences remain atomic replacements. An overlay may
+not contribute the configured schema version or replace an ancestor of that
+version with a scalar or null. These rules apply even if the supplied version
+text would be unchanged. Overlays cannot mask a base parse, composition, or
+version failure because those checks run first.
 
 ## Explicit CLI overrides
 
@@ -668,10 +1101,16 @@ must never contain the raw argument or configured value. General environment
 variable overlays are an explicit non-goal: environment access is reserved for
 typed secret references.
 
-Overrides apply after the selected overlay and before typed defaults/decoding, so
-the input precedence is application defaults < composed base < selected overlay <
-CLI. Recorded resolution, typed secret lookup, and semantic validation follow the
-effective override tree. Overrides also work with `NoSource`.
+Overrides are prepared after source assembly and inheritance, then assigned after
+typed defaults and decoding and before application resolution. The order is:
+defaults and decoded sources < CLI < application resolution. A CLI mapping
+recursively overlays the decoded mapping; a CLI sequence replaces the complete
+decoded sequence regardless of source collection operations. Inheritance and
+typed defaulting do not run again after CLI assignment. Earlier parsing,
+composition, inheritance, and decoding failures remain errors even if an override
+would replace the failing value. Recorded resolution, typed secret lookup, and
+semantic validation use the effective post-override values. Overrides also work
+with `NoSource`.
 
 Values are converted directly into ConfigKit's owned tree without YAML parsing or
 string coercion. Supported values are `nil`, built-in strings and booleans, all
@@ -682,7 +1121,8 @@ false, and zero remain intentional; empty sequences replace whole lists; and map
 values recursively overlay existing mappings. Map keys are sorted canonically and
 sequence order is preserved.
 
-All records are detached and validated before any is applied. Root and explicit
+All records are detached and validated before any is applied. They cannot create
+`configkit` or `extends` controls. Root and explicit
 sequence-index paths are invalid; parsed numeric tokens remain mapping keys. A path
 may create missing mapping ancestors but cannot traverse an existing scalar, null,
 or sequence. Duplicate and ancestor/descendant-overlapping records are rejected.
@@ -699,8 +1139,12 @@ unsupported types, nonfinite floats, and limit excesses fail with safe `override
 diagnostics. Diagnostics and provenance identify the safe `Flag` source, never raw
 arguments or values.
 
-Override origins are explicit CLI origins. History retains replaced base/overlay
-definitions and records equal-value replacements as well as structural changes.
+Override origins are explicit CLI origins. Resolution receives CLI values with
+their explicit-presence and source information. Absence-based defaults preserve
+those values, while recorded normalization and derivation may change them. History
+retains replaced source definitions and records equal-value replacements as well
+as structural changes. Secret lookup and validation use post-resolution values
+with their effective source locations.
 Caller-owned maps and slices are snapshotted during loading and must not be mutated
 concurrently with the call. The executable `ExampleOverride` shows translating a
 parser's explicit-set state into records while preserving an explicitly supplied
@@ -708,12 +1152,22 @@ zero.
 
 ## Explicit imports
 
-Imports are opt-in. With `Options.Imports` enabled, a document may reserve its root
-`configkit` mapping for an `imports` sequence. ConfigKit removes that mapping before
-application decoding. Each sequence item is a mapping with exactly one nonempty
-string `path` or `glob` plus an optional boolean `optional`; all other shapes and
-fields are `import.invalid`. A `configkit` key below the root is invalid. With
-imports disabled, any occurrence remains `import.disabled`.
+Imports are opt-in. With `Options.Imports` enabled, a document may use its root
+`configkit` mapping for `imports` and `sequences` controls. Ordinary `Load` may use
+`sequences` without enabling imports, but an `imports` field still requires
+`Options.Imports`. ConfigKit removes recognized root controls before application
+decoding. Each imports sequence item is a mapping with exactly one `path`,
+`dir`, or `glob` selector. `path` and `dir` accept a string or a list of strings;
+`glob` accepts one path pattern string. The optional `required` boolean controls
+whether an empty selection or a missing named path is an error. Its default is
+false, so missing imports are absent unless `required: true` is specified. The
+removed `optional` field is invalid. `destination` is an optional sequence of
+strings. Every destination string is one literal mapping key and follows the
+input-destination validation rules above; quote numeric-looking keys so YAML
+reads them as strings. Null, scalar destinations, and non-string members are
+`import.invalid`, as are all other unsupported shapes and fields. A `configkit`
+key below the root is invalid. With imports disabled, root imports and nested
+`configkit` keys remain `import.disabled`.
 
 The directive key is `configkit`, not the `_configkit` spelling used in v0.1.0.
 There is no legacy alias: `_configkit` is ordinary application data, subject to
@@ -722,25 +1176,118 @@ directive keys in root files and imported fragments, and rename any application
 fields previously named `configkit` to avoid the newly reserved key.
 
 Path declarations use slash-separated relative paths. Absolute paths and network
-URLs are invalid. Each path resolves from its declaring file's canonical directory,
-and files are traversed depth-first in declaration order: a file contributes its own
-body before its imports. Order stabilizes source records and diagnostics but never
-permits replacement. Required missing paths fail; `optional: true` suppresses only
-target nonexistence, not invalid paths, containment failures, directories,
-permissions, parse errors, or composition errors.
+URLs are invalid. Each path resolves from its declaring file's canonical directory.
+An empty `path: []` selects nothing; it succeeds even when `required: true` because
+there are no named targets to be missing. A nonempty required path list verifies
+every named target, including targets later excluded from selection.
+
+Directory declarations search the listed literal directory roots. Missing roots
+contribute no candidates; supplied inaccessible or invalid roots fail. `pattern`
+accepts one basename pattern or a list, matching filenames only. Omission selects
+all filenames; an explicit empty list selects none. `exclude` accepts one basename
+pattern or a list and removes matching final candidates. Patterns use Go's
+case-sensitive `path.Match` grammar. `recursive` enables descendant traversal;
+`depth` is valid only when recursion is effective, with direct files at depth 1 and
+depth 0 or omission meaning unlimited traversal subject to library budgets.
+`follow` controls discovered symlink entries and `hidden` controls discovered
+names beginning with a dot. Both default to true. Explicit roots retain their
+normal access behavior.
 
 Glob declarations use the slash-based Go `path.Match` grammar: `*`, `?`, character
-classes, and backslash escapes are supported, while recursive `**` syntax is invalid.
-Backslashes escape pattern characters rather than acting as directory separators.
-ConfigKit validates the complete pattern before filesystem traversal, walks only
-directories selected by preceding segments, and reads wildcard directories in
-bounded batches. Every examined entry consumes the separate visit budget, including
-nonmatches. Matches are sorted lexicographically by slash-relative spelling within
-each declaration before depth-first source traversal. A required zero-match glob is
-`import.missing`; `optional: true` suppresses only zero matches. Invalid patterns,
-walk failures, matched directories, invalid documents, escapes, and repeated targets
-remain errors. Final candidates are bounded by the remaining source limit before
-their contents are accumulated.
+classes, and backslash escapes are supported, while recursive `**` syntax is
+invalid. Backslashes escape pattern characters rather than acting as directory
+separators. ConfigKit validates the complete pattern before filesystem traversal,
+walks only directories selected by preceding segments, and reads wildcard
+directories in bounded batches. Every examined entry consumes the separate visit
+budget, including nonmatches. Literal components are explicit path components;
+`follow` and `hidden` filter only wildcard-discovered entries. `exclude` filters
+final candidate basenames.
+
+`Options.ImportDefaults` and `LayeredOptions.ImportDefaults` provide one
+application-owned load-wide policy for enabled imports in the main source,
+underlays, and transitive imports; they do not apply to selected overlays or
+discover sources by themselves. Each applicable field resolves independently:
+an explicit declaration value wins, then the application default, then the
+library default (`required: false`, `follow: true`, `hidden: true`,
+`recursive: false`, and unrestricted filenames). A declaration's `pattern`
+replaces rather than extends the default pattern. Nil `Pattern` leaves directory
+filenames unrestricted; a non-nil empty slice selects none. Defaults do not make
+a field applicable to selectors that do not support it, and load calls snapshot
+the supplied pointers and slice before source acquisition.
+
+A source's `configkit.sequences` declarations apply independently of import
+enablement; a `configkit.imports` declaration still requires `Options.Imports`.
+Sequence identities can be supplied either by these YAML declarations or by
+application registrations. The [strict composition contract](#strict-document-composition)
+defines their agreement, operation, and lifetime rules.
+
+Acquisition visits declarations in authored order and each declaration's selected
+paths in namespace-path lexical order, preserving source-record and version-callback
+order. Ordinary mapping contributions in a main tier are assembled dependency-first:
+ConfigKit selects the lexically smallest stable source key among the contributions
+whose imports are complete. Independent branches may interleave, and the main body
+may be assembled after its imports. Sequence-item contributions in that tier are
+nodes in the same schedule; each selected item first assembles its own strict
+mapping scope, with helpers before the bodies that import them. Those local scopes
+do not use application collection policies and are not flattened into the outer
+schedule. Item positions and diagnostic paths follow the resulting assembly order.
+Each underlay group likewise acquires its inputs before dependency-ready assembly
+in one strict scope. Independent groups remain separate precedence tiers; at a
+declared or registered destination, the assembled higher sequence uses the
+resolved collection operation against the lower sequence. Other sequences retain
+whole-value replacement between tiers. Assembly order never permits replacement.
+
+Acquisition order and resolution order serve different purposes: acquisition
+assigns source records and callback order, while the shared scheduler determines
+composition order among currently dependency-ready contributions. Inheritance
+uses the same eligibility rule within each destination-local sibling scope; it
+does not reorder the authored output mapping. The private namespace and stable-key
+construction is described in the [import implementation overview](internal/imports.md).
+
+An empty selection is allowed by default. `required: true` makes an empty directory
+or glob selection fail with `import.missing`; it also makes missing named paths
+fail. Requiredness does not suppress invalid paths, containment failures,
+directories selected as files, permissions, parse errors, or composition errors.
+
+For inspection, `SourceInfo.ImportSelections` records one outcome for each
+completed optional declaration whose entire selected set is empty. The record is
+attached to its declaring source, keeps the declaration path local to that source,
+identifies the declaration mapping (including its `SourceID`) and selector, and
+uses outcome `none`. Records remain in declaration order and are available
+through the existing result, load
+error, and callback source snapshots. They do not enumerate candidates, explain
+why a path was absent, or describe partial omissions. Empty `path: []` succeeds
+regardless of requiredness: it produces a `none` record when effectively optional,
+and no record when effectively required. Snapshots detach the record paths and
+slices.
+
+Each declaration may include `placement: mapping` or
+`placement: sequence-item`. Omission means `mapping` for that declaration,
+including every match of a glob; placement is not inherited. Sequence-item
+placement requires a nonempty effective destination and wraps the imported
+mapping as one element of a sequence there. It does not split source sequences
+or infer identity fields. Repeated sequences combine only when the destination
+has an applicable YAML identity declaration or application registration; source
+placement remains independent of collection identity and operation. Sequence
+records retain source locations through placement and history; indexes describe
+their assembled positions while source coordinates continue to identify the
+declaring files.
+Header/version policy retains its existing owner and timing. Application
+registrations also enforce identity on the final effective typed configuration;
+YAML-only declarations are enforced at structural combination and inheritance
+boundaries. The [configuration contract](#strict-document-composition) defines
+these validation boundaries.
+
+An item import first opens a local mapping scope. Its body and ordinary helper
+imports compose within that record. An omitted helper destination or
+`destination: []` places the helper at the record root; other destinations are
+relative to the immediate importing source. A helper may supply fields such as
+the record's `id`. An explicit nested `sequence-item` import creates a sequence
+inside the record, and its placement is independent of the enclosing item.
+At the record root, an item import must name a nonempty nested destination.
+Physical paths continue to resolve from the file that declares each import.
+An actual empty mapping contributes an empty item, while an unmatched declaration
+with default or explicit `required: false` contributes nothing.
 
 For host files, the configuration root defaults to the selected file's canonical
 directory. A host file may supply `FilesystemContext.Root` to widen that root but
@@ -759,8 +1306,13 @@ ConfigKit canonicalizes symlinks,
 checks containment with relative paths, and opens the canonical root-relative target
 through a library-owned root handle. Contained relative or absolute symlink targets
 are accepted; escapes are rejected. Canonical paths define identity, so hard links at
-distinct paths remain distinct sources. Native filesystems use slash-path operations
-and their own symlink behavior; imports stay in their selected namespace and root.
+distinct paths remain distinct sources. Native filesystems use slash-path
+operations. Where the supplied `fs.FS` implements `fs.ReadLinkFS`, ConfigKit uses
+that capability to inspect and resolve discovered links; otherwise it follows
+the filesystem's ordinary `fs.FS` behavior. ConfigKit does not require every
+native filesystem to expose link inspection, and native filesystem behavior is
+not a general sandbox guarantee. Imports stay in their selected namespace and
+root.
 
 An active-stack repeat is `import.cycle` and reports the complete cycle's import
 sites. A previously visited canonical path is `import.repeated` and reports both

docs/consumers/README.md

@@ -6,6 +6,30 @@ applications. Consumers still own their configuration structs, defaults, semanti
 validation, schema-version policy, CLI parsing, runtime matching and conversion,
 resource checks, remote acquisition, and service initialization.
 
+## Choosing a loading entry point
+
+- Use [`Load`](../../lifecycle.go) with one source for an ordinary configuration.
+  [`SearchConfig`](../../discovery.go) is also a source, so its explicit
+  discovery result can be passed to `Load` or used as the main input to
+  [`LoadLayered`](../../layered.go).
+- Use `LoadLayered` when inputs must be composed in underlay groups before the
+  main source. Each [`Input`](../../layered.go) owns its source, filesystem,
+  destination, and import choice. The overlay filesystem context is selected
+  independently in [`LayeredOptions`](../../layered.go).
+- Load-wide rules belong to [`Options`](../../lifecycle.go) or
+  [`LayeredOptions`](../../layered.go): the application definition (including
+  version policy), import defaults, and explicit overrides. These do not turn
+  on imports or discovery by themselves. See the
+  [configuration contract](../config.md#layered-loading) and
+  [discovery contract](../config.md#configuration-discovery) for exact rules.
+
+The root [README example](../../README.md) shows the small `Load` path.
+[`layered.go`](../../examples/consumers/layered.go) demonstrates the bundled
+file and layered cases. For import-focused adoption, see the
+[prompt catalog](../../examples/consumers/prompt_catalog.go); for explicit
+discovery composed with sequences, see the
+[sequence catalog](../../examples/consumers/sequence_catalog.go).
+
 The maintained [`examples/consumers`](../../examples/consumers) package demonstrates
 sanitized shapes using only ConfigKit and the Go standard library:
 
@@ -17,7 +41,11 @@ sanitized shapes using only ConfigKit and the Go standard library:
 | Seriatim-like | Independent strict ordered speaker and autocorrect rule files | First-match and ordered replacement algorithms; no central application configuration file |
 | Bundled file | Simple native `fs.FS` source through ordinary `Load` | Selection and lifetime of the application-owned embedded filesystem |
 | Layered catalog | Embedded defaults with an import, a mapping underlay, an independent host main file, explicit false/zero input, and source-relative resource access | Domain validation and opening/interpreting the template resource |
+| Mounted collections | Entry-declared relative destinations, nested metadata/prompt/profile imports, local inheritance, and a version-only higher-tier override | Collection IDs, profile meaning, and opening/interpreting prompt resources |
 | Named presets | One-load sibling inheritance, explicit false, nested destination resolution, and source/derivation inspection | Choosing and executing a preset |
+| Effective configuration diff | Same-type frozen effective comparisons for two selected files or base-versus-overlay impact | Source/options selection and consumer-owned safe presentation |
+| Shared source-version checks | One `Definition.Version` policy for accepted declarations across participating source documents, with optional or required declarations and sanitized evidence | Choosing which declarations the application accepts; final assembled schema and domain validation |
+| Sequence catalogs | Declared collection operations, record inheritance, typed CLI and resolution, inspection, exact and wildcard application registrations | Catalog meaning and final domain validation |
 
 The YAML under [`testdata`](../../examples/consumers/testdata) is loaded by the
 executable examples and ordinary tests. It contains no private application data,
@@ -51,6 +79,70 @@ both without redefining their syntax. ConfigKit overlays are configuration overl
 they do not select or replace an application's LLM, model, pipeline, report, or
 execution profile concepts.
 
+## Explicit main-file discovery
+
+Applications choose the search conventions and their precedence, then pass the
+resulting `SearchConfig` source to the same `Load` or `LoadLayered` entry point.
+For example, the maintained sequence catalog puts a caller-selected file before
+a custom directory fallback while keeping its underlay and overlay choices
+independent. See [`sequence_catalog.go`](../../examples/consumers/sequence_catalog.go)
+and its [behavior tests](../../examples/consumers/sequence_catalog_test.go), plus
+the [working-directory executable example](../../examples/consumers/discovery_test.go).
+Together, the root-owned [discovery API](../../discovery.go) and [configuration
+discovery contract](../config.md#configuration-discovery) define ordered
+concatenation, directory-first Cartesian expansion, optional exhaustion, and
+selected-file behavior. [ADR 0019](../adr/0019-explicit-configuration-discovery.md)
+records why discovery is explicitly selected.
+
+## Import defaults and selection inspection
+
+`Options.ImportDefaults` and `LayeredOptions.ImportDefaults` apply one
+load-wide policy to enabled main, underlay, and transitive imports. Each field
+is resolved independently: a declaration's explicit value wins, then the
+application default, then the library default. `Required`, `Follow`, `Hidden`,
+and `Recursive` use nil pointers for unspecified values. `Pattern` applies to
+directory searches: nil leaves filenames unrestricted, while a non-nil empty
+slice selects no files. A declaration's own `pattern` replaces the application
+default. The defaults do not enable imports or change selected-overlay behavior.
+
+`Result.Sources`, `LoadError.Sources`, and `Resolution.Source` expose each
+declaring source's `ImportSelections`. These records identify the source-local
+declaration path, its authored location and selector, and the `none` outcome;
+they do not include candidate names or a guessed reason for absence. Returned
+records are detached snapshots. The [configuration contract](../config.md#explicit-imports)
+owns selector applicability and exact absence semantics.
+
+## Shared source-version checks
+
+[`LoadVersionedService`](../../examples/consumers/source_version.go) demonstrates
+if-present and required modes with one `Definition.Version` rule across a mounted
+main input and an underlay. The example reads only sanitized
+`SourceInfo.VersionCheck` fields; its executable test is in
+[`layered_test.go`](../../examples/consumers/layered_test.go). The
+[configuration contract](../config.md#shared-source-version-checks) defines
+participation, source-local lookup and diagnostics, non-main header consumption,
+main-path protection, default behavior, and evidence disclosure. Accepted
+declarations do not replace final assembled decoding or application validation.
+
+## Effective configuration diffs
+
+[`CompareFiles`](../../examples/consumers/diff.go) loads two caller-selected files
+with one application definition, handles each load error, calls `configkit.Diff`,
+and presents the safe records through consumer-owned `FormatDifferences`.
+[`CompareRolloutOverlay`](../../examples/consumers/diff.go) loads the same base
+once without an overlay and once with the selected overlay, so omitted lower
+priority values remain present. Its resolver records a derived endpoint URL after
+overlay application, and its nested typed secret reference is classified before
+the resulting records are presented.
+
+The maintained [behavior tests and executable examples](../../examples/consumers/diff_test.go)
+use temporary files and disposable secret values. Presenters should use
+`ValueSnapshot.Present` for absence, record membership for change, and
+`HiddenChanged` for a withheld change indicator. Do not parse the redaction marker
+or add separately rendered full contexts that lack combined sensitivity. The
+[configuration contract](../config.md#effective-configuration-diffs) owns shared
+equality, granularity, and disclosure semantics.
+
 ## Named sibling inheritance
 
 [`LoadPresets`](../../examples/consumers/inheritance.go) loads the maintained
@@ -88,6 +180,79 @@ and may fail independently of the completed configuration load. The
 [configuration contract](../config.md#source-resource-contexts) owns exact
 lifetime and precedence rules.
 
+[`LoadPromptBundle`](../../examples/consumers/layered.go) is the ordinary-root
+counterpart: its entry opts into imports through `Load`, imports metadata at the
+root, and places a prompt beneath `prompts` using YAML. The executable example
+and behavior test are in
+[`layered_test.go`](../../examples/consumers/layered_test.go).
+
+## Input destinations and mounted collections
+
+Main and underlay inputs use the same destination field. An import without a
+`destination` inherits its declaring document's effective destination; an
+explicit sequence of key strings appends a relative suffix. Nested imports
+append to their immediate parent's effective destination, while filesystem
+paths remain relative to the declaring file. Destinations are application
+mapping paths, independent of filesystem context. The configuration contract
+owns the complete behavior and validation rules.
+
+The small [`Example_inputDestination`](../../examples/consumers/destinations_test.go)
+mounts one mapping source beneath literal collection/prompt keys and loads a
+second input without a destination to show that root placement remains the
+default. Consumers pass paths, not filenames or namespace strings for ConfigKit
+to interpret.
+
+[`LoadCollectionBundle`](../../examples/consumers/destinations.go) assembles an
+embedded collection manifest mounted beneath a literal collection path. The
+entry's YAML places nested metadata imports at the collection root and prompt
+and profile peers beneath their own keys. A higher-priority input changes only
+one prompt's ordinary `version` field. The main document alone supplies the
+application schema version. A prompt's local `extends` relationship resolves
+after all tiers, so `custom` inherits the final `brief` content and fields while
+retaining its own profile selection.
+
+The resolver follows each effective `content_file` origin to its declaring
+source context. This also works for `custom`'s inherited resource. The
+application-owned [`ReadCollectionPromptContent`](../../examples/consumers/destinations.go)
+demonstrates the same lookup after loading; ConfigKit does not open or interpret
+the resource. [`destinations_test.go`](../../examples/consumers/destinations_test.go)
+checks source destinations and authored coordinates, presence, inherited values,
+relative resources, and strict peer conflicts. The
+[configuration contract](../config.md#input-destinations) owns shared placement,
+import, source-location, and ownership behavior.
+
+## Identified prompt sequences
+
+[`LoadPromptCatalog`](../../examples/consumers/prompt_catalog.go) shows the
+smaller import-focused adoption path: a registered `prompts.id` identity,
+directory and glob selection, a helper import that supplies an item's ID, and
+resource lookup through the effective source context. Its executable examples
+and tests are in
+[`prompt_catalog_test.go`](../../examples/consumers/prompt_catalog_test.go), with
+complete [catalog](../../examples/consumers/testdata/prompt-catalog/config.yml)
+and [required-file](../../examples/consumers/testdata/prompt-catalog/simple.yml)
+fixtures.
+
+[`LoadSequenceCatalog`](../../examples/consumers/sequence_catalog.go) integrates
+an underlay declaration, imported main fragments, identified-record inheritance,
+selected overlay `append`, `upsert`, and `replace`, a typed CLI override, and
+recorded application resolution. Its tests and runnable example demonstrate
+stable output order, whole-record upsert, provenance/history inspection,
+resource lookup for inherited records, and secret-safe explanation in
+[`sequence_catalog_test.go`](../../examples/consumers/sequence_catalog_test.go).
+[`LoadNamespacedMapCatalog`](../../examples/consumers/sequence_catalog.go) adds
+an exact root registration and a wildcard namespace registration, showing that
+each concrete sequence has an independent identity scope. The same tests show
+that a YAML identity declaration without an application registration does not
+create final-result identity enforcement.
+
+The examples keep prompt meaning and domain checks in the consuming application.
+The [configuration contract](../config.md#strict-document-composition) owns
+declaration, identity, operation, ownership, and final-validation semantics;
+[sibling inheritance](../config.md#sibling-inheritance) defines record
+relationships, and the [CLI contract](../config.md#explicit-cli-overrides) owns
+override ordering.
+
 Known module catalogs should use canonical object entries with a module identifier
 and typed options. A small union struct is sufficient when the available option
 fields are known; validation selects which fields apply to each module. Runtime-
@@ -167,7 +332,9 @@ focused behavioral owner:
 | Canonical module objects and dynamic option handling | WeatherReporter- and Notarius-shaped examples and migration tests |
 | `LoadSecretsDir`, bootstrap reporting, typed lookup, CLI explanation | `Example_weatherStartup` |
 | Native filesystem, grouped underlays, programmatic mapping, retained resource context | `Example_nativeFilesystem`, `Example_layeredCatalog`, and their focused consumer tests |
+| Literal input destinations, root-default compatibility, and mounted peer collections | `Example_inputDestination`, `LoadCollectionBundle`, and `destinations_test.go` |
 | Named sibling inheritance, presence, and derivation | `Example_inheritedPresets` and `TestPresetCatalogOneLoadAndDerivation` |
+| Effective configuration diffs and consumer-owned presentation | `ExampleDiff`, `ExampleCompareFiles`, and `ExampleCompareRolloutOverlay` |
 | Results, metadata, diagnostics, effective output, and writer failures | Focused root-package load, provenance, rendering, and explanation tests |
 
 The package-level import-boundary test covers the consumer examples too: supported

docs/development.md

@@ -39,9 +39,11 @@ public behavior.
 | Task | Read |
 | --- | --- |
 | Repository orientation or component responsibility | [Component overview](internal/overview.md) |
+| Input destinations | [Configuration contract](config.md#input-destinations), [ADR 0011](adr/0011-input-destinations.md), and the destination tests below |
+| Shared source-version checks | [Configuration contract](config.md#shared-source-version-checks), [ADR 0013](adr/0013-shared-source-version-checks.md), and the version/source-version tests below |
 | Documentation changes | The owning document |
 | Licensing, dependencies, attribution, or release-document changes | [Project grant](../COPYRIGHT) and [attribution](oss/attribution.md); release changes also follow the release procedure |
-| Architecture decisions or future work | [Deferred scope](roadmap/deferred.md) and relevant ADRs listed below |
+| Architecture decisions or future work | [Future features](roadmap/future.md), [remediation roadmap](roadmap/roadmap.md), and relevant ADRs listed below |
 | Code and test changes | The owning contract and tests from the map below |
 | Examples or copyable assets | The owning implemented contract and [consumer adoption guide](consumers/README.md) |
 | Release preparation | [Release procedure](release.md) |
@@ -49,11 +51,19 @@ public behavior.
 Accepted decisions:
 
 - [Separate composition and overlay](adr/0001-separate-composition-and-overlay.md)
+- [Check source versions with one shared policy](adr/0013-shared-source-version-checks.md)
 - [Own YAML semantics behind a private parser boundary](adr/0002-own-yaml-semantics.md)
 - [Preserve presence and record application resolution](adr/0003-preserve-presence-and-resolution-history.md)
 - [Secret references and explicit bootstrap](adr/0004-secret-references-and-explicit-bootstrap.md)
 - [Explicit overrides and safe explanation](adr/0005-explicit-overrides-and-safe-explanation.md)
 - [Keep the public facade root-owned](adr/0006-public-facade.md)
+- [Compare frozen effective configuration snapshots](adr/0010-effective-configuration-diff.md)
+- [Place inputs at literal application mapping paths](adr/0011-input-destinations.md)
+- [Share dependency-ready ordering](adr/0015-shared-dependency-ready-ordering.md)
+- [Separate import selection controls](adr/0016-independent-import-controls.md)
+- [Define identity and merge rules for each collection](adr/0017-collection-sequence-policy.md)
+- [Apply CLI overrides after inheritance and decoding](adr/0018-cli-after-structural-resolution.md)
+- [Keep main-file discovery explicit and composable](adr/0019-explicit-configuration-discovery.md)
 
 For cross-cutting changes, follow every applicable row. Completed implementation
 and remediation plans are preserved in Git history at `cb5b435`; use current
@@ -64,8 +74,9 @@ git show cb5b435:docs/roadmap/remediation.md
 ```
 
 Release-candidate, publication, and verification work follows the operational
-[release procedure](release.md). Its single publication command validates the
-committed candidate and publishes only the intended refs.
+[release procedure](release.md). Its module-selected ReleaseKit `publish`
+invocation validates the committed candidate and publishes only the intended
+tag; those release checks remain distinct from the everyday development check.
 
 The completed string-and-integer schema-version implementation and remediation
 records are preserved in Git history at `244104f`. The
@@ -93,9 +104,114 @@ preserved in Git history at `73ce6e9` under `docs/roadmap/`. Current overlay
 behavior belongs to the [configuration contract](config.md#selected-overlays),
 and release operations belong to the [release procedure](release.md).
 
-Install Go 1.25.0 or a supported patched/newer Go toolchain. Acquire declared
-dependencies explicitly before entering an offline development or validation
-environment:
+The completed ReleaseKit adoption, startup performance request and experiments,
+benchmark evidence, performance audit, implementation plan, and remediation
+record are preserved in Git history at `8e91a2b` under `docs/roadmap/`. For example:
+
+```sh
+git show 8e91a2b:docs/roadmap/startup-benchmark.md
+```
+
+Remaining optimization candidates belong to
+[future performance experiments](roadmap/future.md#5-performance-experiments).
+
+The completed effective-configuration diff roadmap, implementation plan, and
+resolved review findings are preserved in Git history at `1c2610c` under
+`docs/roadmap/`. The
+[configuration contract](config.md#effective-configuration-diffs) owns current
+comparison and disclosure behavior, and
+[ADR 0010](adr/0010-effective-configuration-diff.md) records the decision.
+
+The completed input-destination, relative-import-destination, and shared
+source-version roadmaps, associated consumer request, implementation plan, and
+resolved review findings are preserved in Git history at `e4270c9` under
+`docs/roadmap/`. For example:
+
+```sh
+git show e4270c9:docs/roadmap/implementation.md
+git show e4270c9:docs/roadmap/remediation.md
+```
+
+Current behavior belongs to the configuration contract's
+[input destinations](config.md#input-destinations),
+[imports](config.md#explicit-imports), and
+[shared source-version checks](config.md#shared-source-version-checks) sections.
+Explicit sequence-item placement, collection operations, identity registration,
+and final registered-identity validation are owned by the
+[strict composition](config.md#strict-document-composition) section;
+[imports](config.md#explicit-imports) owns acquisition and scheduling. The
+[maintained sequence catalog](consumers/README.md#identified-prompt-sequences)
+shows the consumer boundary, and [ADR 0017](adr/0017-collection-sequence-policy.md)
+records the collection decision.
+
+The completed sequence-composition roadmap, implementation plan, and resolved
+remediation record are preserved in Git history at `008d2ba` under
+`docs/roadmap/`. For example:
+
+```sh
+git show 008d2ba:docs/roadmap/sequence-composition.md
+git show 008d2ba:docs/roadmap/implementation.md
+git show 008d2ba:docs/roadmap/remediation.md
+```
+
+The completed import-discovery roadmap, implementation plan, and resolved
+remediation record are preserved in Git history at `5747ec4` under
+`docs/roadmap/`. For example:
+
+```sh
+git show 5747ec4:docs/roadmap/imports.md
+git show 5747ec4:docs/roadmap/implementation.md
+git show 5747ec4:docs/roadmap/remediation.md
+```
+
+The subsequent import-integration audit and completed remediation record are
+preserved at `6903c9e:docs/roadmap/remediation.md`; retrieve them with
+`git show 6903c9e:docs/roadmap/remediation.md`.
+
+The completed sequence-identity/collection-policy roadmap and implementation
+plan are preserved in Git history at `d0a605a`:
+
+```sh
+git show d0a605a:docs/roadmap/sequences.md
+git show d0a605a:docs/roadmap/implementation.md
+```
+
+Current behavior belongs to the configuration contract's
+[strict composition](config.md#strict-document-composition),
+[sibling inheritance](config.md#sibling-inheritance), and
+[CLI overrides](config.md#explicit-cli-overrides) sections.
+The completed remediation record, including the algorithm-audit findings, is
+preserved at `ea46001:docs/roadmap/remediation.md`.
+
+The completed configuration-discovery roadmap, implementation plan, and
+remediation record are preserved in Git history at `0369111`, for example
+`git show 0369111:docs/roadmap/remediation.md`. Current behavior belongs to
+[configuration discovery](config.md#configuration-discovery); the public
+[discovery declarations](../discovery.go) own the Go API.
+
+The completed consolidation proposal, implementation plan, and resolved audit
+findings are preserved in Git history at `34cbc0d` under `docs/roadmap/`. For
+example:
+
+```sh
+git show 34cbc0d:docs/roadmap/implementation.md
+git show 34cbc0d:docs/roadmap/remediation.md
+```
+
+Potential enhancements are tracked in [future features](roadmap/future.md).
+The completed audit remediation's scope is recorded in the
+[remediation roadmap](roadmap/roadmap.md). The consolidated audit, including
+findings, evidence, and settled scope decisions, is preserved at
+`453da68:docs/roadmap/audit.md`; the original nine reports are preserved at
+`ecd5d5d` under `docs/roadmap/`. Retrieve the consolidated record with:
+
+```sh
+git show 453da68:docs/roadmap/audit.md
+```
+
+Install the Go version required by `go.mod` or a compatible newer toolchain.
+Acquire declared dependencies explicitly before entering an offline
+development or validation environment:
 
 ```sh
 go mod download
@@ -125,11 +241,10 @@ Use focused checks while iterating, with validation proportionate to the change
 under the testing policy. Documentation-only work normally needs the local-link
 check below and whitespace review; behavior-bearing examples and commands also
 need execution. Use `go test -race ./...` for ownership or concurrency changes.
-The [publication command](release.md#4-publish-the-committed-candidate) invokes
-the release checker once. It performs one minimum-toolchain offline race-suite
-run, one full vet pass, and Linux cross-builds using readonly module mode.
-It already covers examples, links, and package boundaries; do not additionally
-run the everyday checker just to repeat those checks. CI never gates a release.
+The [release procedure](release.md#4-publish-the-committed-candidate) performs
+its selected validation through one `publish` invocation. The everyday checker
+remains a development gate and is not a separate release prerequisite. CI never
+gates a release.
 
 ## Task-to-test map
 
@@ -140,22 +255,29 @@ code changes. Use the component overview to find the corresponding implementatio
 | Task | Behavioral test owners | Focused command |
 | --- | --- | --- |
 | YAML syntax, diagnostics, parser limits | [Adapter tests](../internal/yaml/parse_test.go), [source translation](../internal/engine/parse_test.go) | `go test ./internal/yaml` and `go test ./internal/engine -run '^TestParseSource'` |
-| File/reader access, imports, globs, containment | [Sources](../internal/engine/source_test.go), [FIFO](../internal/engine/source_fifo_test.go), [imports](../internal/engine/imports_test.go), [globs](../internal/engine/glob_test.go) | `go test ./internal/engine -run 'Source\|Import\|Glob\|Filesystem\|Nonregular\|RegularFile'` |
+| File/reader access, imports, globs, containment | [Import internals](internal/imports.md), [source acquisition](../internal/engine/source_test.go), [FIFO](../internal/engine/source_fifo_test.go), [import traversal](../internal/engine/imports_test.go), [import evidence](../internal/engine/import_evidence_test.go), [relative destinations](../internal/engine/destination_import_test.go), [globs](../internal/engine/glob_test.go) | `go test ./internal/engine -run 'Source\|Import\|Glob\|Filesystem\|Nonregular\|RegularFile'` |
+| Explicit main-file discovery and absent-input loading | [Candidate selection](../internal/engine/discovery_test.go), [load integration](../internal/engine/discovery_load_test.go), [public facade](../facade_discovery_test.go), and maintained [catalog](../examples/consumers/sequence_catalog_test.go) and [preset](../examples/consumers/discovery_test.go) examples | `go test ./internal/engine -run 'Discovery|Search'`, `go test . -run 'Search'`, and `go test ./examples/consumers -run 'WorkingDirectoryPreset|SequenceCatalog'` |
+| Import defaults, empty-selection evidence, and dependency-ready ordering | [Defaults and facade evidence](../facade_import_defaults_test.go), [engine import evidence](../internal/engine/import_evidence_test.go), [scheduler](../internal/engine/dependency_order_test.go), [scheduling integrations](../internal/engine/imports_test.go) | `go test ./internal/engine -run 'Import|ScheduleDependencies|Inheritance'`, `go test . -run 'ImportDefaults|FacadeImport'`, and `go test ./examples/consumers` |
 | Native filesystem and mapping sources | [Native acquisition](../internal/engine/native_filesystem_test.go), [filesystem facade](../facade_filesystem_test.go), [mapping facade](../facade_mapping_test.go) | `go test ./internal/engine -run 'NativeFilesystem'` and `go test . -run 'TestFacade'` |
 | Composition, selected overlays, CLI precedence | [Composition](../internal/engine/compose_test.go), [overlay](../internal/engine/overlay_test.go), [selected overlays](../internal/engine/selected_overlay_test.go), [overrides](../internal/engine/override_test.go) | `go test ./internal/engine -run 'Compose\|Overlay\|Override'` |
-| Sibling inheritance, provenance, history, sensitivity | [Expansion](../internal/engine/inheritance_test.go), [destination scopes](../internal/engine/inheritance_walk_test.go), [origins](../internal/engine/inheritance_origin_test.go), [history](../internal/engine/inheritance_history_test.go), [secrets](../internal/engine/inheritance_secret_test.go), [public loading](../facade_inheritance_test.go) | `go test ./internal/engine -run 'Inheritance\|Inherited'` and `go test . -run '^TestFacadeInheritance\|^TestFacadeInheritedSecrets'` |
+| Sequence identity, declarations, and operations | [Identity](../internal/engine/sequence_identity_test.go), [registrations](../internal/engine/sequence_registration_test.go), [declarations](../internal/engine/sequence_declaration_test.go), [combination](../internal/engine/sequence_combine_test.go), [final facade validation](../facade_final_sequence_identity_test.go), and [consumer catalog](../examples/consumers/sequence_catalog_test.go) | `go test ./internal/engine -run 'Sequence\|Collection'`, `go test . -run 'SequenceIdentity\|SequenceRegistration'`, and `go test ./examples/consumers` |
+| Sibling inheritance, provenance, history, sensitivity | [Implementation flow](internal/overview.md#implementation-flow), [expansion](../internal/engine/inheritance_test.go), [destination scopes](../internal/engine/inheritance_walk_test.go), [origins](../internal/engine/inheritance_origin_test.go), [history](../internal/engine/inheritance_history_test.go), [secrets](../internal/engine/inheritance_secret_test.go), [public loading](../facade_inheritance_test.go) | `go test ./internal/engine -run 'Inheritance\|Inherited'` and `go test . -run '^TestFacadeInheritance\|^TestFacadeInheritedSecrets'` |
 | Layered loading and source equivalence | [Grouped lifecycle](../internal/engine/load_request_test.go), [layered facade](../facade_layered_test.go), [equivalence](../facade_equivalence_test.go) | `go test ./internal/engine -run '^TestGrouped'`, `go test . -run '^TestFacadeLayered'`, and `go test . -run '^TestEquivalentSourceForms'` |
-| Schema, defaults, projection, cloning | [Schema](../internal/engine/schema_test.go), [decoding](../internal/engine/decode_test.go), [projection](../internal/engine/projection_test.go) | `go test ./internal/engine -run 'Schema\|Decode\|Projection\|Clone\|Inline\|CompareDocument'` |
+| Input and relative-import destinations, mounted sources, and source ownership | [Destination engine tests](../internal/engine/destination_test.go), [tier tests](../internal/engine/destination_tiers_test.go), [import traversal and version tests](../internal/engine/destination_import_test.go), [accounting tests](../internal/engine/destination_accounting_test.go), [lifecycle inspection tests](../internal/engine/destination_import_lifecycle_test.go), [public source/error tests](../facade_destination_test.go), [resource contexts](../facade_resource_context_test.go), and [consumer examples](../examples/consumers/destinations_test.go) | `go test ./internal/engine -run 'Destination|Import|Glob|Mounted'`, `go test . -run 'Destination|Import|SourceContext|FacadeMounted'`, and `go test ./examples/consumers` |
+| Shared source-version modes and evidence | [Engine source-version tests](../internal/engine/source_version_test.go), [public facade tests](../facade_source_version_mode_test.go), and [consumer example tests](../examples/consumers/layered_test.go) | `go test ./internal/engine -run 'SourceVersion'`, `go test . -run 'SourceVersion'`, and `go test ./examples/consumers` |
+| Schema, defaults, projection, cloning | [Implementation flow](internal/overview.md#implementation-flow), [schema](../internal/engine/schema_test.go), [decoding](../internal/engine/decode_test.go), [projection and typed cloning](../internal/engine/projection_test.go) | `go test ./internal/engine -run 'Schema\|Decode\|Projection\|Clone\|Inline\|CompareDocument'` |
 | Lifecycle, resolution, validation, callback ownership | [Loading](../internal/engine/load_test.go), [resolution](../internal/engine/resolution_test.go), [validation](../internal/engine/validation_test.go), [ownership](../internal/engine/callback_ownership_test.go), [independence](../internal/engine/independence_test.go) | `go test ./internal/engine -run 'Load\|Resolution\|Validation\|Validator\|Hook\|Callback\|Parallel'` |
 | Paths, metadata, provenance | [Paths](../internal/engine/path_test.go), [metadata](../internal/engine/metadata_test.go), [provenance](../internal/engine/provenance_test.go) | `go test ./internal/engine -run 'Path\|Metadata\|Provenance\|Queries'` |
 | Passive resource contexts | [Engine contexts](../internal/engine/resource_context_test.go), [public contexts](../facade_resource_context_test.go) | `go test ./internal/engine -run 'Resource'` and `go test . -run 'SourceContext'` |
 | Secrets, bootstrap, redaction, explanation | [References](../internal/engine/secret_reference_test.go), [bootstrap](../internal/engine/secrets_test.go), [rendering](../internal/engine/render_test.go), [explanation](../internal/engine/explain_test.go) | `go test ./internal/engine -run 'Secret\|Bootstrap\|Effective\|Explain\|Explanation\|History\|Sensitivity'` |
-| Consumer examples and migration fixtures | [Consumer tests](../examples/consumers/consumers_test.go), [native and layered examples](../examples/consumers/layered_test.go), [API examples](../example_test.go) | `go test ./examples/consumers` and `go test . -run '^Example'` |
+| Effective configuration diff | [Engine comparison and masking tests](../internal/engine/diff_test.go), [engine operation tests](../internal/engine/diff_operation_test.go), [root facade tests](../facade_diff_test.go), and [consumer examples](../examples/consumers/diff_test.go) | `go test ./internal/engine -run 'Diff|Explain|History|Secret|Independent'`, `go test . -run 'Diff|FacadeInspection|PublicFacadeBoundary|DependencyImportBoundary'`, `go test ./examples/consumers`, and `go test . -run '^ExampleDiff$'` |
+| Consumer examples and migration fixtures | [Consumer tests](../examples/consumers/consumers_test.go), [native, layered, and ordinary relative-import examples](../examples/consumers/layered_test.go), [mounted collection examples](../examples/consumers/destinations_test.go), [API examples](../example_test.go) | `go test ./examples/consumers` and `go test . -run '^Example'` |
 | Assembly and layered scaling evidence | [Assembly benchmarks](../internal/engine/assembly_bench_test.go), [layered benchmarks](../internal/engine/underlay_bench_test.go) | `go test ./internal/engine -run '^$' -bench '^BenchmarkUnderlay' -benchmem`; repeat with `^BenchmarkCompose` and `^BenchmarkOverlay` |
+| Identified-sequence scaling evidence | [Mixed-upsert benchmark](../internal/engine/sequence_combine_bench_test.go) | `go test ./internal/engine -run '^$' -bench '^BenchmarkSequenceMixedUpsert$' -benchmem` |
 | Inheritance scaling evidence | [Expansion benchmarks](../internal/engine/inheritance_bench_test.go) | `go test ./internal/engine -run '^$' -bench '^BenchmarkInheritance' -benchmem` |
 | Documentation links and dependency isolation | [Link checks](../documentation_test.go), [import boundary](../import_boundary_test.go), [public boundary](../public_boundary_test.go) | `go test . -run 'TestLocalDocumentationLinks\|TestDependencyImportBoundary\|TestPublicFacadeBoundary'` |
-| Development and release checker behavior | [Shell regression](../release_check_test.go) | `go test . -run 'TestCheckScripts\|TestDevelopmentCheck'` |
-| Guarded release publication and recovery | [Publication regression](../release_publish_test.go) | `go test . -run '^TestReleasePublication'` |
+| Development check behavior | [Development checker test](../development_check_test.go) | `go test . -run '^TestDevelopmentCheckHandlesFilenamesAndStopsOnFailure$'` |
+| Release profile wiring | [Release profile test](../releasekit_profile_test.go) | `go test . -run '^TestReleaseKitProfile$'` |
 
 ## Fuzz validation
 

docs/internal/imports.md

@@ -0,0 +1,125 @@
+# Import Discovery and Scheduling Internals
+
+The [configuration contract](../config.md#explicit-imports) owns selector syntax,
+defaults, requiredness, placement, inspection evidence, and observable ordering.
+This note records implementation flow and private identities for maintainers.
+
+## Discovery and acquisition
+
+Each source's root directive is parsed into presence-aware declarations and
+resolved against load-wide defaults. `acquisition.go` owns shared acquisition
+policy, load-wide accounting, and tier-scoped repetition/namespace state;
+`import_acquisition.go` owns reading candidate bytes, declaration traversal,
+and import admission. Candidate
+selection in `import_selection.go` preserves a normalized
+namespace path separately from canonical access identity: overlapping roots
+coalesce the same candidate spelling in that namespace, while aliases reached
+through distinct spellings retain separate inclusion routes for admission.
+Directory searches walk each distinct normalized root once, apply all filename
+patterns to discovered names, and form a set union of matching candidates across
+overlapping roots. Exclusions are applied as set difference after discovery and
+candidate normalization. They do not suppress discovery or resource errors.
+
+Directory traversal tracks canonical directories on the active traversal path.
+It adds an identity when descending and removes it when the frame is popped, so
+only a back-edge to an active directory is a traversal cycle; a later alias is
+not mistaken for one. This is separate from import dependency-cycle detection.
+For host sources, discovery uses rooted `Lstat` and canonical containment.
+`filesystem_context.go` owns resolved access facts and live access setup; the
+acquisition operation closes its live handles before composition. Native sources
+use `fs.ReadLinkFS` inspection when supplied; otherwise traversal follows the
+capabilities and behavior of the provided `fs.FS`, without claiming a general
+sandbox guarantee. `source_context.go` derives passive resource contexts from
+resolved sources for later caller-owned reads; those records do not retain live
+acquisition handles.
+
+Host and native link bindings share one iterative link-frame evaluator. It
+tracks active links for cycle detection and reuses successfully completed link
+targets while resolving the remaining path components; each binding retains its
+own namespace and target-containment rules. Directory reads distinguish an
+initial-open failure from failures after opening: only an initial missing
+directory can be treated as optional, while enumeration and close failures
+remain selection errors, including errors from the standard `fs.ReadDir`
+fallback.
+
+Glob selection keeps traversal coordination in `selectGlobImportCandidates`:
+it advances literal and patterned segments, reads directories, unions and sorts
+candidates, validates final targets, then applies exclusions and requiredness.
+For each patterned directory entry, `prepareGlobEntry` handles matching and
+hidden filtering, containment, target and link inspection, intermediate-directory
+eligibility, host/native access-path selection, and candidate identity and label.
+It returns a candidate, an ordinary skip, or a diagnostic without mutating the
+selector's candidate set or traversal state. Directory entry preparation remains
+separate because its matching, absence, recursion, and traversal-state rules
+differ.
+
+Acquisition remains authored: declarations are visited in source order and each
+selected candidate set in namespace-path lexical order. This preserves source
+records, IDs, source-local version checks, and callback ordering. The acquired
+inclusion graph is retained separately from composition scopes so cycles that
+cross item boundaries remain detectable. Source discovery does not itself choose
+sequence assembly order.
+
+## Stable keys and scheduling
+
+The shared scheduler compares immutable tuples component by component using Go's
+bytewise string order; if one tuple is a prefix of another, the shorter tuple
+sorts first. The import key is
+`(namespace-kind, namespace-anchor, resource-kind, resource-path)`:
+
+- Namespace kind is `host` or `native`. Host resource paths are canonical
+  absolute paths rendered with slash separators; native paths retain their
+  namespace-relative identity.
+- Anchors come from explicit input positions, not discovery order: main is
+  `(0,0)` and underlay/input positions are `(group+1,input+1)`, each encoded as
+  fixed-width decimal. Host accesses in one scheduled tier share its host anchor.
+  A native filesystem namespace uses its earliest explicit selecting position
+  in that tier. Filesystem tokens distinguish namespace equality, not sort order,
+  so separately constructed filesystems with the same label remain distinct.
+- Resource kind is `file` or `synthetic`. A synthetic root uses its explicit
+  input anchor and an empty resource path rather than a display label or token
+  address.
+
+Inheritance keys are
+`("definition", scope-pointer, definition-pointer)`, where both pointers use
+canonical escaped semantic path spelling. Source IDs are not part of the key.
+
+Both integrations use the same Kahn scheduler: it deduplicates dependency edges,
+keeps currently ready nodes in a min-heap, and selects the smallest ready key
+after every completion. Its scheduling cost is `O(E + V log V)` time and
+`O(V + E)` additional memory, excluding key construction, tuple copying, and
+tuple comparisons. On a stalled graph, deterministic DFS over the residual graph
+returns a closed cycle rotated to its smallest key; blocked dependents are not
+reported as cycle members.
+
+Import edge preparation builds an importer-to-target adjacency index once for
+an acquisition's outer and item-local scheduling, and once for grouped
+scheduling. Each scope gathers edges from its participating identities and
+retains only edges with both endpoints in its local slots. The retained
+inclusion graph still owns cycle diagnostics and source relationships.
+
+Import composition schedules a main tier or one underlay group as a single
+strict-scope graph. Applicable collection declarations are gathered for the
+scope before peer composition, including identity inherited from lower tiers.
+Identified peer sequences then combine distinct records in assembled source and
+list order; after the strict scope completes, layered assembly applies the
+resolved collection operation against the lower value, as defined by the
+[configuration contract](../config.md#strict-document-composition). Item
+contributions own nested local strict-mapping scopes; their helpers complete
+before the item is placed in its outer scope. Inheritance uses the same
+ready-node rule inside each sibling mapping scope, while preserving the authored
+mapping order in its output.
+These scopes do not flatten into one global graph, and independent precedence
+tiers remain separate.
+
+The owners are [`acquisition.go`](../../internal/engine/acquisition.go),
+[`import_selection.go`](../../internal/engine/import_selection.go),
+[`import_acquisition.go`](../../internal/engine/import_acquisition.go),
+[`directory_walk.go`](../../internal/engine/directory_walk.go),
+[`import_graph.go`](../../internal/engine/import_graph.go),
+[`import_projection.go`](../../internal/engine/import_projection.go),
+[`dependency_order.go`](../../internal/engine/dependency_order.go), and
+[`inheritance_schedule.go`](../../internal/engine/inheritance_schedule.go).
+Their focused behavior is covered by import-selection, acquisition, scheduler,
+and scheduling integration tests under `internal/engine`; see the
+[development test map](../development.md#task-to-test-map).

docs/internal/overview.md

@@ -7,20 +7,44 @@ private state; public signatures and fields do not expose engine types.
 
 Implemented component owners:
 
-- `internal/engine` owns host and native-filesystem source acquisition,
-  grouped strict composition, recursive precedence assembly, and destination-local
-  sibling inheritance before typed decoding. It also owns schema
+- `internal/engine` owns host and native-filesystem source acquisition and
+  explicitly selected configuration discovery. A search resolves once during
+  shared acquisition; the selected concrete file or absent-input state then
+  follows the existing composition, metadata, and resource-context lifecycle.
+  It also owns grouped strict composition, recursive precedence assembly,
+  per-input mapping and sequence-item placement, YAML collection declarations
+  and operations for identified sequences, and application identity registrations for strict
+  composition, inheritance, and final validation. It applies the selected
+  collection operation at precedence boundaries and owns dependency-ready,
+  destination-local inheritance before typed decoding. CLI assignment follows
+  inheritance and decoding, before application resolution. The engine also owns
+  shared source-version checking with sanitized declaration evidence.
+  Explicit import discovery supports declared paths, directory searches, and
+  globs; root-owned defaults are converted to the engine's single defaults input
+  and snapshotted during load preflight before acquisition, while source-local
+  empty-selection evidence crosses the facade without changing acquisition
+  order. The root constructs placement directly on engine options and inputs
+  using explicit enum conversion; public types and fields remain root-owned.
+  Import and inheritance resolution share dependency-ready scheduling within
+  their composition scopes.
+  It retains each source's effective placement alongside physical source
+  identity and owns schema
   inspection, lifecycle coordination, callback auditing,
   provenance/history, passive source-context retention, secret resolution and
-  bootstrap, and safe inspection rendering. It never imports the root package.
+  bootstrap, operation-local secret failure-location attribution, safe
+  inspection rendering, and effective-value comparison over
+  frozen inspection snapshots. It never imports the root package.
 - Root `Source`, `Filesystem`, `Input`, `Underlay`, `Path`, `Secret`, records,
   lifecycle callbacks, results, and bootstrap reports are root-defined wrappers
   or records. Callback wrappers pass only live engine context during a callback;
   `Result.Config` is detached and its private inspection remains frozen
   independently.
-- Root conversion files explicitly cross the path, record, callback, secret, and
-  error boundaries. There are no public aliases, anonymous engine embeddings, or
-  public engine bridge methods.
+- Root conversion files explicitly cross the path, record, callback, secret,
+  search-description, and error boundaries. There are no public aliases,
+  anonymous engine embeddings, or public engine bridge methods.
+- Root `Diff` and `Difference` provide the public comparison boundary and
+  convert detached engine records to root-owned paths and snapshots; consumer
+  examples own presentation of the safe changed regions.
 - `internal/document` owns the parser-independent ordered tree and cloning.
   `internal/yaml` is the sole upstream YAML adapter and parser/dumper boundary.
 - Engine mechanism tests own source, composition, inheritance, decoding,
@@ -37,10 +61,100 @@ these constraints and rejects exported aliases, fields, or signatures involving
 internal types.
 
 The [configuration contract](../config.md) owns implemented YAML and lifecycle
-semantics. The [consumer adoption guide](../consumers/README.md) owns application
-guidance. The [architecture policy](../policy/architecture.md) owns contributor
-boundaries, and [ADR 0006](../adr/0006-public-facade.md) records the facade
+semantics, including
+[configuration discovery](../config.md#configuration-discovery),
+[input destinations](../config.md#input-destinations),
+[shared source-version checks](../config.md#shared-source-version-checks), and
+[effective configuration diffs](../config.md#effective-configuration-diffs).
+The [strict composition contract](../config.md#strict-document-composition)
+owns sequence declarations, identity, operations, ownership, and final
+validation; [CLI overrides](../config.md#explicit-cli-overrides) own override
+ordering. The [consumer guide](../consumers/README.md#identified-prompt-sequences)
+links maintained workflow examples.
+The [consumer adoption guide](../consumers/README.md) owns application guidance.
+The [architecture policy](../policy/architecture.md) owns contributor
+boundaries; [ADR 0006](../adr/0006-public-facade.md) records the facade decision
+and [ADR 0010](../adr/0010-effective-configuration-diff.md) records the diff
+decision; [ADR 0011](../adr/0011-input-destinations.md) records the placement
 decision. See the [development guide](../development.md) for validation commands
 and [release procedure](../release.md) for source-candidate gates. [ADR
 0009](../adr/0009-sibling-inheritance.md) records why inheritance is a built-in
-post-assembly structural operation.
+post-assembly structural operation. [ADR 0012](../adr/0012-relative-import-destinations.md)
+records the relative-import extension; [ADR 0013](../adr/0013-shared-source-version-checks.md)
+records the source-version decision.
+[ADR 0014](../adr/0014-sequence-placement-and-composition.md) records explicit
+sequence placement and composition. [ADR 0015](../adr/0015-shared-dependency-ready-ordering.md)
+records the shared dependency-ready ordering decision; [ADR
+0016](../adr/0016-independent-import-controls.md) records the independent
+selection controls. [ADR 0017](../adr/0017-collection-sequence-policy.md) and
+[ADR 0018](../adr/0018-cli-after-structural-resolution.md) record collection
+operations and CLI ordering. [ADR 0019](../adr/0019-explicit-configuration-discovery.md)
+records explicit main-file discovery. The [import internals](imports.md)
+document candidate and namespace identities, discovery and scheduling, and the
+bounded algorithms.
+
+## Implementation flow
+
+The root facade converts caller-owned values into engine-owned requests. Its
+`ImportDefaults` conversion maps fields without taking ownership; engine load
+preflight snapshots the defaults once before source acquisition or callbacks.
+Ordinary and grouped loading then carry that read-only value through their
+normalized `loadRequest` and acquisition policy, while declaration resolution
+copies values needed by its independently owned effective policy. Placement is
+constructed directly on the engine's options and input records after explicit
+root-to-engine enum conversion. See
+[`lifecycle.go`](../../lifecycle.go), [`layered.go`](../../layered.go), and
+[`load_request.go`](../../internal/engine/load_request.go). Load-wide definition,
+version, default-import, and override policy stays distinct from each input's
+source, filesystem, destination, and import settings.
+
+[`acquisition.go`](../../internal/engine/acquisition.go) defines acquisition
+policy, load-wide accounting, and tier-scoped repetition/namespace state. Each
+input is acquired under that policy; grouped requests retain acquired underlay
+scopes until their tier is composed in
+[`load_request.go`](../../internal/engine/load_request.go). Completed tiers then
+share one assembly handoff that applies collection declarations and projects
+only that tier's source/version evidence; the caller keeps main-version
+protection and its projection explicit. Live
+filesystem access handles are confined to acquisition and closed there. Resolved
+filesystem facts used to authorize access are separate from passively retained
+resource contexts. Non-import validation retains those facts on `assembledInput`
+for passive-context construction and ordinary-overlay handling: see
+[`filesystem_context.go`](../../internal/engine/filesystem_context.go) and
+[`source_context.go`](../../internal/engine/source_context.go).
+
+Import responsibilities are split across candidate selection and path
+normalization in [`import_selection.go`](../../internal/engine/import_selection.go),
+source reading and traversal admission in
+[`import_acquisition.go`](../../internal/engine/import_acquisition.go),
+directory walking in [`directory_walk.go`](../../internal/engine/directory_walk.go),
+dependency graph scheduling in [`import_graph.go`](../../internal/engine/import_graph.go),
+and projection of acquired evidence into final source coordinates in
+[`import_projection.go`](../../internal/engine/import_projection.go). The
+[import internals](imports.md) page describes the boundaries and identities.
+
+After acquisition handles are closed, grouped composition and inheritance run on
+the retained ordered documents. `inheritance.go` coordinates traversal and
+[`inheritance_schedule.go`](../../internal/engine/inheritance_schedule.go) applies
+dependency-ready templates;
+[`inheritance_prepare.go`](../../internal/engine/inheritance_prepare.go)
+prepares child and parent inputs,
+[`inheritance_combine.go`](../../internal/engine/inheritance_combine.go)
+returns the combined value with its origin, policy, sequence, and diagnostic
+evidence, and [`inheritance_finalize.go`](../../internal/engine/inheritance_finalize.go)
+accounts for and finalizes the result. Projection and schema work have separate
+operation owners: [`value_operation.go`](../../internal/engine/value_operation.go)
+holds per-operation state, [`projection.go`](../../internal/engine/projection.go)
+builds projected values, [`schema.go`](../../internal/engine/schema.go) owns
+descriptor caching and type policy, and
+[`schema_fields.go`](../../internal/engine/schema_fields.go) owns ordinary and
+inline field assembly. [`decode.go`](../../internal/engine/decode.go) consumes
+those descriptors for typed decoding.
+
+Behavioral ownership follows the same handoffs: imports and grouped lifetimes are
+covered by `imports_test.go`, `load_request_test.go`, and focused source,
+discovery, and destination tests; inheritance by the inheritance test files;
+reflection behavior by `schema_test.go`, `projection_test.go`, and
+`decode_test.go`. The [development test map](../development.md#task-to-test-map)
+links focused commands. [`clone.go`](../../internal/engine/clone.go) owns typed
+copying, using the shared descriptors and per-operation traversal helpers.

docs/oss/attribution.md

@@ -2,11 +2,12 @@
 
 ## Standard Repository Policies
 
-Material: `docs/policy/meta.md`, `docs/policy/oss.md`,
+Material: `docs/policy/release.md`, adapted from
+`e833faddf120c789a9b45348edaf75720de16348`; `docs/policy/meta.md`, `docs/policy/oss.md`,
 `docs/policy/documentation.md`, `docs/policy/agents.md`,
-`docs/policy/architecture.md`, `docs/policy/testing.md`, and
-`docs/policy/release.md`; root `AGENTS.md`; and the Policy Reading Guide in
-`docs/development.md`, adapted for this repository.
+`docs/policy/architecture.md`, `docs/policy/testing.md`; root `AGENTS.md`; and the
+Policy Reading Guide in `docs/development.md`, adapted from
+`d84a38ebab26f048cafbe59a2246700c35d29843` for this repository.
 
 Copyright (C) 2026 Eric Rakestraw
 
@@ -14,10 +15,14 @@ SPDX-License-Identifier: CC-BY-SA-4.0
 
 License and notices: [CC BY-SA 4.0](licenses/CC-BY-SA-4.0.txt) |
 [COPYRIGHT](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/COPYRIGHT) |
-[LICENSE](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/LICENSE).
+[LICENSE](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/LICENSE) |
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/policy/src/commit/e833faddf120c789a9b45348edaf75720de16348/COPYRIGHT) |
+[LICENSE](https://gitea.maximumdirect.net/eric/policy/src/commit/e833faddf120c789a9b45348edaf75720de16348/LICENSE).
 
 [Source](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843).
 
+[Source](https://gitea.maximumdirect.net/eric/policy/src/commit/e833faddf120c789a9b45348edaf75720de16348).
+
 ## go-yaml v4
 
 Selected module: `go.yaml.in/yaml/v4` at `v4.0.0-rc.6`.

docs/policy/architecture.md

@@ -2,7 +2,7 @@
 
 **Base Policy Version:** 1.3
 
-**Last Updated**: 2026-09-29
+**Last Updated**: 2026-10-22
 
 **Upstream Source:** [Standard Repository Policies](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/docs/policy/architecture.md)
 
@@ -19,8 +19,8 @@ to ADRs for decision rationale.
 ConfigKit is an opinionated Go library for consistent YAML configuration loading
 and resolution across consuming applications. The [component overview](../internal/overview.md)
 owns implementation inventory, the [configuration contract](../config.md) owns
-implemented behavior, and the [deferred scope](../roadmap/deferred.md) preserves
-deferred scope.
+implemented behavior, and the [future feature list](../roadmap/future.md) preserves
+candidates for future consideration.
 
 ## Adopted Constraints And Implemented Structure
 
@@ -32,6 +32,12 @@ structure as implemented when its changes land.
 Constraints must reflect deliberately selected requirements and decisions.
 Do not promote incidental implementation choices into architectural invariants.
 
+The implemented collection-policy and CLI-ordering constraints below reflect
+accepted [ADR 0017](../adr/0017-collection-sequence-policy.md) and
+[ADR 0018](../adr/0018-cli-after-structural-resolution.md). The configuration
+contract owns their current behavior; these architectural constraints describe
+the system boundaries and guarantees that guide changes to that behavior.
+
 ## Success Criteria
 
 ConfigKit prioritizes a coherent configuration model: consistency and
@@ -91,17 +97,82 @@ boundaries, and keep simple consumers independent of optional composition
 features. The [consumer adoption guide](../consumers/README.md) demonstrates the
 implemented surface against representative simple and complex shapes.
 
+## Simplicity, Defaults, And Explicit Controls
+
+These principles apply throughout ConfigKit: to configuration syntax, the public
+Go API, defaults, and feature design. Evaluate their effects on both configuration
+authors and consuming applications.
+
+**Keep straightforward workflows straightforward.** Common configuration
+workflows should require little configuration text and minimal API setup.
+Support more demanding workflows through explicit, composable options, without
+requiring simple consumers to understand or configure those options.
+
+**Prefer simple, permissive defaults with explicit controls.** Choose predictable
+defaults that minimize required setup and accommodate valid configurations
+without unnecessary restrictions. Provide selectable options for workflows that
+need additional capabilities, narrower behavior, or stronger requirements.
+
+Measure simplicity by what consumers must understand and maintain, including
+configuration text and application-side plumbing. Concise syntax is useful, but
+hidden inference and interacting meanings can make a shorter declaration harder
+to understand. Prefer explicitness where it makes behavior more predictable.
+A terse configuration format that requires substantial application-side
+infrastructure does not satisfy the simplicity principle.
+
+Define what permissiveness means for each behavior. Allowing an optional
+contribution to be absent or accepting valid inputs without an unnecessary
+filename restriction can reduce required setup. Invalid declarations, ambiguous
+composition, and unreadable selected sources must still produce meaningful
+errors under the applicable contracts. Permissiveness does not authorize silent
+recovery from invalid supplied configuration or weaken configuration integrity.
+
+Options should express coherent choices with understandable purposes and fit
+ConfigKit's shared configuration model. Support powerful controls when they
+serve concrete needs; these principles do not require a switch for every
+implementation behavior or conceivable consumer preference.
+
+Defaults require judgment when simplicity and permissiveness pull in different
+directions. Broader discovery can accommodate more layouts while increasing the
+implicit search scope. Evaluate such tradeoffs explicitly; these principles do
+not mechanically select the broadest behavior for every option. Changes to
+existing defaults remain subject to compatibility review, and proposed behavior
+remains in the roadmap until implemented.
+
 ## Responsibility Boundaries
 
 ConfigKit owns configuration input processing and the coordination of decoding,
 resolution, validation, and inspection. Applications own their configuration
 types, field meanings, defaults, normalization, semantic validation, and schema
-version policy. Applications also own main-file search conventions, CLI wiring,
-remote acquisition, runtime initialization, and resource checks.
+version policy. Applications choose and order their main-file search conventions;
+ConfigKit provides explicitly selected search descriptions that resolve during
+loading. Applications also own CLI wiring, remote acquisition, runtime
+initialization, and resource checks. See the [configuration discovery
+contract](../config.md#configuration-discovery) and [ADR
+0019](../adr/0019-explicit-configuration-discovery.md).
+
+The engine applies each input's mapping destination within the existing source
+assembly, applies immediate-parent-relative YAML destinations to imports, and
+retains each effective placement with source identity and inspection records;
+the root facade owns the public fields and conversions. The
+[configuration contract](../config.md#input-destinations) defines the observable
+placement behavior. [ADR 0011](../adr/0011-input-destinations.md) records input
+placement, and [ADR 0012](../adr/0012-relative-import-destinations.md) records
+the import extension.
 
 Library code must not depend on consuming applications or initialize their
-runtime resources. It must not perform implicit network access, discover
-application-specific paths, or change process-global state during ordinary `Load`.
+runtime resources. It must not perform implicit network access, implicitly
+discover application-specific paths, or change process-global state during
+ordinary `Load`.
+Import discovery is available only through declarations in selected sources:
+explicit `path`, `dir`, and `glob` selectors may search their declared scope, but
+ConfigKit does not search conventional application paths unless the application
+explicitly supplies a discovery description, and it never discovers a main
+configuration file implicitly. Imports and sibling inheritance share
+dependency-ready scheduling within their existing scopes; acquisition order and
+precedence-tier boundaries remain distinct. The [configuration contract](../config.md#explicit-imports)
+owns observable behavior; [ADR 0015](../adr/0015-shared-dependency-ready-ordering.md)
+and [ADR 0016](../adr/0016-independent-import-controls.md) record the decisions.
 The explicit secrets-directory bootstrap is a narrow, opt-in exception: applications
 invoke it during startup, and it reports writes, skips and partial failures. It
 must never run implicitly as part of configuration loading.
@@ -124,6 +195,29 @@ The [configuration contract](../config.md#completion-and-resource-lifetime) owns
 the precise completion and ownership behavior. The proposed decision record is
 [ADR 0008](../adr/0008-complete-on-return-loading.md).
 
+### Accepted Processing Order
+
+Complete source composition and precedence, including selected overlays, before
+sibling inheritance. Then apply typed defaults and decode the configuration.
+Apply explicit CLI overrides to that configuration and check their types before
+running application resolution. Effective secret lookup and final validation
+follow resolution, before publication. [ADR 0018](../adr/0018-cli-after-structural-resolution.md)
+records this ordering.
+
+CLI overrides are the highest-priority input to application resolution. Supplied
+sequences replace whole sequences, and mappings combine recursively. Inheritance
+and typed defaulting do not run again after CLI application. CLI paths and values
+must not introduce structural directives; earlier source, inheritance, and
+decoding failures remain errors even if a CLI assignment would replace the value.
+Semantic validation assesses the final configuration after overrides and resolution.
+
+Applications own flag parsing and domain resolution. ConfigKit owns override
+application, validation, and source attribution. Resolution receives CLI values
+with their explicit-presence and source information. Absence-based defaults
+preserve explicit values; recorded normalization and derivation can still change
+them. Typed defaults and application resolution retain their own assignment
+semantics, independently of collection operations during source assembly.
+
 ## Public API And Dependency Direction
 
 Keep one small, coherent public surface for the supported lifecycle. Simple
@@ -152,31 +246,44 @@ framework on this library.
 
 ## Configuration Integrity
 
-Within a configuration tier, independent contributions use strict composition:
-shared mappings combine recursively, while overlapping non-mapping definitions
-are conflicts, including equal values. Between precedence tiers, intentional
-overlay combines mappings recursively, retains omitted lower-priority children,
-and replaces other present values according to the existing overlay rules.
-Tier precedence is explicit; traversal order within a tier establishes stable
-reporting, not permission to override. The distinction and its rationale are
-recorded in [ADR 0001](../adr/0001-separate-composition-and-overlay.md).
+Within a configuration tier, independent contributions use strict composition.
+Shared mappings combine recursively. Sequences with an applicable identity
+combine distinct records and reject duplicate identities, including equal
+records. Other overlapping non-mapping values remain conflicts, including equal
+values. Conflicting identity or operation declarations among peers also fail.
+Traversal order establishes stable reporting and record order; it does not grant
+precedence to peers. [ADR 0001](../adr/0001-separate-composition-and-overlay.md)
+records the distinction between strict composition and intentional overrides.
 
 The main document and its imports form a strictly composed tier. Explicit
-underlay groups compose strictly within each group, then combine recursively
-from lowest to highest beneath the main tier. Source acquisition kind does not
-determine its precedence role. Input-specific constraints, including CLI path
-rules and opt-in version ownership, remain part of the configuration contract. Consistent combination
-semantics do not require identical acquisition or lifetime behavior from every
-source kind.
+underlay groups compose strictly within each group, then combine from lowest to
+highest beneath the main tier. Between tiers, mappings combine recursively and
+retain omitted lower-priority children. Sequence values follow collection policy;
+other present values retain the established replacement rules. Source acquisition
+kind does not determine precedence. Input-specific constraints, including version
+ownership and path protections, remain part of the configuration contract.
+Consistent combination semantics do not require identical acquisition or lifetime
+behavior from every source kind.
 
-After source precedence is complete, built-in sibling `extends` relationships
-resolve structurally in the same parser-independent representation, using the
-existing recursive overlay rules. This adds neither an acquired source nor a
-consumer registration mechanism. Enclosing relationships precede nested
-destination-local resolution; typed defaults, application resolution, and
-secret lookup retain their existing lifecycle roles. The [configuration
-contract](../config.md#sibling-inheritance) owns the exact syntax and presence
-rules, and [ADR 0009](../adr/0009-sibling-inheritance.md) records the rationale.
+Built-in sibling `extends` relationships resolve after source precedence in the
+parser-independent document representation. Preserve existing mapping-sibling
+scopes and allow an identified sequence to provide a sibling scope for its
+records. A record may inherit from one parent in that sequence, selected by
+complete identity. The child supplies and keeps its own identity while inheriting
+the parent's other settings. Parents remain ordinary records subject to normal
+validation.
+
+Use the shared dependency ordering and preserve assembled sequence order.
+Resolve enclosing relationships before nested destination-local relationships.
+Mappings combine recursively; sequences use the inheriting collection's policy.
+Whole-record upsert replaces records during combination; explicit inheritance
+uses the records that remain. Inheritance remains built-in and creates no
+acquired source.
+[ADR 0009](../adr/0009-sibling-inheritance.md) records the
+original scope and continuing ownership decisions, with sequence combination
+and identified-sequence inheritance defined by
+[ADR 0017](../adr/0017-collection-sequence-policy.md) and CLI ordering by
+[ADR 0018](../adr/0018-cli-after-structural-resolution.md).
 
 ConfigKit must define supported YAML semantics explicitly rather than inherit
 parser defaults. Processing order and diagnostics must be deterministic for the
@@ -185,13 +292,14 @@ configuration. Do not run application hooks on invalid typed configuration.
 
 User-controlled precedence is application defaults, ordered underlays, composed
 main input, selected overlay, then explicitly supplied CLI overrides.
-Applications own flag parsing; ConfigKit owns overlay semantics and provenance.
-Domain resolution follows these
-inputs and must remain recorded, not become a hidden override channel.
+This priority order is distinct from processing order: typed defaults are applied
+at decoding, and application resolution follows CLI input under the lifecycle
+rules above.
 
-Preserve node-based composition until the effective input is ready for typed
-decoding. Preserve presence and source information through processing. Ordinary Go zero
-values must not be used to infer whether input was absent. Final configuration,
+Preserve node-based composition through source assembly and inheritance until
+typed decoding. Preserve presence and source information through later CLI and
+application-resolution changes as well. Ordinary Go zero values must not be used
+to infer whether input was absent. Final configuration,
 metadata, and effective inspection must agree, including application resolution
 changes; see [ADR 0003](../adr/0003-preserve-presence-and-resolution-history.md).
 Safe historical values and an explanation renderer are part of the implemented contract,
@@ -201,6 +309,79 @@ See [ADR 0005](../adr/0005-explicit-overrides-and-safe-explanation.md).
 Detailed syntax and operation tables belong to the
 [current configuration contract](../config.md).
 
+The frozen inspection boundary also supports comparing completed same-type
+results without reopening sources or rerunning the load lifecycle. The
+[configuration contract](../config.md#effective-configuration-diffs) owns
+comparison and disclosure behavior, the [component overview](../internal/overview.md)
+owns implementation responsibility, and [ADR 0010](../adr/0010-effective-configuration-diff.md)
+records the accepted decision.
+
+## Collection Identity And Combination
+
+ConfigKit owns a common collection policy for imported records, inline sequences,
+selected overlays, and sibling inheritance. Configuration authors may declare
+identity and combination rules; applications may register authoritative identity
+constraints. Configuration declarations must agree with application constraints.
+Keep source placement explicit and separate from policy, retaining the placement
+decision in [ADR 0014](../adr/0014-sequence-placement-and-composition.md).
+[ADR 0017](../adr/0017-collection-sequence-policy.md) owns the rationale for
+collection policy and its relationship to application authority.
+
+Gather declarations for the relevant destination and source or item-local scope
+before strict composition. Applicable identity from accumulated lower tiers must
+be available when composing higher-tier peers. Higher-tier declarations cannot
+repair invalid lower-tier composition. Preserve the distinction between source
+acquisition, dependency-ready scheduling within scopes, and precedence order.
+
+Strictly compose each tier's contributions before applying one operation for the
+collection against its lower value. Support whole-sequence replacement,
+duplicate-rejecting append, and whole-record replacement by identity (`upsert`).
+Append and upsert require identity; do not infer it from record fields or position.
+Duplicate identities among peers fail under every mode. Source precedence and
+inheritance use the same operations and deterministic ordering rules. Whole-record
+upsert replaces the matching record's fields together; recursive record merging
+and unkeyed sequence concatenation are outside this operation model.
+
+An explicit mode persists at its applicable destination until a higher explicit
+mode changes it. With no explicit mode, effective identity selects upsert and
+its absence selects whole-sequence replacement. Inferred defaults do not become
+persistent instructions. The inheriting destination chooses its operation; copying
+a parent collection does not automatically copy its mode to that destination.
+
+Track YAML identity and explicit modes according to their separate lifetimes.
+Actual whole-value replacement discards lower YAML identity while preserving
+application constraints and any new identity declaration. Item-local policies
+belong to their records: retaining a record retains its rules, and discarding it
+discards those rules. Policies independently applicable at a destination remain
+in effect. Inheritance can copy YAML identity subject to agreement and replacement
+rules; application registrations continue to constrain their registered paths.
+Numeric positions alone do not establish policy ownership.
+
+Allow declarations without accompanying records. A key declaration validates a
+retained sequence immediately or remains pending in its owning scope when no
+sequence exists. A mode-only declaration affects later combinations without
+recombining earlier tiers. Neither declaration alone clears a collection, and
+declaring replace alone does not authorize changing a retained collection's
+identity. These rules do not independently impose path-existence or shape
+requirements.
+
+Enforce YAML identity during source composition, precedence, and identified
+inheritance operations. Application registrations additionally constrain final
+effective sequences after defaults, CLI overrides, and application resolution.
+The application schema and validation own existence, nullability, and allowed
+shape. The final identity guarantee covers actual sequences on successful load;
+it does not cover every intermediate callback state or later caller mutations.
+YAML-only declarations carry no such final guarantee through defaults, CLI
+assignments, or application resolution.
+
+Preserve source attribution, presence, history, sensitivity, and resource context
+as records survive, move, or are replaced. Identity diagnostics use paths and
+available origin information without exposing identity values. The guarantee
+requires no new mutation-history system or exact reconstruction of causes for
+records supplied by defaults or callbacks. Exact identity equality, declaration
+syntax, presence cases, and operation tables belong in the
+[configuration contract](../config.md#strict-document-composition).
+
 ## Ownership And Confidentiality
 
 Returned configuration and metadata must not share mutable state across loads.
@@ -227,7 +408,7 @@ for safe messages supplied by their hooks.
 Do not introduce automatic directory discovery, network providers, arbitrary
 merge strategies, additional configuration languages, or application runtime
 management without concrete consumer evidence and an explicit design decision.
-The [deferred scope](../roadmap/deferred.md#explicit-non-goals-and-deferred-scope)
-owns the deferred-feature inventory. Significant boundary
-changes require an ADR and corresponding policy/contract updates; implementation
-convenience alone is not a reason to broaden the public API.
+Record significant decisions to exclude capabilities, and their rationale, in
+ADRs. Significant boundary changes require an ADR and corresponding
+policy/contract updates; implementation convenience alone is not a reason to
+broaden the public API.

docs/policy/documentation.md

@@ -2,7 +2,7 @@
 
 **Base Policy Version:** 1.11
 
-**Last Updated**: 2026-09-28
+**Last Updated**: 2026-10-16
 
 **Upstream Source:** [Standard Repository Policies](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/docs/policy/documentation.md)
 
@@ -38,9 +38,10 @@ roadmap owns implementation status and sequencing until the decision is
 implemented. Current architecture, user, operator, integration, and internal
 documentation are updated when the behavior lands.
 
-`docs/roadmap/deferred.md` owns deliberate exclusions; task-specific plans
-own execution order and progress. Completed plans retire during release
-preparation.
+`docs/roadmap/future.md` owns potential enhancements and their open questions;
+task-specific plans own execution order and progress. Record significant
+architectural exclusions and their rationale in ADRs. Do not maintain a roadmap
+inventory of rejected proposals. Completed plans retire during release preparation.
 
 ### Audience And Detail
 
@@ -91,7 +92,7 @@ secret values.
 | Inventory of implemented internal packages | `docs/internal/overview.md` | Implemented packages and components, responsibilities, and links to focused internal docs. | Normative architecture, contributor reading policy, external contracts. |
 | Behavior and architecture of internal packages | Other files under `docs/internal/` | Implementation flow, collaborators, state transitions, package-local guarantees and failures, and relevant tests. | Global architecture invariants, configuration definitions and defaults, external schemas, operator procedures. |
 | Architectural decision records | `docs/adr/` | Significant decisions, context, alternatives, rationale, consequences, and supersession history. | Current behavior reference, implementation status, task sequencing. |
-| Deferred scope | `docs/roadmap/deferred.md` | Deliberate exclusions and conditions for reconsideration. | Task execution and current behavior. |
+| Potential future features | `docs/roadmap/future.md` | Candidate enhancements and open questions for future consideration. | Rejected proposals, architectural decisions, task execution, and current behavior. |
 | Future implementation plans | Task-specific files under `docs/roadmap/` | Unfinished work, sequencing, acceptance, and progress; completed plans retire during release preparation. | Implemented behavior and ADR rationale. |
 | Release-note history | `docs/releases/` | One checked-in summary per source release under the procedure; the note at its immutable tag is the release record. | Current commands, behavior, contracts, and compatibility definitions. |
 | Complete examples | `examples/` and executable Go examples | Maintained copyable artifacts and Go API examples. | Configuration reference, symbol contracts, and prose explanation. |
@@ -156,8 +157,10 @@ Use sequentially numbered ADR filenames such as
 
 Treat the decision content of an accepted ADR as immutable. When a decision
 changes, create a new ADR and update the earlier ADR's status to superseded.
-Rejected architectural alternatives belong in the ADR; rejected product ideas
-belong in the roadmap. Routine choices need no ADR.
+Rejected architectural alternatives and significant architectural exclusions
+belong in ADRs. Roadmaps preserve plausible future work, not rejected product
+ideas. Routine choices and rejected suggestions need no ADR unless they
+represent a significant architectural decision.
 
 ## Maintenance
 

docs/policy/release.md

@@ -1,10 +1,10 @@
 # Release Policy
 
-**Base Policy Version:** 1.4
+**Base Policy Version:** 1.6
 
-**Last Updated**: 2026-09-28
+**Last Updated**: 2026-10-09
 
-**Upstream Source:** [Standard Repository Policies](https://gitea.maximumdirect.net/eric/policy/src/commit/d84a38ebab26f048cafbe59a2246700c35d29843/docs/policy/release.md)
+**Upstream Source:** [Standard Repository Policies](https://gitea.maximumdirect.net/eric/policy/src/commit/e833faddf120c789a9b45348edaf75720de16348/docs/policy/release.md)
 
 ## Purpose
 
@@ -12,6 +12,9 @@ This policy defines a proportionate workflow for validating and publishing an
 intended source revision under a fresh version. Releases should require little
 routine intervention, avoid repeated work, and preserve published contents.
 
+Adopt stronger verification or reproducibility requirements through an explicit
+project-specific decision justified by the intended audience and support needs.
+
 The repository-specific commands belong in `docs/release.md`. Historical release
 summaries belong in `docs/releases/`. The procedure must agree with the adopted
 policy and link to it rather than introduce additional release requirements.
@@ -54,10 +57,11 @@ authoritative primary branch, completes the release. The source commit may be
 pushed before the tag or published together with it through an atomic push.
 The procedure must choose and document the supported approach.
 
-ConfigKit uses sequential explicit pushes: first the captured candidate to
-authoritative `main`, then the annotated tag. Its
-[release procedure](../release.md) names `origin` and checks that its fetch and
-push destinations resolve to the authoritative URL.
+ConfigKit uses sequential publication: first push the committed candidate to
+authoritative `main`, then invoke ReleaseKit once to validate that candidate and
+publish its annotated tag. The [release procedure](../release.md) names
+`origin`; ReleaseKit checks its exact effective push URL before validation and
+tag publication.
 
 Release commands must finish at successful publication. They must not wait for
 or poll CI, require successful CI status, inspect downloadable artifacts, or
@@ -76,8 +80,9 @@ the tagged source. The procedure must identify those deliverables and the
 responsibility of their automation. Published deliverables must be traceable
 to that source and identify the same version.
 
-ConfigKit currently declares no additional deliverables. Its source-tag push
-is the completion point.
+ConfigKit currently declares no additional deliverables. Confirmed publication
+of the intended source tag is the completion point. CI, macOS validation,
+hosted delivery, and consumer retrieval are outside the release gates.
 
 ## Versioning And Compatibility
 
@@ -96,7 +101,8 @@ requires a deliberate update to the policy and procedure where needed.
 For releases at or after `v1.0.0`:
 
 - a major release may make intentional incompatible changes;
-- a minor release may add backward-compatible behavior; and
+- a minor release may add backward-compatible behavior or raise the Go consumer
+  minimum as described below; and
 - a patch release may make backward-compatible fixes.
 
 Before `v1.0.0`, a minor release may change a documented public, operational,
@@ -108,13 +114,53 @@ Additional compatibility promises and deliberate exceptions belong in the
 adopted policy or canonical contract documentation. Release notes explain
 changes; they do not redefine current compatibility contracts.
 
+## Consumer Minimum And Release Toolchain
+
+Distinguish the minimum toolchain supported for consumers from the toolchain
+maintainers use to build and validate a release. Record the consumer minimum in
+the language's standard metadata and explain its support meaning in canonical
+documentation. Define the selected release toolchain and its setup commands in
+`docs/release.md`. Toolchain selection and upgrade cadence are repository-specific
+choices.
+
+For Go repositories, normally declare the initial stable release of the selected
+series as the consumer minimum, such as `go 1.26.0`. Use a higher patch minimum
+when a dependency or an identified correctness fix requires it, and document
+the reason.
+
+For release-policy purposes, successful build and release validation on the
+selected Go patch release are sufficient to presume support across that Go
+series, subject to the declared consumer minimum and concrete evidence of
+incompatibility. Investigate specific compatibility concerns as they arise.
+
+Use the selected release toolchain for candidate validation and release builds,
+including CI jobs performing those tasks. A newer validation patch version alone
+leaves the consumer minimum unchanged.
+
+A minor release may raise the Go consumer minimum, including from one Go series
+to the next, both before and after `v1.0.0`. This is an explicit exception to the
+backward-compatibility requirement for minor releases. Patch releases preserve
+the consumer minimum. Explain an increase's impact and required consumer action
+in the release note. Other incompatible changes remain subject to the versioning
+rules above.
+
+### ConfigKit Toolchain Selection
+
+For minor and major releases, select the newest stable Go series available in
+the development environment. For patch releases, retain the series recorded in
+`go.mod` and its consumer minimum. Use the latest locally available compatible
+patch in the selected series, with `GOTOOLCHAIN=local`, to build the
+module-selected ReleaseKit CLI and validate the candidate. The same Go
+executable builds the tool and runs its validation commands.
+
 ## Release Documentation And Preparation
 
 Every new release must include `docs/releases/<version>.md` in the tagged
 source tree. Use the complete version tag as the filename and begin with a
 title containing the product name and that exact version.
 
-ConfigKit requires the first line `# ConfigKit <version>`.
+ConfigKit requires the first line `# ConfigKit <version>`. ReleaseKit's
+publication guard checks the release-note H1 for the product and version.
 
 Include a short summary and any relevant compatibility impact or upgrade
 instructions. Choose sections suited to the release and keep the note
@@ -146,10 +192,9 @@ checks publication state, and publishes the intended refs. A validation-only
 mode or command may be provided for diagnosis, but must not be a required step
 before publication repeats the same checks.
 
-ConfigKit uses `./scripts/release.sh vMAJOR.MINOR.PATCH`. The separate
-`scripts/release-check.sh` remains an optional diagnostic command; the
-publication command invokes it once and does not use a persistent candidate
-record.
+ConfigKit uses the module-selected ReleaseKit CLI's `publish
+vMAJOR.MINOR.PATCH` command. Its `check` command is optional for diagnosis; it
+is not a required preliminary pass to publication.
 
 Run the complete deterministic, offline default test suite once for the final
 candidate, following the [testing policy](testing.md). For repositories that
@@ -157,13 +202,17 @@ publish builds, verify the required build targets. Select those targets for
 the application's actual support commitments; native testing on each platform
 and access to particular hardware are not default release prerequisites.
 
-ConfigKit's one maintained Go 1.25 baseline runs the full suite with the race
-detector and test-time vet disabled. This protects library behavior, ownership,
-and concurrent use; the suite already includes examples, consumer fixtures,
-documentation links, and package boundaries. One full vet pass adds static
-checks beyond the test-time subset. CGO-disabled Linux amd64 and arm64 builds
-cover supported configurations distinct from native race compilation. macOS
-and newer-toolchain checks are optional.
+Prepare dependencies and build the tool before offline validation; the
+publication command is the required validation run, with no separate successful
+test or CI run required before it.
+
+The selected profile runs the complete deterministic, offline default suite
+once with the race detector and test-time vet disabled. The suite already
+includes examples, consumer fixtures, documentation links, and package
+boundaries. One full vet pass adds static checks beyond the test-time subset.
+CGO-disabled Linux amd64 and arm64 package compilation covers supported
+configurations distinct from native race compilation. macOS and newer-toolchain
+checks are optional.
 
 Additional gates must address a concrete risk that the existing checks do not
 adequately cover. Do not automatically require linting, race testing, module
@@ -172,9 +221,10 @@ expanded platform matrices solely because they are possible. Retain such
 checks when their distinct protection justifies their cost. A one-time migration
 check must not silently become a permanent release requirement.
 
-The selected Go commands use readonly module mode with `GOPROXY=off`; acquire
-the baseline toolchain and dependencies before validation. Formatting and
-whitespace remain ordinary development checks in `scripts/check.sh`.
+ReleaseKit runs its selected Go commands with readonly module mode and
+`GOPROXY=off`; acquire dependencies and build the tool before candidate
+validation. Formatting and whitespace remain ordinary development checks in
+`scripts/check.sh`.
 Review module tidiness when dependencies change, not as a release gate.
 
 Release tests must not require real credentials, mutable application services,
@@ -206,16 +256,19 @@ Automate the checks needed to publish the intended source under a fresh version:
   authoritative remote;
 - the worktree and index are clean on the primary release branch;
 - the source being published is the exact candidate that passed validation,
-  including its matching release note;
+  including its matching release note H1;
 - the remote primary branch can be advanced to that candidate without
   discarding remote history, and is published before the tag; and
 - the tag identifies the validated candidate and only the intended refs are
   published.
 
-ConfigKit also checks that `origin` has exactly the authoritative fetch and
-push destination, including configured push URLs. The command checks the
-candidate and unused version again after validation, and checks local state
-before creating the tag. Ignored local files do not block publication.
+ReleaseKit checks that `origin`'s effective push URL exactly matches the
+authoritative URL before candidate validation. This does not assert the
+configured fetch URL or require the release-note H1 to be the first line; the
+required `# ConfigKit <version>` first-line format remains a content rule. The
+command checks the candidate and unused version again after validation, and
+checks local state before creating the tag. Ignored local files do not block
+publication.
 
 Fetch the authoritative branch and tag state as needed for these checks.
 Do not force branch or tag updates to bypass a failed guard. Publish explicit
@@ -248,10 +301,10 @@ release without replacing its contents or reusing its version.
 ## Repository Release Procedure
 
 Maintain `docs/release.md` as a short, executable maintainer runbook. It must
-identify prerequisites, the primary branch and authoritative remote, version
-and note preparation, one publication command with local checks and guards,
-source-only deliverables, the successful-tag-push completion point, and
-partial or unknown-state recovery.
+identify prerequisites, the selected release toolchain, the primary branch and
+authoritative remote, version and note preparation, one publication command with
+local checks and guards, source-only deliverables, the successful-tag-push
+completion point, and partial or unknown-state recovery.
 
 Use copyable commands where practical. Keep optional diagnostics and consumer
 verification clearly optional. The procedure must not turn preparation work,

docs/release.md

@@ -1,7 +1,7 @@
 # ConfigKit Release Procedure
 
 This is the maintainer runbook for ConfigKit's [release policy](policy/release.md).
-ConfigKit publishes one source-only Go module. The immutable annotated tag and
+ConfigKit publishes a source-only Go module. The immutable annotated tag and
 matching checked-in release note are the deliverables; there is no binary,
 container image, generated package, or separate registry publication.
 
@@ -9,183 +9,221 @@ container image, generated package, or separate registry publication.
 
 - Module: `gitea.maximumdirect.net/eric/ConfigKit`
 - Primary branch: `main`
-- Authoritative remote: `origin`, with fetch and push destination
-  `https://gitea.maximumdirect.net/eric/ConfigKit.git`
+- Authoritative remote: `origin`
+- Authoritative push URL: `https://gitea.maximumdirect.net/eric/ConfigKit.git`
 - Tags: stable Semantic Versions such as `v0.5.0`; no prereleases
-- Minimum Go line: Go 1.25, validated with one maintained patch
+- Minimum Go version: the first stable patch of the selected Go series, or a
+  higher patch when required by a dependency or specific correctness fix, as
+  recorded in `go.mod`; see the
+  [consumer minimum policy](policy/release.md#consumer-minimum-and-release-toolchain)
 - Release-gated targets: Linux amd64 and arm64
 - Best-effort targets: macOS amd64 and arm64, under the
   [platform support policy](policy/release.md#platform-support)
 
-The checker runs one offline race suite, one full vet pass, and CGO-disabled
-builds for both Linux targets. It reports skipped filesystem cases. macOS checks
-and newer-toolchain tests are optional. CI is never a readiness or completion
-gate; do not wait for or poll it.
+Release validation runs one offline race-enabled Go suite, one full vet pass,
+and CGO-disabled package compilation for both Linux targets. macOS and CI are
+not release gates; do not wait for or poll CI.
 
-## 1. Prepare access and tools
+## 1. Select Go and prepare the tool
 
-Use a clean checkout with Git, Bash 3.2 or newer, a maintained Go 1.25 patch,
-and a native compiler for race tests. Configure private-host access through the
-normal credential manager. Keep credentials and environment dumps outside the
-repository and command transcript.
+Use a clean checkout with Git, Bash 3.2 or newer, and a native compiler for race
+tests. Choose the release type before selecting Go. For a minor or major release,
+select the newest stable Go series available in the development environment.
+For a patch release, retain the series recorded in `go.mod` and leave its consumer
+minimum unchanged. Follow the
+[toolchain selection policy](policy/release.md#configkit-toolchain-selection).
 
-Verify access without displaying credentials:
+Put the selected series' latest locally available compatible patch first on
+`PATH`, ensuring it meets `go.mod`'s minimum. The same `go` executable must build
+ReleaseKit and run its child validation commands. Set local toolchain selection
+and verify it:
 
 ```sh
-git remote get-url origin
-git ls-remote origin HEAD
+export GOTOOLCHAIN=local
+go version
+go env GOVERSION
+command -v go
 ```
 
-The fetch URL and all configured push destinations must resolve to the
-authoritative URL above. The publication command checks this again. Acquire the
-selected toolchain and dependencies explicitly before offline validation:
+Only when a minor or major release advances the consumer minimum to a newer Go
+series, derive that series' initial stable release from the selected executable:
 
 ```sh
-export CONFIGKIT_BASELINE_TOOLCHAIN=go1.25.14
-GOTOOLCHAIN="$CONFIGKIT_BASELINE_TOOLCHAIN" go version
-GOTOOLCHAIN="$CONFIGKIT_BASELINE_TOOLCHAIN" go mod download
+selected_go=$(GOTOOLCHAIN=local go env GOVERSION)
+selected_series=${selected_go#go}
+selected_series=${selected_series%.*}
+release_minimum="$selected_series.0"
+go mod edit -go="$release_minimum"
 ```
 
-A maintained patch may replace the example pin. No per-release latest-version
-survey is required. Dependencies must be cached; the checker sets `GOPROXY=off`.
-Publication requires user authorization. Standing authorization for local
-commits does not authorize a push or release.
+If a dependency or identified correctness fix requires a higher patch minimum,
+use that required version instead and document the reason. When the series is
+unchanged, leave the minimum unchanged unless a minor or major release
+deliberately changes it under the
+[consumer minimum policy](policy/release.md#consumer-minimum-and-release-toolchain).
+For example, validating with Go 1.26.8 leaves `go 1.26.0` unchanged. There is no
+separate requirement to validate with Go 1.26.0.
 
-## 2. Select the version and review compatibility
+Prepare dependencies whether or not the minimum changes:
 
-Choose an unused stable version. Compare the candidate with the previous tag.
-Review exported Go signatures and ownership, YAML and scalar semantics,
-precedence, diagnostic codes, minimum Go version, and platform support. A
-signature-preserving change may still affect consumers. Before `v1.0.0`,
-intentional incompatible minor-release changes need a note explaining impact
-and consumer action; patch releases remain compatible within their minor line.
-Review the maintained consumer examples and migration guide.
+```sh
+go mod download
+```
 
-When dependencies change, review parser behavior and migration impact as part
-of that change. Module tidiness belongs to dependency-change review, not the
-release checker. Do not repeat a dependency-version survey for every release.
+Build the tool selected by `go.mod` into temporary storage outside the checkout.
+Ensure `$TMPDIR` names a system temporary location, not the repository. These
+commands never install another ReleaseKit version:
 
-## 3. Write and commit the release note
+```sh
+releasekit_tmp=$(mktemp -d "${TMPDIR:-/tmp}/configkit-releasekit.XXXXXX")
+releasekit_bin="$releasekit_tmp/releasekit"
+go build -o "$releasekit_bin" gitea.maximumdirect.net/eric/ReleaseKit/cmd/releasekit
+go list -m gitea.maximumdirect.net/eric/ReleaseKit
+"$releasekit_bin" version
+```
 
-Create `docs/releases/<version>.md` for the selected version. Its first line
-must be `# ConfigKit <version>`. Give a meaningful summary and include
-compatibility impact or upgrade instructions when applicable. Link to current
-contracts for exact behavior. Review the note and affected canonical documents
-as ordinary preparation, then commit the complete candidate.
+The module-selected executable is ReleaseKit v0.4.2. Use the built executable
+for publication to preserve its exit status and publication-state diagnostics.
+`go run` maps a nonzero child exit status to 1. ReleaseKit's child Go commands
+use the `go` found on `PATH` with `GOTOOLCHAIN=local`; keep the selected
+executable first on `PATH` through the publication command.
 
-End every new note with these links, replacing `VERSION` with the complete tag:
+Verify access to the authoritative remote without displaying credentials:
+
+```sh
+git ls-remote https://gitea.maximumdirect.net/eric/ConfigKit.git HEAD refs/heads/main
+```
+
+Publication requires explicit authorization. Local implementation commits do
+not authorize a push or release.
+
+## 2. Prepare and commit the candidate
+
+Choose an unused stable version locally and on the authoritative remote. Compare
+the candidate with the previous tag and review exported Go signatures and
+ownership, YAML and resolution semantics, diagnostic contracts, minimum Go
+version, and platform support. Review maintained consumer examples and the
+migration guidance. Explain any consumer-minimum increase and required consumer
+action under the
+[consumer minimum policy](policy/release.md#consumer-minimum-and-release-toolchain).
+
+Create `docs/releases/<version>.md`. Its first line must be exactly
+`# ConfigKit <version>`, and it must end with the license and attribution footer:
 
 ```markdown
 [COPYRIGHT](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/VERSION/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/VERSION/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/VERSION/docs/oss/attribution.md)
 ```
 
-Confirm the three target files exist in the candidate and review the version in
-the URLs. The future tag need not resolve remotely yet; this review adds no
-network gate. ConfigKit uses checked-in notes without hosted release objects.
-If a hosted page is used later, reuse the note and preserve its footer. Do not
-rewrite historical notes or add downloadable license assets.
+Replace `VERSION` with the full tag and confirm the three target files exist in
+the candidate. Retire completed roadmaps, preserve unfinished and deferred
+work, and repair affected links. Review the complete candidate for unintended
+files and credentials, then commit it on `main`. These are ordinary preparation
+tasks, not separate approval gates. Any `go.mod` change belongs in this
+candidate.
+
+## 3. Push the candidate to `main`
+
+Push only that committed candidate to the literal authoritative URL. This does
+not follow local tags:
+
+```sh
+git push --no-follow-tags https://gitea.maximumdirect.net/eric/ConfigKit.git HEAD:refs/heads/main
+```
+
+ReleaseKit requires authoritative `main` to equal this candidate before it
+validates. A failure before this push means ReleaseKit publication has not
+started. If the push result is uncertain, inspect authoritative `main` and
+continue only when it is confirmed at the candidate. If validation later fails,
+no release tag is published; correct the issue in a new commit, push the new
+candidate to `main`, and invoke validation again.
 
 ## 4. Publish the committed candidate
 
-Before publication, review compatibility, release-note accuracy, and the file
-list for unintended content or credentials. Retire completed roadmaps as part of
-release preparation, preserving unfinished and deferred work and repairing
-links. These are ordinary preparation tasks, not separate human sign-offs.
-
-With the release explicitly authorized, run the one publication command.
-The script resolves the repository root from its own path, so it may be invoked
-from another directory by using its path there. Replace the sample version with
-the selected version:
+With the authorized candidate already on authoritative `main`, invoke the
+module-selected native executable exactly once:
 
 ```sh
-./scripts/release.sh v0.5.0
+"$releasekit_bin" publish vMAJOR.MINOR.PATCH
 ```
 
-The command requires `main`, a clean index/worktree, an unused version locally
-and on the authoritative remote, and a matching note in the captured commit.
-Ignored local files do not block it. It checks remote ancestry, runs
-[`release-check.sh`](../scripts/release-check.sh) once against that commit,
-rechecks candidate and version state, pushes only that commit to `main`, then
-creates and pushes one annotated tag. The tag message is the checked-in note.
-It never commits or repairs the checkout.
-
-A successful intended tag push completes publication. The command reports the
-version and commit immediately. No CI, module retrieval, remote read-back, or
-hosted release object is required. The checker may be run separately for
-diagnosis, but it is not a prerequisite: a fresh publication invocation
-validates its candidate once.
-
-## 5. Recover from a failed or uncertain push
-
-The command performs one bounded authoritative-state check after a failed push.
-If the main push failed but authoritative main equals the candidate, it proceeds.
-If the tag push failed but its exact tag object reached the remote, it reports
-success. Otherwise it reports branch-only, conflicting, or unknown state and
-retains any local tag.
-
-Inspect exact refs before any retry. Set `release_version` to the attempted
-version and compare main with the candidate from the command's report:
+Replace the placeholder with the selected version. ReleaseKit checks the
+effective push URL against `expected_push_url`, verifies the candidate on
+`main`, and checks the release-note H1 for the product and version. It then
+runs the selected race-enabled standard suite once with test-time vet disabled,
+one full vet pass, and CGO-disabled package compilation for Linux amd64 and
+arm64. It publishes only the annotated tag for that candidate. Confirmed
+publication of the intended tag completes the release. The tag points to the
+captured commit and uses the complete checked-in note as its annotation message.
+The optional `check`
+command is available for diagnosis but is not a separate required pass:
 
 ```sh
-release_version=v0.5.0
-authoritative=https://gitea.maximumdirect.net/eric/ConfigKit.git
-if remote_main=$(git ls-remote --refs --exit-code "$authoritative" refs/heads/main); then
-  printf 'remote main: %s\n' "$remote_main"
-else
-  result=$?
-  if [ "$result" -eq 2 ]; then
-    printf 'authoritative main confirmed absent\n' >&2
-  else
-    printf 'remote main state unknown; inspect before retrying\n' >&2
-  fi
-fi
-ref="refs/tags/$release_version"
-if remote_tag=$(git ls-remote --refs --exit-code "$authoritative" "$ref"); then
-  printf 'remote tag: %s\n' "$remote_tag"
-else
-  result=$?
-  if [ "$result" -eq 2 ]; then
-    printf 'remote tag confirmed absent\n'
-  else
-    printf 'remote tag state unknown; do not reuse the version\n' >&2
-  fi
-fi
-if local_tag=$(git rev-parse --verify "refs/tags/$release_version" 2>/dev/null); then
-  printf 'local tag object: %s\n' "$local_tag"
-else
-  printf 'no local tag object\n'
-fi
+"$releasekit_bin" check vMAJOR.MINOR.PATCH
 ```
 
-A query failure is not proof that the tag is absent. Compare the remote tag
-object ID with the local annotated tag object ID; its peeled commit alone is
-insufficient to establish identical publication. A conflicting published tag
-must not be moved or replaced. If remote main advanced but the tag was not
-published, report that partial progress. Only after confirming the local tag
-was never published or used by another channel may a maintainer remove it:
+Status 0 confirms publication, status 1 reports a known publication failure or
+conflict, status 2 reports usage, profile, or version errors, and status 3 means
+the push outcome remains unknown. Preserve both the native exit status and its
+diagnostics.
+
+The command does not push or repair `main`, wait for CI, or publish binaries or
+a hosted release page. A successful intended-tag publication does not require
+remote read-back or consumer retrieval; those checks remain optional.
+
+## 5. Recover from a failed or uncertain tag push
+
+The executable reports the publication state, phase, version, candidate commit,
+tag-object ID, effective endpoint, and error on standard error. It performs one
+bounded authoritative reconciliation after a failed tag push. Exit status 3
+means the tag outcome remains unknown; do not blindly retry or move a tag.
+
+For an uncertain result, use the exact endpoint, candidate commit, version, and
+tag-object ID from the report. Copy the captured endpoint into the variable
+below using normal shell quoting. Check that Git still resolves it to itself,
+then query the authoritative refs once:
+
+```sh
+authoritative='COPY_CAPTURED_ENDPOINT'
+release_version='vMAJOR.MINOR.PATCH'
+test "$(git ls-remote --get-url -- "$authoritative")" = "$authoritative" &&
+  git ls-remote --refs -- "$authoritative" refs/heads/main "refs/tags/$release_version"
+```
+
+A failed URL-equality check or remote query leaves the outcome unresolved.
+Correct transport configuration before continuing; a rewritten destination
+cannot establish absence at the captured endpoint.
+
+Compare the `main` object ID with the reported candidate commit and the tag
+object ID with the reported tag-object ID. `--refs` returns the annotated tag
+object itself, not merely its peeled commit. If the tag object matches, the
+release is published. If the query fails, the result is conflicting, or the
+endpoint or candidate differs, stop and resolve the state before retrying. If
+the endpoint is reachable, the tag is confirmed absent, and `main` still equals
+the reported candidate, remove a retained local tag only after confirming it
+was not published or used by another channel; then rerun `publish`, which
+revalidates the candidate:
 
 ```sh
 git tag -d "$release_version"
+"$releasekit_bin" publish "$release_version"
 ```
 
-A subsequent invocation revalidates the candidate; a correctly published tag
-needs no rerun. Do not roll back main or automatically delete a tag. Transport
-or authentication failure alone does not require a new version. A defect in
-published source requires a corrected commit, a new semantically appropriate
-version and note, and the release procedure again.
+A validation failure after the candidate reached `main` leaves an untagged
+candidate there. Correct it with a new commit rather than rewriting or rolling
+back `main`. A conflicting published tag must never be moved or reused. A
+defect in published source requires a corrected commit, a new semantically
+appropriate version, and a matching release note.
 
 ## 6. Optional consumer verification
 
-Only when useful or requested, read back the remote tag or retrieve the exact
-version into a new module without a local `replace` directive. These checks
-are optional and non-gating. Do not poll CI or wait for module propagation.
-Private-host credentials must already be configured. For direct-origin
-verification, set the version to the release just published (replace the sample
-below), then use a temporary consumer module and check the exact module path
-and version reported by `go list -m`:
+When useful or requested, read back the remote tag or retrieve the exact
+version into a new module without a local `replace` directive. These checks are
+optional and non-gating. Do not poll CI or wait for module propagation. For
+direct-origin verification, use the exact published version in a temporary
+consumer module:
 
 ```sh
-release_version=v0.5.0
+release_version=vMAJOR.MINOR.PATCH
 consumer_dir=$(mktemp -d)
 (
   cd "$consumer_dir"

docs/releases/v0.10.0.md

@@ -0,0 +1,77 @@
+# ConfigKit v0.10.0
+
+This minor release adds explicit, composable discovery of a configuration file.
+Applications choose the search conventions and their order; ConfigKit selects one
+file during loading and returns the final configuration with its normal metadata.
+Optional searches that find no file share the expanded `NoSource()` lifecycle.
+
+## Changes
+
+- Add `SearchConfig`, immutable `Search` and `FilenameGroup` descriptions, and
+  search-wide requiredness through `SearchOptions`. Compose explicit `Paths`,
+  custom `SearchDirectory` descriptions, and built-in Unix system, user-config,
+  home, or working-directory presets with `Searches`. See the
+  [public declarations](../../discovery.go) and
+  [discovery contract](../config.md#configuration-discovery).
+- Expand searches in supplied order, trying a directory's filename groups before
+  moving to the next directory. Named and generic groups provide conventional
+  YAML filenames; custom groups provide literal names. Directory searches
+  distinguish omitted filename selection from an explicitly empty selection.
+  Constructors perform no filesystem access, and each load observes current
+  directory bases and file availability.
+- Select the first existing candidate. Access errors and failures while loading
+  the selected file terminate the load rather than trying another configuration.
+  Selected files retain physical source identity, placement, provenance, imports,
+  and relative-resource context. Discovery works with ordinary loading and
+  layered inputs, including underlays, without adding a precedence tier.
+- Permit enabled imports with an absent main input: there are no main import
+  declarations to traverse, while actual underlay sources retain normal imports.
+  Support ordinary selected overlays with an absent main and explicit filesystem
+  context, including a root-only host context. Layered overlays retain their
+  independent `OverlayFilesystem` requirement. See
+  [single-source loading](../config.md#single-source-loading) and
+  [layered loading](../config.md#layered-loading).
+- Preserve ordinary file-context behavior after discovery and include the
+  failing candidate's file identity in access diagnostics. Use Go's
+  platform-aware Windows component rules and native drive-root-relative path
+  resolution. Isolate user-config tests according to the host Go conventions.
+
+## Compatibility and upgrade
+
+Existing explicit source constructors and loading APIs remain available.
+`File(path)` remains an exclusive selection that fails for a missing file;
+discovery is activated only by explicitly supplying `SearchConfig`.
+
+Searches are optional by default. Set `SearchOptions{Required: true}` when the
+application must find a file. An exhausted optional search contributes no main
+document or source record; application version checks, defaults, resolution,
+and validation still run under their existing rules and may reject incomplete
+configuration. A selected overlay is still required to exist.
+
+The former `NoSource()` rejections of enabled imports and explicit filesystem
+context are relaxed. An active ordinary overlay with no main file now loads
+when explicit context is supplied. Applications that relied on those combinations
+being rejected should review their option handling. Selected host files derive
+their own base directory; do not supply an explicit `BaseDir` or native filesystem
+context for a search that selects a host file.
+
+Applications may replace their own fallback-loading loops with the new search
+descriptions. The maintained [consumer examples](../consumers/README.md#explicit-main-file-discovery)
+show simple preset selection and custom-first discovery alongside existing
+configuration tiers. Composition, collection operations, inheritance, CLI
+precedence, and the complete-on-return ownership model retain their existing
+semantics.
+
+Select this version and rebuild:
+
+```sh
+go get gitea.maximumdirect.net/eric/ConfigKit@v0.10.0
+```
+
+The consumer minimum remains Go 1.26.0. Release validation uses Go 1.26.10.
+The YAML dependency remains `go.yaml.in/yaml/v4 v4.0.0-rc.6`, and the maintainer
+tool remains ReleaseKit v0.4.1. Linux amd64 and arm64 remain release-validated
+targets; macOS remains best-effort. Publication follows the existing
+[release procedure](../release.md).
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.10.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.10.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.10.0/docs/oss/attribution.md)

docs/releases/v0.10.1.md

@@ -0,0 +1,42 @@
+# ConfigKit v0.10.1
+
+This patch release consolidates ConfigKit's internal loading machinery and
+reduces repeated import bookkeeping. Public APIs and configuration semantics
+remain unchanged from v0.10.0.
+
+## Changes
+
+- Consolidate ordinary and layered input acquisition, including import requests
+  and selected-overlay handling, while retaining their distinct context and
+  precedence rules. Remove unused private reverse conversions.
+- Separate import acquisition, scoped composition, and evidence projection;
+  clarify inheritance preparation, combination, and finalization; and simplify
+  reflective projection and schema-field assembly responsibilities.
+- Retain resolved filesystem facts from non-import validation for later passive
+  resource contexts and ordinary overlays. Remove duplicate absent-host context
+  derivation while preserving validation timing and handle cleanup.
+- Record empty optional-import selections directly against their declaring
+  source, avoiding repeated searches through acquired sources. Index import
+  dependencies for scope-local edge preparation instead of scanning the whole
+  inclusion graph for every item scope. Stable dependency-ready ordering,
+  provenance, and source destinations are preserved.
+- Expand regression coverage for import evidence ownership, scope boundaries,
+  and broad main/underlay item imports. Improve
+  [consumer loading guidance](../consumers/README.md) and internal documentation.
+
+## Compatibility and upgrade
+
+No consumer code or configuration migration is required. Composition, collection
+operations, inheritance, discovery, CLI precedence, source-version checks, and
+inspection retain their existing contracts.
+
+Select this version and rebuild:
+
+```sh
+go get gitea.maximumdirect.net/eric/ConfigKit@v0.10.1
+```
+
+The consumer minimum remains Go 1.26.0. Dependencies and platform support are
+unchanged. Publication follows the existing [release procedure](../release.md).
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.10.1/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.10.1/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.10.1/docs/oss/attribution.md)

docs/releases/v0.10.2.md

@@ -0,0 +1,70 @@
+# ConfigKit v0.10.2
+
+This patch release fixes edge-case loading failures, resource cleanup, and limit
+accounting, and simplifies ConfigKit's internal loading machinery. Public APIs
+and documented configuration semantics remain unchanged from v0.10.1.
+
+## Fixes
+
+- Terminate pointer-only recursive type traversal safely during decoding and
+  schema-path resolution, preserving nullable values and supported recursive
+  structures. Roll back failed dynamic descriptor construction so subsequent
+  operations do not reuse incomplete state or panic.
+- Accept finite, non-cyclic slice views that share backing storage while
+  retaining cycle detection and detached output.
+- Deactivate resolution, validation, and version-check contexts when callbacks
+  panic, and close library-owned import roots during panic unwinding. The
+  original application panic continues to propagate.
+- Report directory enumeration and close failures instead of treating them as
+  optional absence, including failures reported as not-exist after a successful
+  open. Genuinely missing optional directories retain their existing behavior.
+- Reuse completed symlink dependencies during host and native import discovery,
+  avoiding exponential repeated expansion of finite link graphs while retaining
+  cycle detection and containment checks.
+- Check existing projection budgets before bulk container allocation and account
+  for the complete retained inspection-history graph, including replay and
+  separately copied inheritance evidence.
+- Apply the documented append/upsert sequence requirements to non-sequence
+  contributions at retained collection-policy destinations.
+
+## Internal improvements
+
+- Index collection-policy retirement and prepare sparse incoming policy changes
+  instead of repeatedly scanning or copying unrelated retained policy.
+- Reuse successful dynamic-type eligibility checks within an operation and index
+  tier-history lookups, compound identity restoration, and secret-failure source
+  attribution to avoid repeated scans.
+- Remove unused snapshots, evidence, helpers, fields, and test-only production
+  paths. Clarify acquisition handoffs and separate collection preparation,
+  application, and finalization while sharing sequence-history construction.
+- Separate patterned glob entry preparation from traversal. Consolidate engine
+  import defaults, snapshot them during load preflight, and construct placement
+  directly across the facade while keeping public types root-owned.
+- Add regression coverage for the corrected failures, public historical paths,
+  defaults snapshots, and resource ownership. Strengthen fuzz detachment checks
+  and replace redundant or implementation-coupled assertions.
+- Update release tooling from ReleaseKit v0.4.1 to v0.4.2.
+
+These changes remove redundant work and improve scaling in the identified paths;
+no end-to-end performance improvement has been quantified.
+
+## Compatibility and upgrade
+
+No consumer API or configuration-format migration is required. The
+[configuration contract](../config.md) remains authoritative.
+
+Loads that previously succeeded despite directory acquisition failures, invalid
+append/upsert contributions, or incomplete retained-history accounting may now
+return the documented structured errors. These corrections enforce existing
+requirements and limits rather than introducing new ones.
+
+Select this version and rebuild:
+
+```sh
+go get gitea.maximumdirect.net/eric/ConfigKit@v0.10.2
+```
+
+The consumer minimum remains Go 1.26.0. The YAML dependency and platform support
+are unchanged. Publication follows the existing [release procedure](../release.md).
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.10.2/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.10.2/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.10.2/docs/oss/attribution.md)

docs/releases/v0.6.0.md

@@ -0,0 +1,63 @@
+# ConfigKit v0.6.0
+
+This minor release reduces avoidable configuration-loading and inspection work,
+adopts ReleaseKit for source publication, and raises the consumer Go minimum to
+1.26.0. The public configuration model remains unchanged from v0.5.0.
+
+## Changes
+
+- Reduce temporary path allocations by sizing pointer-formatting buffers,
+  combining final metadata traversal, reusing parsed sensitivity paths, and
+  avoiding path construction when detecting inheritance. Effective and
+  historical metadata indexes retain independently owned paths.
+- Replace repeated resolution-coverage, metadata-update, CLI-overlap, and
+  sensitivity-propagation scans with private indexes and incremental processing.
+  Preserve diagnostic ordering, history, contextual path shapes, and redaction.
+- Make presence-only metadata queries avoid materializing provenance. Prepare
+  resolution replay descriptors and immutable mapping and inheritance-history
+  lookups for inspection, and use ordered membership indexes for provenance
+  collections. Returned records remain detached; explanations still replay
+  complete trees and render on demand.
+- Add a maintained public-API startup construction benchmark. The initial path
+  formatting and metadata traversal experiments reproduced allocation reductions
+  across local and downstream workloads. Later algorithmic improvements were
+  validated behaviorally without a new benchmark campaign. Busy-host timings
+  were inconclusive; no fixed speedup or allocation threshold is promised.
+- Replace ConfigKit-owned release scripts with a small profile and the
+  module-selected ReleaseKit v0.3.0 tool. The
+  [release procedure](../release.md) documents candidate preparation, tag-only
+  publication, and failed-push recovery. Existing release checks and the
+  successful-tag-push completion point remain unchanged; CI is not a gate.
+- Adopt release policy 1.6, distinguishing the consumer Go minimum from the
+  validation toolchain. Retire completed planning and measurement records into
+  Git history while preserving [performance candidates](../roadmap/future.md#5-performance-experiments).
+
+## Compatibility and upgrade
+
+The consumer minimum increases from Go 1.25.0 to **Go 1.26.0**. Install Go 1.26.0
+or newer before upgrading, and update application build and CI environments that
+still use Go 1.25. The `go.mod` directive records the consumer minimum; using a
+newer validation patch does not itself raise that minimum. See the
+[consumer minimum policy](../policy/release.md#consumer-minimum-and-release-toolchain).
+
+Select this version and rebuild:
+
+```sh
+go get gitea.maximumdirect.net/eric/ConfigKit@v0.6.0
+```
+
+No application-code or configuration migration is required from v0.5.0. Public
+Go signatures, strict within-tier composition, recursive tier precedence,
+sibling inheritance, presence/null distinctions, provenance, secret handling,
+and synchronous completion retain their existing contracts. See the
+[configuration contract](../config.md) and
+[consumer guide](../consumers/README.md).
+
+The YAML dependency remains `go.yaml.in/yaml/v4 v4.0.0-rc.6`. ReleaseKit is a
+declared Go tool dependency for maintainer commands, not a library runtime
+import; its requirements participate in ConfigKit's module graph. Linux amd64
+and arm64 remain release-validated targets, and macOS remains best-effort.
+Maintainers should use the current release procedure instead of the retired
+`scripts/release.sh` and `scripts/release-check.sh` commands.
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.6.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.6.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.6.0/docs/oss/attribution.md)

docs/releases/v0.7.0.md

@@ -0,0 +1,58 @@
+# ConfigKit v0.7.0
+
+This minor release adds structured comparison of completed configurations.
+Applications can inspect changes between two configuration files or the
+effective impact of a selected overlay, with ConfigKit handling comparison and
+safe value disclosure while applications own presentation.
+
+## Changes
+
+- Add `Diff[T](left, right *Result[T]) ([]Difference, error)` through the public
+  facade. It compares frozen effective inspection values from successful loads,
+  including defaults and recorded application resolution, rather than source
+  text or later mutations of `Result.Config`.
+- Return detached changed-region records containing a semantic path, safe left
+  and right `ValueSnapshot` values, and `HiddenChanged`. Mappings compare
+  recursively, sequences report atomic changes, and records follow canonical
+  pointer order. Presence distinguishes additions and removals from explicit
+  null; equal effective values produce no records regardless of provenance or
+  loading history.
+- Protect both snapshots with sensitivity from both loads, including historical
+  and dependency-derived classification. Compatible regions retain public
+  siblings; incompatible structural replacements use complete-region masking
+  where required. `HiddenChanged` identifies actual changes withheld by masking,
+  rather than merely indicating that a snapshot contains redacted content.
+- Preserve visible public scalar changes when only historical sensitive
+  descendants remain, including additions and removals. Historical explanation
+  redaction and masking of incompatible present shapes remain intact.
+- Add maintained file-comparison and base-versus-overlay examples with
+  consumer-owned formatting, plus an executable public API example. See the
+  [consumer guide](../consumers/README.md#effective-configuration-diffs) and
+  [configuration contract](../config.md#effective-configuration-diffs).
+
+## Compatibility and upgrade
+
+No application-code or configuration migration is required from v0.6.0.
+Adopting the new API is optional. Loading, composition, inheritance, provenance,
+existing inspection, and secret-handling contracts remain unchanged.
+
+Load both configurations successfully before calling `Diff`; nil or
+uninitialized results return an error with no records. Comparison does not
+reopen sources, rerun callbacks, or reread environment variables. Typed secret
+references compare by configured reference name, not resolved credential bytes,
+so credential rotation behind an unchanged name is not reported. Returned
+snapshots describe changed regions, not cross-protected full-document context.
+
+Select this version and rebuild:
+
+```sh
+go get gitea.maximumdirect.net/eric/ConfigKit@v0.7.0
+```
+
+The consumer minimum remains Go 1.26.0. The YAML dependency remains
+`go.yaml.in/yaml/v4 v4.0.0-rc.6`, and the maintainer tool dependency remains
+ReleaseKit v0.3.0. Linux amd64 and arm64 remain release-validated targets;
+macOS remains best-effort. Publication follows the existing
+[release procedure](../release.md).
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.7.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.7.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.7.0/docs/oss/attribution.md)

docs/releases/v0.7.1.md

@@ -0,0 +1,22 @@
+# ConfigKit v0.7.1
+
+This patch updates the maintainer release tool to ReleaseKit v0.4.1. ConfigKit's
+runtime behavior and public API are unchanged.
+
+## Compatibility and upgrade
+
+No application-code or configuration migration is required from v0.7.0. The
+consumer minimum remains Go 1.26.0, and the YAML dependency is unchanged.
+
+Select this version and rebuild:
+
+```sh
+go get gitea.maximumdirect.net/eric/ConfigKit@v0.7.1
+```
+
+ReleaseKit v0.4.1 is a maintainer tool dependency and does not change consumer
+configuration or runtime behavior. Linux amd64 and arm64 remain
+release-validated targets; macOS remains best-effort. Publication follows the
+existing [release procedure](../release.md).
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.7.1/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.7.1/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.7.1/docs/oss/attribution.md)

docs/releases/v0.8.0.md

@@ -0,0 +1,70 @@
+# ConfigKit v0.8.0
+
+This minor release adds explicit input placement, relative import destinations,
+shared source-version checking, and opt-in composition of identified sequence
+records. Applications can assemble independently authored configuration fragments
+into one final configuration while retaining source locations, provenance,
+history, and sensitivity information.
+
+## Changes
+
+- Add `Destination` to ordinary load options and layered inputs. A source mapping
+  can contribute beneath literal application keys without being rewritten by the
+  consumer. Imports inherit that destination or append a relative YAML
+  `destination`. Physical import paths and resource contexts remain relative to
+  their declaring sources. See [input destinations](../config.md#input-destinations),
+  [import syntax](../config.md#explicit-imports), and the
+  [mounted collection example](../consumers/README.md#input-destinations-and-mounted-collections).
+- Add `SourceVersionModeIfPresent` and `SourceVersionModeRequired` to apply one
+  application-owned version checker across main, underlay, imported, and selected
+  overlay documents. Enabled modes check source-local declarations and consume
+  accepted non-main headers before composition. Detached, sensitivity-aware
+  evidence is available through `SourceInfo.VersionCheck`. Acceptance remains
+  application policy; ConfigKit does not require declarations to equal the main
+  version. See [shared source-version checks](../config.md#shared-source-version-checks).
+- Add explicit `PlacementModeSequenceItem` placement for ordinary and layered
+  inputs, and `placement: sequence-item` for path and glob imports. A source
+  mapping becomes one item at its destination; ordinary helper imports compose
+  within that item and can supply its identity. See
+  [imports](../config.md#explicit-imports).
+- Add `Definition.SequenceCompositions` registrations for strict composition at
+  selected literal paths using one string identity field. Distinct records append
+  in contribution order, while duplicate identities conflict. Higher-priority
+  sequences still replace lower-priority sequences wholesale. Registration does
+  not add identity-based overlays or final identity validation after defaults and
+  application resolution. See the [composition contract](../config.md#strict-document-composition)
+  and [maintained prompt catalog](../consumers/README.md#identified-prompt-sequences).
+- Extend source inspection with destination and placement evidence, including
+  whether a nested destination is unresolved in an intermediate or failed-load
+  snapshot. Preserve source-specific history, consumed-header sensitivity, and
+  explanations of descendants removed by sequence replacement as records are
+  assembled. Source locations continue to identify authored coordinates.
+
+## Compatibility and upgrade
+
+The new behavior is opt-in: zero-valued destinations retain root placement,
+mapping placement remains the default, unregistered sequences retain strict
+conflict behavior, and the default source-version mode preserves the existing
+main-only version policy. Recursive mapping precedence, whole-sequence
+replacement between tiers, and application-owned final validation remain intact.
+
+Public option, definition, input, and source-information structs gain fields.
+Consumers using unkeyed literals for affected structs must switch to keyed
+literals or update them. `SourceInfo` now contains a `Path` and is no longer
+comparable with `==` or usable directly as a map key; compare the relevant fields
+or select an application-appropriate key instead. Existing keyed literals need
+no changes to retain the previous loading behavior.
+
+Select this version and rebuild:
+
+```sh
+go get gitea.maximumdirect.net/eric/ConfigKit@v0.8.0
+```
+
+The consumer minimum remains Go 1.26.0. The YAML dependency remains
+`go.yaml.in/yaml/v4 v4.0.0-rc.6`, and the maintainer tool dependency remains
+ReleaseKit v0.4.1. Linux amd64 and arm64 remain release-validated targets;
+macOS remains best-effort. Publication follows the existing
+[release procedure](../release.md).
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.8.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.8.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.8.0/docs/oss/attribution.md)

docs/releases/v0.9.0.md

@@ -0,0 +1,111 @@
+# ConfigKit v0.9.0
+
+This minor release expands import selection, adds author-declared collection
+identity and operations, and supports inheritance between identified records.
+It also moves CLI assignments after inheritance and typed decoding. These changes
+include breaking API and configuration behavior changes from v0.8.0; review the
+migration guidance below before upgrading.
+
+## Changes
+
+- Extend imports with named-file and directory lists, directory filename patterns,
+  exclusions, recursion and depth controls, and explicit hidden-file and symlink
+  discovery controls. Add `ImportDefaults` to ordinary and layered options, with
+  authored declarations taking precedence over application defaults. Imports are
+  optional by default. `SourceInfo.ImportSelections` exposes completed optional
+  declarations that selected no sources. See the
+  [import contract](../config.md#explicit-imports) and
+  [consumer guidance](../consumers/README.md#import-defaults-and-selection-inspection).
+- Use shared dependency-ready ordering for import composition and sibling
+  inheritance. Dependencies complete before dependents, with deterministic
+  selection among currently eligible contributions. Source acquisition and
+  callback order remain separate from assembly order. See the
+  [ordering decision](../adr/0015-shared-dependency-ready-ordering.md).
+- Add root `configkit.sequences` declarations for collection identity and
+  `replace`, `append`, or `upsert` operations, including in selected overlays.
+  Strict peers combine distinct identities and reject duplicates; precedence
+  and inheritance apply the destination's collection operation. Upsert replaces
+  a whole matching record at its retained position. Identity supports compound
+  field sets and scalar strings, numbers, booleans, and null. See the
+  [collection contract](../config.md#strict-document-composition).
+- Add `Definition.SequenceIdentities`, exact `SequenceIdentityAt` registrations,
+  and dynamic `SequenceIdentityMatching` selectors. Application registrations
+  govern structural collection operations and check existing matching final
+  typed sequences, including values from defaults, CLI assignments, and
+  application resolution. YAML-only declarations govern structural operations
+  without adding that final-result check.
+- Support record inheritance within an identified sequence using an `extends`
+  mapping containing the parent's complete identity. The child retains its own
+  identity, relationships use surviving assembled records, and output sequence
+  order is preserved. See [sibling inheritance](../config.md#sibling-inheritance)
+  and the maintained [sequence catalog](../consumers/README.md#identified-prompt-sequences).
+- Apply explicit CLI overrides to the decoded configuration before application
+  resolution. Mappings retain omitted fields and their presence metadata;
+  sequences replace whole lists independently of collection operations. See the
+  [CLI contract](../config.md#explicit-cli-overrides).
+- Correct native import alias and contained-symlink handling, source-version
+  path projection, and provenance/history through collection movement,
+  replacement, inheritance, and CLI assignment. Reduce repeated scans,
+  accumulated policy copying, and sparse-projection storage during construction.
+
+## Compatibility and upgrade
+
+This pre-v1 minor release intentionally changes the v0.8.0 contracts:
+
+1. **Migrate import requiredness.** The YAML `optional` field is removed and is
+   rejected. Replace `optional: true` with `required: false`; omission is equivalent
+   when the application does not default imports to required. Replace
+   `optional: false` with `required: true`. Imports that previously relied on
+   implicit requiredness now need `required: true`, or an application
+   `ImportDefaults.Required` default set to true. Review discovery filters:
+   `follow` and `hidden` default to true, and directory searches select all
+   filenames unless a pattern is supplied.
+2. **Replace sequence registrations.** `SequenceComposition` and
+   `Definition.SequenceCompositions` are removed without compatibility aliases.
+   For a previous registration at `prompts` using `id`, use:
+
+   ```go
+   rules.SequenceIdentities = []configkit.SequenceIdentity{
+       configkit.SequenceIdentityAt(configkit.Keys("prompts"), "id"),
+   }
+   ```
+
+   These registrations now also enforce identity in final typed sequences. Fix
+   missing or duplicate identities introduced by defaults, CLI assignments, or
+   application resolution. Registration does not require a collection to exist
+   or determine its allowed shape; schema and application validation retain
+   those responsibilities.
+3. **Review collection precedence and ordering.** An identified collection now
+   defaults to upsert across layers and inheritance when no explicit mode
+   applies, rather than unconditional whole-sequence replacement. Declare
+   `mode: replace` at the relevant destination where replacement is intended.
+   With append or upsert, `[]` contributes no records and does not clear the
+   collection; null is invalid. Use `mode: replace` with `[]` to clear it.
+   An explicit mode persists at its destination until changed; a parent's mode
+   does not transfer to a different inheriting destination. Unidentified
+   sequences retain strict peer conflicts and whole-value precedence
+   replacement. Dependency-ready assembly can change sequence positions and
+   diagnostic or history order compared with the former traversal order.
+4. **Review CLI recovery and inheritance assumptions.** CLI overrides can no
+   longer repair values that fail earlier decoding, bypass source or inheritance
+   errors, or add or cancel inheritance relationships. Supply structural
+   directives in source configuration. Application resolution still receives
+   explicit CLI values and may record subsequent normalization or derivation.
+5. **Update affected Go literals and diagnostic handling.** Use keyed literals
+   for expanded option and source-information structs. Collection validation adds
+   `CodeSequenceInvalid` (`sequence.invalid`) and `CodeSequenceConflict`
+   (`sequence.conflict`); update diagnostic handling where applicable.
+
+Select this version and rebuild:
+
+```sh
+go get gitea.maximumdirect.net/eric/ConfigKit@v0.9.0
+```
+
+The consumer minimum remains Go 1.26.0. The YAML dependency remains
+`go.yaml.in/yaml/v4 v4.0.0-rc.6`, and the maintainer tool dependency remains
+ReleaseKit v0.4.1. Linux amd64 and arm64 remain release-validated targets;
+macOS remains best-effort. Publication follows the existing
+[release procedure](../release.md).
+
+[COPYRIGHT](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.9.0/COPYRIGHT) | [LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.9.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/ConfigKit/src/tag/v0.9.0/docs/oss/attribution.md)

docs/roadmap/deferred.md

@@ -1,34 +0,0 @@
-# Deferred Scope
-
-This document owns intentionally excluded features, not a committed delivery plan.
-The [architecture policy](../policy/architecture.md) governs changes to these
-boundaries. Implemented behavior belongs in the [configuration contract](../config.md)
-and consumer adoption in the [consumer guide](../consumers/README.md).
-
-## Explicit non-goals and deferred scope
-
-General environment-variable configuration overlays are not supported. Typed secret
-references and explicit directory bootstrap are the environment convention. Literal
-credential configuration and implicit secret-directory discovery are unsupported.
-
-The following remain deliberately excluded:
-
-- automatic `conf.d` discovery;
-- remote imports, arbitrary providers, and additional configuration formats;
-- custom merge strategies, sequence merging, reset operators, multiple-parent
-  inheritance, and cross-scope references;
-- named overlay inheritance;
-- parser-specific or custom structural-decoding extension APIs;
-- generic dynamic-subtree conversion and module-registry APIs;
-- automatic schema migrations;
-- hot reload, runtime resource replacement, and remote resource validation;
-- a bundled CLI framework or application commands; and
-- configuration diffs and fingerprints.
-
-Applications may build their own validate, effective-value, source, bootstrap, and
-explanation commands around the library. Broadening these boundaries requires
-concrete consumer evidence, compatibility review, and an explicit design change.
-
-Built-in sibling inheritance using `extends` is implemented as described in the
-[configuration contract](../config.md#sibling-inheritance). It is distinct from
-inheritance between selected overlay files, which remains excluded above.

docs/roadmap/future.md

@@ -0,0 +1,93 @@
+# Potential Future Work
+
+This document lists potential enhancements and maintenance work for consideration
+in future ConfigKit releases. Entries are not accepted contracts, implementation
+plans, or release commitments. Adopting a candidate requires an explicit design
+decision and a separate plan. Significant architectural exclusions and their rationale belong
+in ADRs; this document does not maintain a list of rejected proposals.
+
+## 1. Separate glob entry preparation from traversal
+
+Implemented in the [internal-boundary roadmap](internal-boundaries.md#glob-entry-preparation) (C03).
+
+## 2. Simplify redundant facade-to-engine transport
+
+Implemented in the [internal-boundary roadmap](internal-boundaries.md#facade-to-engine-transport) (F07).
+
+## 3. Recursive body merging for identified records
+
+Status: deferred. The implemented
+[collection policy](../config.md#strict-document-composition) uses `upsert` to
+replace a complete matched record. Recursive merging of fields inside matching
+records remains a separate candidate and would revisit the whole-record update
+boundary selected by [ADR 0017](../adr/0017-collection-sequence-policy.md).
+
+Consider this only in response to a concrete consumer need. A proposal would need
+to define conflicts for overlapping body fields, removal of omitted fields,
+presence and ordering, and how provenance, history, sensitivity, and resource
+context follow partially retained records. Any change requires a new decision and
+implementation plan; this entry does not alter current upsert behavior.
+
+## 4. Required version declarations for selected sources
+
+Status: deferred. PromptKit requested a way to require version declarations on
+a selected subset of source files while permitting other fragments to omit them.
+
+### Motivation and proposed behavior
+
+PromptKit could require each collection's root file (for example,
+`collection.yml` or `meta.yml`, with naming still undecided) to declare an
+accepted specification version. Other files would use the existing
+[check-when-present behavior](../config.md#shared-source-version-checks).
+This would enforce a rule that otherwise remains a documentation convention.
+
+All declarations would use the same application-owned `Definition.Version`
+path and acceptance policy. The proposal concerns where declarations are
+required; it does not introduce independently versioned collections or subtrees.
+A root declaration would not prove that headerless descendants declare a
+compatible version or replace final configuration validation.
+
+### Questions to resolve before implementation
+
+- **Selection API:** should callers mark explicit inputs, select sources by
+  path or predicate, or annotate imports? An input-level requirement is simpler,
+  but cannot by itself select a root reached only through an import.
+- **Scope:** does a requirement apply only to the selected document or also to
+  its imports? Applying it only to that document best fits PromptKit's request;
+  imported fragments could still declare versions checked by the shared policy.
+- **Mode interaction:** how should selection interact with disabled source
+  checks? Require an enabled mode, or activate checking for selected sources?
+  Selection would strengthen check-when-present behavior and add no requirement
+  beyond the load-wide required mode.
+- **Source handling and evidence:** how should selection work across file,
+  reader, byte, and mapping inputs, glob matches, and repeated imports? Define
+  behavior for absent optional inputs and preserve source-local diagnostics,
+  inspection evidence, and existing sensitivity handling.
+
+## 5. Performance experiments
+
+Status: deferred pending evidence. The completed performance audit identified
+four internal optimization candidates for separate investigation:
+
+- Branch-local explanation replay instead of the current whole-tree replay.
+- Redesigning the private metadata-copy handoff while preserving detached public
+  records.
+- Caching schema descriptors across loads.
+- Indexing complex contributor-trace membership while preserving its equality
+  rules and compact linked traces.
+
+These are experiments, not committed implementation work or promised speedups.
+Reconsider them with evidence that the benefit justifies the ownership,
+retained-memory, or equivalence risks. The completed audit and implementation
+evidence are preserved in Git history at `8e91a2b` under `docs/roadmap/`.
+
+### Questions to resolve before implementation
+
+- Which representative workloads show enough remaining cost to justify each
+  optimization, and what measurements would establish a useful improvement?
+- Can narrower replay and membership indexes preserve explanation results,
+  ordering, provenance, and sensitivity behavior?
+- Can metadata handoff changes preserve detached public records and ownership
+  across loads without introducing aliasing?
+- What cache lifetime and concurrency design would make schema reuse worthwhile
+  without excessive retained memory?

docs/roadmap/implementation.md

@@ -0,0 +1,296 @@
+# Import Discovery And Facade Simplification Implementation Plan
+
+Status: complete.
+Baseline: `bb5540b62e382106d59678e2edb7655acfd5c960`.
+Audience: a Luna coding agent implementing the stages sequentially.
+
+## Scope And Execution
+
+Implement **C03** and **F07** from the
+[internal-boundary roadmap](internal-boundaries.md). That document owns the target
+state and preservation requirements; this plan owns concrete private choices,
+steps, validation, and progress. The preceding 29-finding remediation sprint is
+complete. Its implementation record remains at
+`bb5540b:docs/roadmap/implementation.md`; this plan does not reopen it.
+
+Read `AGENTS.md`, [agent policy](../policy/agents.md), the [development
+guide](../development.md), and applicable architecture, testing, and documentation
+policies. Use graph-first discovery and verify actual callers before deleting
+helpers. Preserve unrelated work.
+
+This is one end-to-end sprint. Intermediate stages may fail to compile, especially
+the coordinated engine/facade cutover in Stages 2–3. Do not create adapters,
+aliases, duplicate paths, or extra abstractions solely to make intermediate
+stages green. Run focused checks once their affected surface is buildable; record
+pending checks and finish them before final acceptance. Commit focused nonempty
+units under repository policy and record stage status, actual validation, and
+known commit identifiers here. Completion requires the integrated result, not
+just checked-off stages.
+
+### Boundaries
+
+- Preserve public root-owned declarations and signatures, YAML syntax, loading
+  semantics, ordinary/layered differences, callback order, diagnostics, source
+  identity, inspection, confidentiality, and ownership.
+- Preserve glob literal-versus-discovered rules and directory/glob differences.
+  Keep completed-link reuse, directory failure phases, containment, and existing
+  accounting. This is not a discovery algorithm or permission-policy change.
+- Keep `resolvedImportPolicy`, registration compilation, normalized request
+  responsibilities, and conversions with real ownership or domain meaning.
+  Simplification is limited to defaults transport and placement construction.
+- Add no dependencies, caller controls, caches, limits, background work, release
+  gates, or performance promises. Do not publish a release, update consumers, or
+  revise accepted ADR decisions.
+
+### Selected Ownership Model
+
+Use the existing engine `ImportDefaults` name and its exported `Required`,
+`Follow`, `Hidden`, `Recursive`, and `Pattern` fields as the sole engine
+input-defaults type. Root `ImportDefaults` stays separately defined. Root
+conversion maps the fields without cloning them; the public loading wrappers
+immediately enter engine preflight. `loadSettings` takes the per-load snapshot
+before acquisition or callbacks, as it already does. The normalized request and
+acquisition policy then borrow that load-owned read-only value.
+
+Declaration resolution reads the owned defaults and retains its existing copies
+of selected pattern/exclusion slices and destinations in `resolvedImportPolicy`.
+This separates presence-aware load input from independently owned effective
+declaration policy. No public copy-count or thread-safety guarantee follows.
+
+For placement, expose the engine's existing enum as `PlacementMode uint8` with
+`PlacementModeMapping` and `PlacementModeSequenceItem`, and expose placement
+fields on engine input/options records for direct construction. These remain
+internal-package implementation surfaces. Root types stay root-owned; direct
+casts preserve unknown values for existing preflight validation.
+
+### Validation Conventions
+
+Run from the repository root with an installed compatible Go toolchain and
+dependencies already available. Use `GOTOOLCHAIN=local`, `GOPROXY=off`, and
+`GOTELEMETRY=off`. Keep disposable artifacts outside the repository and do not
+change dependencies for tooling. Focused expressions identify behavior owners;
+include a newly named regression explicitly if needed. Reuse unchanged successful
+checks. No timing, allocation, exact syscall, or helper-shape assertions are
+acceptance criteria.
+
+## Stage 1 — Extract Patterned Glob Entry Preparation
+
+Finding: **C03**. Owners: `internal/engine/glob.go`, existing glob selection,
+link, and directory failure tests. Status: **complete**. Validation passed:
+`env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test ./internal/engine -run 'Glob|Directory|Link|Filesystem|Resource'` and `git diff --check`.
+
+1. Add `prepareGlobEntry` beside the selector. Take the filesystem access,
+   base path, parent `selectedImportCandidate`, `fs.DirEntry`, pattern segment,
+   `last` flag, and `resolvedImportPolicy`. Return a candidate, a boolean meaning
+   selected versus ordinary skip, and a nullable `*Diagnostic`. Use the parent's
+   existing `site`, which is propagated from the selector's root candidate.
+   No candidate-set, accounting, or traversal-state mutation belongs here.
+2. Move only the patterned branch's per-entry body into this helper. Preserve
+   its order: `path.Match`; match/hidden filtering; lexical/access path creation
+   and containment; `importEntryTarget` and existing diagnostic translation;
+   follow filtering; intermediate-directory eligibility; host/native physical
+   path selection and canonicalization fallback; candidate label/site creation.
+   Return existing safe codes, paths, and sites. Do not turn inspection errors
+   into skips or skip validation merely because exclusion will later match.
+3. Replace that inline body with the helper call and candidate insertion.
+   Leave pattern validation, `explicitGlobCandidate`, directory enumeration,
+   per-segment union/sorting, final-file validation, exclusions, requiredness,
+   and source-admission timing in the selector.
+4. Keep directory entry preparation separate. Reuse existing filesystem/link
+   primitives; do not introduce a generic entry strategy or unify differing
+   missing-target, literal-link, filtering, or recursion rules. Remove imports
+   only when genuinely unused after extraction.
+
+Acceptance: the selector now expresses traversal without embedding preparation
+details. Reuse `glob_selection_test.go`, `glob_test.go`, filesystem-link tests,
+and the preceding sprint's directory-phase regressions. They must still cover
+literal hidden/link components versus wildcard filtering, aliases and native
+spelling, escaping/cyclic links, matched-directory errors before exclusion,
+selection order, and bounded accounting. Add a behavioral case only for a gap
+identified during extraction, not a unit test asserting the new helper exists.
+
+Focused check:
+`go test ./internal/engine -run 'Glob|Directory|Link|Filesystem|Resource'`.
+
+## Stage 2 — Consolidate The Engine Defaults Representation
+
+Finding: **F07**, engine representation. Owners: `internal/engine/import_spec.go`,
+`load.go`, `layered.go`, `load_request.go`, `acquisition.go`, `options.go`, and
+their callers/fixtures. Status: **complete**. Validation passed:
+`env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test ./internal/engine -run 'Import|Glob|Directory|SourceVersion'` and `git diff --check`. Root facade callers remain pending Stage 3.
+
+1. Retain engine `ImportDefaults` with its existing five exported fields.
+   Remove private `importDefaults`, without a compatibility alias. Update
+   engine option/request/acquisition fields, helper signatures, readers, and
+   fixture literals to use the canonical type and exported default-field names.
+   Private containing fields such as `loadRequest.importDefaults` can keep their
+   names. Do not rename the distinct fields of `importDeclarationSpec` or
+   `resolvedImportPolicy` just because they have similar meanings.
+2. Expose `ImportDefaults ImportDefaults` on engine `Options[T]` and
+   `LayeredOptions[T]`. Remove `OptionsWithImportDefaults`,
+   `LayeredOptionsWithImportDefaults`, and `privateImportDefaults`. Stage 3 changes
+   root callers; a temporary compile failure is preferable to an interim bridge.
+3. Adapt `snapshotImportDefaults` to the canonical type, retaining nil boolean
+   pointers and nil versus non-nil empty pattern. Keep `cloneBool` and
+   `cloneStrings` where they still have real consumers. Keep current snapshot
+   call sites until Stage 3 establishes their final ownership placement.
+4. Keep effective declaration policy resolution and validation semantics intact:
+   explicit false beats defaults, each control resolves independently, and
+   selector-specific applicability and depth checks stay unchanged. Do not
+   combine defaults with resolved policy or registration compilation.
+
+Acceptance: one engine input-defaults type reaches every production path and
+fixture; no equivalent private record or alias remains. Inspect ordinary,
+layered, transitive-import, directory/glob, and test-helper call sites. The root
+cutover and removal of redundant snapshots are explicitly pending until Stage 3.
+
+Run the import/default focused tests after Stage 3 if the root cutover prevents
+building the package here; do not report that deferred check as passed.
+
+## Stage 3 — Simplify Facade Defaults Construction And Snapshot Ownership
+
+Finding: **F07**, defaults handoff. Owners: root `conversion.go`, `lifecycle.go`,
+`layered.go`; engine defaults/preflight/coordinator owners; `facade_import_defaults_test.go`
+and `internal/engine/import_spec_test.go`. Status: **complete**. Validation passed:
+`env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test ./internal/engine -run 'Import|Glob|Directory|SourceVersion'`, `env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test . -run 'ImportDefaults|FacadeImport|SourceVersion'`, `env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test ./...`, `env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test -race ./...`, and `git diff --check`.
+
+1. Set engine `ImportDefaults` fields directly in root `toEngineOptions` and
+   `toEngineLayeredOptions`. Retain `toEngineImportDefaults` as the explicit
+   root-to-engine type mapping, but map its booleans/pattern without copies.
+   Remove root `cloneImportDefaultBool` and `cloneImportDefaultStrings` after
+   confirming no other caller. Keep public root types and declaration contracts.
+2. Keep `snapshotImportDefaults` in `loadSettings` as the load-entry ownership
+   operation for both APIs. Remove repeated snapshots in engine `LoadLayered`
+   request construction, `executeLoadRequest` acquisition-policy construction,
+   and `resolveImportPolicy`. Carry the owned defaults read-only thereafter.
+   Review all callers of those helpers, including test setup; direct declaration
+   resolution can read its input because it returns its own concrete policy.
+3. Keep per-policy copies of the selected pattern, exclusions, and destination.
+   Do not borrow caller-owned storage beyond preflight, mutate the shared owned
+   defaults, or remove copying from unrelated option/record paths. Each new load
+   snapshots its own current inputs; add no reuse cache or ownership flag.
+4. Replace the private conversion-snapshot fragment in
+   `TestImportDefaultsPreserveEmptyPatternAndRequiredness`, which converts options,
+   mutates inputs, and calls engine `Load`. It should no longer demand a snapshot
+   from a private conversion helper. Keep that test's public nil/empty,
+   requiredness, and inspection-evidence checks.
+5. Protect the public snapshot contract with a compact ordinary/layered table:
+   use a main document with a version and imports, enable
+   `SourceVersionModeIfPresent`, and use a configuration type with version and
+   imported-value fields. In the main version-check callback, mutate the caller's
+   recursive default from true to false and its existing pattern element from
+   `*.cfg` to `*.skip`. This callback runs after preflight but before imported
+   discovery. Verify the original recursive/pattern selection still supplies a
+   value from a nested imported file. Use existing fixture helpers and no
+   concurrent mutation. A subsequent load with those changed defaults must omit
+   that imported contribution rather than reuse the prior snapshot. Keep fixtures
+   reusable or create fresh readers for each load.
+6. Reuse or extend the small snapshot/policy tests for all four optional booleans,
+   explicit false, pattern mutation, nil, and non-nil empty pattern. Preserve
+   existing invalid-pattern-before-read/hook tests and declaration override,
+   transitive, underlay, and selected-overlay coverage. Do not assert exact copy
+   counts, pointer identities, or the private conversion boundary as public policy.
+
+Acceptance: the engine/facade cutover builds; public defaults remain detached
+before acquisition/callbacks, reused loads remain independent, and only genuine
+ownership copies remain. Confirm the absence of full defaults snapshots at the
+three removed downstream sites by source review, not production instrumentation.
+
+Focused checks:
+`go test ./internal/engine -run 'Import|Glob|Directory|SourceVersion'` and
+`go test . -run 'ImportDefaults|FacadeImport|SourceVersion'`.
+
+## Stage 4 — Construct Placement Directly Across The Facade
+
+Finding: **F07**, placement handoff. Owners: engine `destination.go`, `load.go`,
+`layered.go`, normalized input and placement consumers; root `lifecycle.go`,
+`layered.go`, and placement fixtures. Status: **complete**. Validation passed:
+`env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test ./internal/engine -run 'Destination|Placement|Item|SourceVersion'`, `env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test . -run 'Placement|Destination|SourceInfo|SourceVersion|PublicFacadeBoundary|DependencyImportBoundary'`, `env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test ./...`, and `git diff --check`.
+
+1. Rename the existing engine enum and its constants to `PlacementMode`,
+   `PlacementModeMapping`, and `PlacementModeSequenceItem`, preserving underlying
+   `uint8` values and zero semantics. Use the same type throughout engine
+   placement parsing, normalization, acquisition, source evidence, and version
+   projection. Do not add a second raw-mode record or type alias.
+2. Expose engine `Options[T].Destination` and `.PlacementMode`, and
+   `Input.PlacementMode`. Normalized `loadInput` and other private owner fields
+   can retain their names. Populate the normalized input from those fields.
+   Remove `OptionsWithPlacement` and `InputWithPlacement` and migrate all callers,
+   including the direct engine construction in root `records_test.go`.
+3. In root `toEngineOptions` and `toEngineInput`, set destination using the existing
+   `toEnginePath` conversion and placement using
+   `engine.PlacementMode(options.PlacementMode)` or its input equivalent.
+   Preserve invalid enum values for `normalizeInput` to diagnose before reads.
+   Do not add validation/defaulting to the facade or a new setter/cast function.
+4. Preserve normalization of literal/parsed destination keys, root-item rejection,
+   wrapper accounting, mounted/item source coordinates, and version projection.
+   Root `Path` is immutable and `toEnginePath` carries its private engine value;
+   keep normalization and actual mutable ownership copies. Do not redesign path
+   conversion or add redundant clones solely to replace the deleted decorator.
+5. Leave source-result transport such as `SourceInfoPlacementModeForFacade`,
+   source-version conversion, sequence registration preparation, and opaque
+   records alone. They are not input setters targeted by this step. Adapt tests
+   mechanically to new engine construction without weakening public assertions.
+
+Acceptance: both public loading APIs retain zero/mapping and sequence-item
+placement, errors before input consumption, final values, source metadata, and
+history. Reuse the public placement tests plus destination/accounting,
+item-import, mounted-version, and invalid-option cases. Root signatures still
+contain no engine types, and engine exports remain below the internal boundary.
+
+Focused checks:
+`go test ./internal/engine -run 'Destination|Placement|Item|SourceVersion'` and
+`go test . -run 'Placement|Destination|SourceInfo|SourceVersion|PublicFacadeBoundary|DependencyImportBoundary'`.
+
+## Stage 5 — Integrate, Document, And Verify The Result
+
+Findings: final acceptance for **C03, F07**. Status: **complete**. Validation
+passed on Linux/amd64 (Go 1.26.10):
+`env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off ./scripts/check.sh`,
+`env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test -race ./...`, and
+`git diff --check`. Existing integration coverage was reviewed; no additional
+case was needed.
+
+1. Review both paths together: defaults resolving wildcard discovery in ordinary,
+   layered, and transitive imports; placement carrying the resulting sources
+   through composition and inspection; and failures stopping before later hooks.
+   Existing integration tests own these interactions. Add only a concrete missing
+   case rather than a second copy of the complete matrix or a new fixture framework.
+2. Inspect the resulting handoffs. Confirm the glob selector no longer embeds
+   per-entry preparation, the engine has one defaults input type, direct fields
+   replaced the four input decorators, and the load snapshot is not repeated at
+   each private representation. Keep domain-specific policy/registration and
+   root conversion boundaries, necessary ownership copies, and existing budgets.
+3. Update `docs/internal/imports.md` for the selector/preparation split and
+   `docs/internal/overview.md` for defaults conversion, preflight snapshot ownership,
+   and direct placement construction. Keep algorithms private and avoid public
+   performance or copy-count guarantees. No ADR/policy change is needed.
+4. Replace C03/F07 detail in the first two `docs/roadmap/future.md` entries with
+   short pointers to `internal-boundaries.md`. Retain their heading anchors for
+   existing audit links; leave other future candidates unchanged. Correct old
+   roadmap references that would otherwise mistake this replacement plan for
+   the previous sprint's completed record: use
+   `bb5540b:docs/roadmap/implementation.md` for that historical plan. Do not reopen
+   the prior audit or retire unrelated planning records in this sprint.
+5. Format changed Go files, finish deferred checks, run the final checks below,
+   and inspect the complete diff for API, dependency, semantics, or scope changes.
+   Record actual results and platform limitations here. Commit focused nonempty
+   units and hand off a clean workspace; do not create empty validation commits.
+
+Final validation, following the existing development guide:
+
+```sh
+env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off ./scripts/check.sh
+env GOTOOLCHAIN=local GOPROXY=off GOTELEMETRY=off go test -race ./...
+git diff --check
+```
+
+The everyday check includes consumers, documentation links, and public/dependency
+boundary tests. Confirm those existing tests ran rather than adding redundant
+final invocations. Report filesystem/platform skips accurately; express macOS
+validation, CI completion, release publication, extended fuzz campaigns, and
+benchmarks are not additional acceptance requirements.
+
+No product or architectural question remains open. The concrete choices above
+complete the plan without changing the roadmap's public contracts; private
+formatting and file organization can follow local idiom.

docs/roadmap/interface-remediation.md

@@ -0,0 +1,346 @@
+# Interface Contract Remediation Roadmap
+
+Status: target settled, including shared history ownership; implementation has
+not begun.
+Planning baseline: `1f64673` (v0.10.2 implementation plus documentation changes).
+
+This roadmap covers the four findings in the [interface audit](interfaces.md).
+The audit owns reproductions, evidence, and finding numbers. This document owns
+the combined target, required changes, and general dependency order. A subsequent
+implementation plan should own concrete stages, private implementation choices,
+validation commands, and progress.
+
+## Purpose And Scope
+
+Restore agreement between ConfigKit's presence, metadata, history, and diagnostic
+interfaces under their existing contracts. The first three findings concern
+semantic paths that share a JSON Pointer spelling; the fourth concerns a
+composition claimant order that is lost during diagnostic canonicalization.
+
+Build one shared internal history model from which metadata history and
+explanation are derived. It will own the historical evidence and semantic
+attribution used by both views, replacing their independent reconstruction of
+that attribution. This is the selected resolution of OD-1.
+
+The [Path declarations](../../path.go), [metadata contract](../config.md#layered-metadata),
+[resolution contract](../config.md#recorded-application-resolution), and
+[strict-composition contract](../config.md#strict-document-composition) remain
+authoritative. This work changes the implementation of those contracts without
+selecting a new public API, path syntax, precedence rule, or configuration
+restriction. The existing root-owned facade and engine responsibilities remain.
+
+## Current Baseline And Targeted End State
+
+An explicit mapping-key path and an explicit sequence-index path can both render
+as `/shape/0`. `ParsePath` deliberately leaves numeric tokens contextual, whereas
+`Keys` and `Index` preserve the caller's explicit choice. The engine already has
+shape-aware effective/historical path indexes and a frozen resolution-entry view,
+but some queries and metadata updates still use pointer spelling alone.
+
+History currently reconstructs definition and previous-origin records, while
+explanation replays source contributions, inheritance, decoding checkpoints, CLI
+assignments, and grouped resolution snapshots. Neither public representation
+contains everything needed by the other. The target is shared underlying evidence
+with distinct projections for each interface, preserving their existing behavior.
+
+| Finding | Current discrepancy | Target |
+| --- | --- | --- |
+| 1: resolution presence | `IsExplicit` and `Default` consult string-keyed presence after schema validation, which cannot determine actual shape beneath dynamic values. | Presence decisions use the matching semantic path in the fixed resolution-entry state. |
+| 2: effective explicitness | Resolution retains old field information when pointer spelling survives a container-shape change. | Input explicitness carries forward for the same semantic path; a newly introduced key or index does not inherit another path's explicitness. |
+| 3: history attribution | A historical-shape check admits the query, then `History` returns the entire textual-pointer bucket; explanation selects evidence separately. | Both interfaces derive attribution from one shared history model, selecting exact definitions, previous origins, and applicable ancestor operations by semantic path. |
+| 4: claimant order | Composition constructs related locations in assembly order, then canonicalization sorts them by location. | Structured and rendered composition conflicts preserve the selected claimant order. |
+
+For example, if input contains `shape: {'0': old}` and recorded application
+resolution changes `shape` to a sequence, the old key remains an input path in
+the callback's entry view. The new index is effective in final metadata but was
+not explicitly supplied by that input. Its history must not present the old key's
+definition as its own. This distinction concerns container shape, not identity
+tracking for records that move within a sequence.
+
+## Required Changes
+
+### 1. Use Entry-State Shape For Resolution Presence
+
+In [resolution.go](../../internal/engine/resolution.go), align `IsExplicit` and
+the explicitness decision in `Default` with the frozen, shape-aware entry metadata
+already used by `InputInfo` in
+[load_metadata.go](../../internal/engine/load_metadata.go).
+
+Reuse the existing semantic-path matching machinery. A shared private lookup can
+avoid divergent presence decisions without requiring every public query to have
+identical error behavior. In particular:
+
+- Preserve `IsExplicit`'s schema-path validation and its existing diagnostics.
+  `InputInfo` remains a query that returns false for missing or mismatched paths
+  without those diagnostics.
+- Preserve `Default`'s helper-target validation, assignment, recording, and final
+  helper-value checks. Correct its presence gate; do not bypass the other checks.
+- Use entry-state shape even after application code changes the working value.
+  Explicit false, zero, empty, and null retain their existing presence meaning.
+- Keep callback lifetime and detached metadata behavior intact.
+
+A shape mismatch beneath a schema-valid dynamic path must not become explicit
+merely because its pointer spelling matches an input path. Contextually parsed
+numeric tokens continue to resolve against the entry state's actual containers.
+
+### 2. Preserve Explicitness By Semantic Path During Resolution
+
+Update resolution metadata application and field refresh in
+[load_metadata.go](../../internal/engine/load_metadata.go) to compare the relevant
+entry and final shapes before reusing presence information.
+
+The comparison must account for container kinds along the full path. A mapping
+key becoming a sequence index at an ancestor can change a descendant's semantic
+path even when its last segment and full pointer spelling stay unchanged.
+
+Keep explicitness for matching surviving paths, including descendants covered by
+an ancestor normalization record. New semantic paths introduced by resolution
+remain non-explicit, and retain the applicable recorded resolution origin.
+Removed paths leave effective metadata while remaining available to historical
+queries under the existing contract. Do not clear every descendant's presence
+just because its parent was normalized.
+
+Use the existing before/after projection and entry metadata where possible.
+This work does not require tracking every transient mutation inside a callback;
+the existing net-change and coalesced-record model continues to apply.
+
+### 3. Build A Shared Internal History Model
+
+#### Evidence And Ownership
+
+Give historical evidence one private owner in the engine. Composition,
+inheritance, decoding, CLI assignment, and resolution continue to perform their
+existing work and supply evidence to that owner. The model does not drive a
+second configuration load or repeat application callbacks.
+
+Organize the evidence into ordered groups with the phase boundaries required by
+the existing inspection contracts:
+
+| Evidence group | Information to preserve |
+| --- | --- |
+| Default baseline | Canonical typed application defaults or Go-zero values and their default attribution. |
+| Source contributions | Underlay, main-tier, and selected-overlay contributions in their existing order; exact definitions, composite origins, displaced definitions, raw policy-validated values, and sequence-position projections. |
+| Inheritance | Parent transfers, child contributions including equal-valued replacements, controlling directives and their consumption, identity-field preservation, and contribution origins/dependencies. |
+| Typed checkpoints | Pre-CLI and post-CLI typed projections and attribution needed to reconcile retained defaults and raw contributions with decoded values. |
+| CLI contributions | Explicit assignments, including equal-valued assignments, raw supplied representation, source attribution, and affected paths. |
+| Resolution | Original recorded operations, reasons, dependencies, and before/after evidence, with the grouping needed to explain overlapping records as net subtree changes. |
+
+The logical order remains baseline, source contributions, inheritance, pre-CLI
+decoding reconciliation, CLI contributions, post-CLI typed reconciliation, then
+recorded resolution. The two typed checkpoints belong on their respective sides
+of the CLI contribution phase.
+
+The groups describe logical inspection chronology, not a transcript of every
+physical mutation. Retain the existing net-change and coalescing rules for
+resolution. Preserve superseded raw input that need not decode into the final
+application type; typed checkpoints do not replace that evidence.
+
+Use a small set of ordinary private structures and existing document/path
+mechanisms. Exact types, indexes, and storage layout belong in the implementation
+plan. Complete evidence means enough for the existing history, explanation, and
+internal attribution consumers; it does not add persistence, a public event API,
+or recording of transient application effects.
+
+#### Shared Selection And Attribution
+
+Retain concrete semantic-path evidence when it is available. A query over the
+model must distinguish exact definitions from operations affecting an ancestor
+or descendant, determine the relevant previous origins, and apply sequence
+contribution coordinates at their assigned positions.
+
+The affected semantic path and the operation's reported path are distinct facts.
+For example, a replacement at `/shape` can remove a mapping key and introduce a
+sequence index with the same pointer spelling. Preserve enough before/after shape
+evidence to answer each explicit query correctly. Merely filtering `Change.Path`
+after concrete segment kinds have been discarded is insufficient.
+
+Make this selection and attribution common to both projections, including
+inheritance integration and `Previous` reconstruction. A false definition must
+not survive as a previous origin after it has been excluded from the query's
+history. Preserve genuine ancestor replacement/removal effects and chronology.
+A path that never existed does not acquire history solely because an ancestor
+changed.
+
+Keep the contextual meaning of parsed numeric tokens across historical states
+and the one-entry-per-canonical-pointer enumeration of `HistoricalPaths`.
+Indexes and compact derived records may remain, but their attribution must come
+from the shared evidence. Metadata queries should not require full configuration
+replay or YAML rendering to obtain definition and origin records.
+
+#### Distinct Public Projections
+
+`Metadata.History` produces detached definition, inheritance, and resolution
+records from the model. `Explain` reconstructs and safely renders values from
+the same evidence, retaining its existing decoding and grouping behavior. Neither
+public API should be implemented by translating the other's public result.
+
+Preserve these intentional differences:
+
+| Case | Metadata history | Explanation |
+| --- | --- | --- |
+| Several strict contributions to a mapping | Composite definition/origin information. | Each contribution's effect on the mapping. |
+| Raw input differs from canonical typed values | Definition and attribution records. | The applicable decoding reconciliation transition. |
+| Overlapping resolution records | Specific recorded attribution under the existing coverage rules. | One net subtree event with combined reasons and dependencies. |
+| Equal-valued input or inheritance operations | Explicit definition and operation attribution remains visible. | Relevant operations remain visible even when before and after values are equal. |
+
+Agreement means consistent semantic identity and attribution, not identical
+record lists. Retain original resolution records as evidence from which net
+replay groups are derived; a grouped explanation alone loses information needed
+by metadata.
+
+#### Replace Existing Attribution Machinery
+
+The common model must replace independent historical attribution, rather than
+become another representation alongside it. The implementation plan should map
+the following owners to their replacements:
+
+| Current owner | Required change |
+| --- | --- |
+| History-building portions of `metadataFromDecoded`, `applyOverlayMetadata`, and `appendPostDecodeCLIHistory` in [load_metadata.go](../../internal/engine/load_metadata.go) and [override_context.go](../../internal/engine/override_context.go) | Produce or project shared evidence instead of independently reconstructing pointer-keyed histories. Retain effective-field construction where it is still needed. |
+| Previous-origin reconstruction and inheritance insertion in [inheritance_history.go](../../internal/engine/inheritance_history.go) | Use common semantic selection and attribution while preserving the existing inheritance projection rules. |
+| Relevance and chronology orchestration in [explain.go](../../internal/engine/explain.go) | Consume the shared model; retain useful value reconstruction and redaction helpers. |
+| Separate published inheritance evidence for metadata and explanation in [load_result.go](../../internal/engine/load_result.go) | Consolidate under the frozen history owner where independent copies are unnecessary; keep required output detachment. |
+| Original and coalesced resolution evidence in [history.go](../../internal/engine/history.go) | Own original records in the model and derive the replay grouping without losing specific attribution or maintaining a second independently interpreted history. |
+
+Existing source, sequence-combination, and inheritance evidence producers may
+remain useful. Their algorithms need not be rewritten merely to contribute to
+the model. Likewise, final effective fields, the frozen resolution-entry view,
+source records, and final configuration snapshots retain their responsibilities.
+Findings 1 and 2 still require explicit presence corrections.
+
+#### Other Consumers, Lifetime, And Resource Accounting
+
+Move historical dependency reads in
+[secret classification](../../internal/engine/secret.go) and historical location
+fallbacks in [override_context.go](../../internal/engine/override_context.go) to
+the shared evidence. Do not retain the old history machinery solely for those
+readers. Preserve existing sensitivity propagation and historical classification;
+correcting attribution does not authorize narrowing redaction for paths that
+share a pointer spelling.
+
+Evidence must be available at the lifecycle stages where these internal readers
+need it, including diagnostic attribution before publication. Keep the callback's
+entry metadata fixed while later resolution evidence is added. Finalize the
+model for published queries without introducing lazy mutation of shared state.
+Private frozen evidence may be shared by the two views while their returned
+records remain detached under the existing ownership contract.
+
+Reuse retained trees and existing lookup machinery where practical. Review
+retained-history accounting against the resulting ownership graph, including
+typed checkpoints and inheritance/resolution evidence. The model does not require
+a full document copy per event, a new quota, or a new concurrency guarantee.
+Reduced duplication is a design benefit to evaluate, not a promised memory or
+speed improvement.
+
+### 4. Preserve Composition Claimant Order Through Diagnostics
+
+Keep the order produced by `compositionConflictDiagnostic` in
+[compose.go](../../internal/engine/compose.go) through accumulation, `LoadError`,
+facade conversion, and `WriteDiagnostics` in
+[diagnostic.go](../../internal/engine/diagnostic.go).
+
+Prefer a narrow internal canonicalization rule based on the diagnostic semantics
+already carried across these boundaries. Identify the composition diagnostic
+producers governed by the claimant-order contract rather than changing all
+related-location lists. Preserve canonical ordering of diagnostic records and
+the existing sorting of related locations for other diagnostics.
+
+The root renderer converts public diagnostic records back to engine records and
+canonicalizes them again. An engine-only marker discarded by facade conversion
+would therefore be insufficient. No public ordering option is needed for this
+correction. Keep primary selection, conflict detection, acquisition order, and
+dependency-ready scheduling unchanged.
+
+## General Implementation Order
+
+1. Prepare the concrete shared-evidence design and focused regression cases.
+   Map existing producers and consumers to the new owner, distinguish common
+   attribution from presentation rules, and identify the old paths to remove.
+   Select representations and indexes that preserve existing query and lifetime
+   needs without adding public contracts.
+2. Repair entry-state presence queries and `Default` (finding 1). This is a local
+   correction that can proceed independently of the larger history work.
+3. Repair final explicitness preservation (finding 2), retaining the shape evidence
+   needed by metadata updates. Coordinate this work with history construction so
+   the two fixes do not introduce competing definitions of path identity.
+4. Build the shared model and migrate metadata history, explanation, sensitivity
+   dependencies, and diagnostic-location readers. Preserve phase evidence and
+   derive both public projections through common semantic attribution. Remove
+   the superseded reconstruction paths and reconcile retained-history accounting.
+   Verify finding 3 and its interaction with final presence metadata; repairing
+   pointer buckets alone does not complete the selected architecture.
+5. Complete composition diagnostic ordering (finding 4) as an independent change;
+   it may proceed alongside the path work. Check rendering as well as the
+   structured error before considering it complete.
+6. Review the integrated result, run applicable development checks, and update
+   affected implementation documentation and finding status.
+
+These are high-level dependencies, not a requirement to combine unrelated fixes
+in one commit. The implementation plan should select concrete edits and test
+commands while keeping the four findings individually traceable.
+
+## Validation And Completion
+
+Follow the [testing policy](../policy/testing.md) and use the
+[development guide](../development.md#baseline-validation) for checks. Extend
+existing test owners rather than recreating full internal matrices at the facade.
+The focused cases should cover:
+
+- Both mapping-key/index mismatch directions beneath dynamic values, including
+  explicit paths and contextual parsed pointers. Check callback presence and
+  `Default` behavior, including the fixed entry view after working-value changes.
+- Both container-shape transitions during recorded resolution, a descendant
+  affected by an ancestor shape change, and a same-shape normalization control.
+  Check explicitness and recorded origins through final metadata and explanation.
+- Shape-changing CLI and resolution history, checking exact definitions,
+  `Previous`, and genuine ancestor replacement/removal. Include contextual
+  enumeration/query coverage and preserve existing inheritance and overlay cases;
+  add another case only where those integrations expose an unprotected risk.
+- Shared-model projections across composite mapping origins, typed decoding and
+  default reconciliation, equal-valued writes, inheritance controls/transfers,
+  mapped sequence replacements and removed descendants, and overlapping
+  resolution records. Reuse existing cases where they protect those semantics;
+  check the intentional differences between the public projections as well as
+  agreement on attribution.
+- Historical sensitivity/dependency propagation, diagnostic source-location
+  fallback, detached returned records, and complete retained-history accounting
+  under the new ownership graph. Use small controlled budget cases and existing
+  disclosure fixtures rather than memory or timing thresholds.
+- A composition conflict whose acquisition and assembly orders differ, using the
+  audit fixture. Check both published diagnostics and rendered related-location
+  order, while retaining coverage for ordinary diagnostic canonicalization.
+
+Retain existing checks for invalid schema paths, callback lifetime, detached
+records, safe historical rendering, and default/null presence semantics. Test
+observable behavior rather than a particular private index or number of copies.
+No benchmark campaign, broader audit, or new release gate is part of this roadmap.
+
+Completion requires all four findings to have implementation and regression
+evidence, appropriate development checks, and both history interfaces deriving
+semantic attribution from the shared model. Review the final producer/consumer
+map to confirm superseded independent reconstruction has been removed and
+internal attribution readers have migrated. Passing the four defect regressions
+while retaining the old parallel attribution machinery is not completion.
+
+The audit's previous passing suite is not validation of these future changes.
+Scope here does not promise identical output from different inspection interfaces
+or a universal guarantee about unrelated path-processing code.
+
+## Decision Status
+
+OD-1 is settled: build one shared internal history model with distinct metadata
+and explanation projections. The design investigation found that both views
+require evidence the other's public representation does not preserve. Shared
+ownership can replace independent attribution and redundant retained evidence,
+while keeping the current lifecycle and public contracts.
+
+The narrower repair of stored history events would leave the separate replay
+attribution owner in place. Promoting current explanation output to authoritative
+history would lose composite-definition and specific resolution information.
+The selected common underlying model addresses both concerns. This accepts a
+larger internal refactor than a local shape-filtering fix.
+
+No significant open decision remains at roadmap level. Private representation,
+indexing, and helper choices belong in the implementation plan. This roadmap
+records future implementation ownership; it does not revise accepted ADRs or
+declare the model implemented.

docs/roadmap/interfaces.md

@@ -0,0 +1,202 @@
+# API and Interface Contract Audit
+
+Reviewed baseline: `01b08e76a3fecc22c4414437bfeaa07b4b7a85aa`.
+
+Status: four open findings. This is an audit record, not an implementation plan.
+No production code or repository tests were changed.
+
+The [interface remediation roadmap](interface-remediation.md) describes the
+settled target, required changes, and dependencies, including a shared internal
+history model for metadata history and explanation. Implementation remains pending.
+This audit retains the findings and reproduction evidence.
+
+## Scope and evidence
+
+Reviewed the root-owned public facade and its conversions, engine loading and
+callback boundaries, source/discovery and filesystem ownership, schema and typed
+value handling, metadata/history, safe rendering and comparison, secrets, and
+the document/YAML boundary against declaration comments, the
+[configuration contract](../config.md), and applicable architectural decisions.
+The root delegates to the synchronous engine lifecycle; parser-independent
+documents support source assembly and inheritance, and completed results retain
+detached configuration and frozen inspection state.
+
+The findings below were reproduced through the public API with a disposable Go
+program outside the repository, using labeled byte sources and `fstest.MapFS`.
+The full existing suite passed with Go 1.26.10 on Linux:
+
+```sh
+GOTOOLCHAIN=local GOTELEMETRY=off GOPROXY=off go test ./... -count=1
+```
+
+The existing tests cover many ownership, lifetime, error, and zero-value cases,
+but do not detect these cross-interface discrepancies. No benchmarks, external
+service calls, or new test files were needed. The recommendations restore
+existing contracts rather than introduce new API restrictions or guarantees.
+
+## 1. Resolution presence queries confuse explicit keys with sequence indexes
+
+Priority: P2. Status: open.
+
+Relevant code:
+
+- [`Resolution.IsExplicit`](../../internal/engine/resolution.go), lines 75–82.
+- `Resolution.InputInfo` and `resolutionInput.lookup` in
+  [resolution.go](../../internal/engine/resolution.go) and
+  [load_metadata.go](../../internal/engine/load_metadata.go).
+- `resolveSchemaPath` in [schema.go](../../internal/engine/schema.go).
+
+The public [path contract](../../path.go) distinguishes explicit mapping keys
+from sequence indexes. The [resolution contract](../config.md#recorded-application-resolution)
+defines surviving-input presence and a fixed entry-state metadata view;
+[metadata queries](../config.md#layered-metadata) preserve container shapes.
+
+`IsExplicit` validates against the schema but then looks up presence using only
+`path.String()`. In dynamic `any` state, schema validation cannot establish the
+actual container shape. Consequently these distinct queries collide:
+
+```go
+key := configkit.Keys("shape", "0")
+index := configkit.Keys("shape").Index(0)
+```
+
+Load `map[string]any` from `shape: [input]` with a resolver that queries both
+paths, without changing the configuration. Observed results:
+
+| Query | `Resolution.IsExplicit` | `Resolution.InputInfo` found | Final `Metadata.IsExplicit` |
+| --- | --- | --- | --- |
+| Explicit mapping key | true | false | false |
+| Explicit sequence index | true | true | true |
+
+The mapping-key path never existed and should not be reported as explicit.
+Applications branching on presence receive inconsistent answers depending on
+which public query they use, even during an otherwise unchanged load.
+
+Potential remediation: use the retained entry-state semantic shape when checking
+resolution presence, consistently with `InputInfo` and `Metadata.IsExplicit`.
+Review the same string-only presence lookup in `Default` while preserving its
+existing assignment and callback-lifetime rules. Add focused public coverage for
+both mapping-to-index and index-to-mapping mismatches under dynamic values,
+including contextually parsed pointers. This does not require changing the
+documented treatment of invalid schema paths.
+
+## 2. Resolution can transfer explicitness to a different semantic path
+
+Priority: P2. Status: open.
+
+Relevant code: `applyResolutionMetadata`, `applyResolutionFieldOrigins`, and
+`refreshResolutionFields` in
+[load_metadata.go](../../internal/engine/load_metadata.go), especially lines
+371–387; final shape checking in
+[`Metadata.effectiveField`](../../internal/engine/metadata.go).
+
+The [metadata contract](../config.md#layered-metadata) distinguishes key/index
+shapes and keeps derived values for absent input non-explicit. Recorded
+resolution changes origins without creating input explicitness.
+
+Reproduction:
+
+1. Load `map[string]any` from `shape: {'0': old}`.
+2. In resolution, replace `configuration["shape"]` with `[]any{"new"}` and
+   record normalization at `Keys("shape")`.
+3. Query final metadata using the two paths from finding 1.
+
+The old explicit mapping-key path correctly stops being effective, but the newly
+created sequence index reports `Explicit: true`. That sequence index was never
+supplied by input. `refreshResolutionFields` reuses the old `FieldInfo` whenever
+its textual pointer survives; it updates null state without checking whether
+the old and new containers give that pointer different semantic shapes. Final
+lookup validates only the new shape, so it cannot recover the lost distinction.
+
+This can mislead consumers inspecting whether a final value was explicitly
+configured rather than introduced by application resolution. The converse
+sequence-to-numeric-key transition has the same underlying collision.
+
+Potential remediation: retain entry/effective shape evidence across resolution
+and carry input explicitness forward only for the matching semantic path.
+Preserve existing explicitness for same-shape paths and the recorded resolution
+origin; do not clear every descendant's presence merely because an ancestor was
+normalized. Cover both shape transitions and agreement among `Lookup`,
+`IsExplicit`, and explanation. Fixing finding 1's callback query alone does not
+repair this publication path.
+
+## 3. Metadata history mixes numeric-key and sequence-index definitions
+
+Priority: P2. Status: open.
+
+Relevant code:
+
+- [`Metadata.History`](../../internal/engine/metadata.go), lines 56–65.
+- `metadataFromDecoded`, `applyOverlayMetadata`, and history event construction
+  in [load_metadata.go](../../internal/engine/load_metadata.go).
+- Shape-aware replay in [`Inspection.Explain`](../../internal/engine/explain.go).
+
+The [metadata contract](../config.md#layered-metadata) promises semantic
+key/index distinctions for historical as well as effective queries. The
+historical-path index retains both shapes, but `History` uses that index only to
+check that the query was ever known. It then retrieves the entire history bucket
+by textual pointer, without selecting the query's shape.
+
+Reproduction: load `shape: {'0': old}` into `map[string]any`, then apply a CLI
+override at `Keys("shape")` with `[]any{"new"}`. Both explicit shapes were
+observed and share `/shape/0`.
+
+`History(Keys("shape").Index(0))` incorrectly includes the original base
+definition of the mapping key, and identifies that base origin as the previous
+origin of the CLI operation. The old input defined a key, not an index. In the
+same successful result, `Explain` for the explicit index correctly shows only
+its CLI introduction. `History` for the key and index returns identical records
+despite their different lifetimes.
+
+Consequences include false source attribution and disagreement between the two
+public history interfaces when consumers inspect removed or introduced paths.
+
+Potential remediation: retain enough container-shape evidence with stored
+history events, or select relevant events from existing shape-aware replay
+evidence. Filter exact definitions and previous origins by the queried semantic
+shape while retaining genuine ancestor replacement/removal events. Filtering
+only on the current `Change.Path` is insufficient where construction already
+replaced concrete segment kinds with parsed tokens. Preserve the intentional
+one-pointer-per-entry `HistoricalPaths` enumeration and contextual behavior of
+`ParsePath`; neither public contract needs redesign. Cover both direction changes
+and compare metadata history with explanation.
+
+## 4. Diagnostic canonicalization breaks composition claimant ordering
+
+Priority: P3. Status: open.
+
+Relevant code: `compositionConflictDiagnostic` in
+[compose.go](../../internal/engine/compose.go) and `canonicalDiagnostics` in
+[diagnostic.go](../../internal/engine/diagnostic.go), particularly lines 88–92.
+The latter runs both while constructing `LoadError` and in `WriteDiagnostics`.
+
+The [strict-composition contract](../config.md#strict-document-composition)
+specifies the first claim in scoped assembly order as primary and every remaining
+claim in the same order as related locations.
+
+A native-filesystem fixture demonstrates the discrepancy:
+
+```text
+main.yml imports a.yml, b.yml, c.yml
+a.yml imports z.yml
+b.yml, c.yml, and z.yml each define x
+```
+
+Acquisition assigns main=source-1, a=source-2, z=source-3, b=source-4,
+c=source-5. Dependency-ready assembly considers the three conflicting claims
+in b, c, z order. The composer correctly chooses b as primary and constructs
+related locations c, z. Generic canonicalization sorts those related locations
+by acquisition ID, publishing z, c instead. The reproduced public diagnostic
+has primary source-4 and related source-3, source-5.
+
+All claimants remain present and the conflict still fails loading, so this is
+lower impact than the presence/history defects. It nevertheless violates the
+documented reporting order and gives consumers two different ordering rules
+within one conflict report.
+
+Potential remediation: preserve the selected semantic ordering for composition
+claimant lists through accumulation and rendering, while retaining canonical
+ordering of diagnostic records and the established sorting of related locations
+for diagnostics without a semantic-order contract. Add one public failure-path
+case whose acquisition and claimant orders differ and verify both structured
+diagnostics and rendered output. No scheduler or precedence change is needed.

docs/roadmap/internal-boundaries.md

@@ -0,0 +1,186 @@
+# Import Discovery And Facade Boundary Simplification
+
+Status: target defined; execution is tracked in the [implementation plan](implementation.md).
+Baseline: `80cf23e7102fca2c5d7c7621e75f930a1ef1ecad`.
+
+This roadmap brings together the two refactors deferred until after the audit
+remediation sprint: glob entry preparation (**C03**) and facade-to-engine
+transport simplification (**F07**). Their original motivation is retained in the
+[glob](future.md#1-separate-glob-entry-preparation-from-traversal) and
+[transport](future.md#2-simplify-redundant-facade-to-engine-transport) future-work
+entries and the [consolidated audit](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md).
+
+The goal is clearer responsibility and ownership with fewer redundant internal
+representations. Both changes preserve the public API and configuration model.
+They are independent work areas within one maintenance effort, not a new loading
+workflow. The [implementation plan](implementation.md) owns concrete edits,
+stages, validation commands, and execution progress.
+
+## Baseline And Targeted End State
+
+The preceding sprint repaired directory-failure classification and shared link
+dependency evaluation, clarified acquisition handoffs, and removed unused
+placement bridges. Those changes provide the baseline; this work does not repeat
+or redesign them.
+
+| Area | Current baseline | Target |
+| --- | --- | --- |
+| Glob discovery | `selectGlobImportCandidates` interleaves traversal and set operations with substantial per-entry preparation. | The selector coordinates discovery; a cohesive private operation prepares a wildcard-discovered entry. |
+| Facade transport | Import defaults pass through root conversion, an exported engine transport record, an equivalent private engine record, and repeated snapshots. Placement construction also uses setter-style bridges. | Root-owned public records convert explicitly to one engine defaults representation, with straightforward engine option/input construction and identifiable ownership copies. |
+
+The [architecture policy](../policy/architecture.md), [configuration
+contract](../config.md), and public declarations remain authoritative. [ADR
+0006](../adr/0006-public-facade.md) continues to govern the facade boundary. These
+refactors do not require a new architectural decision or changes to its accepted
+decision. The [component overview](../internal/overview.md) and [import
+internals](../internal/imports.md) own implemented structure.
+
+## Glob Entry Preparation
+
+### Target Responsibility Split
+
+In [glob.go](../../internal/engine/glob.go), the selector continues to own pattern
+validation, component advancement, literal-segment handling, directory
+enumeration and its accounting, candidate union and sorting, final target
+validation, exclusions, and requiredness.
+
+A private glob-specific preparation operation owns the patterned branch's work
+for one directory entry: matching and hidden-entry filtering, lexical/access path
+construction, containment checks, target inspection and discovered-link policy,
+intermediate-directory eligibility, host/native access-path selection, labels,
+and candidate construction. Its result distinguishes a selected candidate, an
+ordinary skip, and a diagnostic. It receives the facts needed for that entry
+rather than mutating the selector's candidate set or surrounding traversal state.
+
+Use the existing filesystem access and link-resolution helpers. Directory
+selection already has `prepareImportDirectoryEntry`; it remains a separate owner
+because its recursion, matching, absence handling, and traversal state differ.
+Share a smaller operation only where it actually expresses the same semantics
+and reduces duplication. No common visitor or discovery strategy is needed for
+this extraction.
+
+### Behavior To Preserve
+
+Preserve the current distinctions and processing order documented by
+[explicit imports](../config.md#explicit-imports) and exercised by the selection
+tests:
+
+- Literal components remain explicit selections; discovery filters apply at
+  their existing wildcard boundaries. Moving matching or hidden filtering after
+  filesystem inspection would change which errors are observable.
+- Keep lexical candidate identity separate from resolved access identity.
+  Native source identity and relative imports retain namespace spelling; host
+  access retains its existing canonicalization and containment behavior.
+- Keep discovered-link follow rules, intermediate-directory eligibility,
+  missing-target handling, and structured diagnostic sites/classifications.
+  Do not make glob and directory rules identical as a side effect of extraction.
+- Preserve deterministic candidate and error order, union/deduplication,
+  directory-entry accounting, and the existing point of source admission.
+  Final selected-file validation remains before exclusions where it is today;
+  exclusions do not hide discovery or resource errors.
+- Retain the preceding sprint's distinction between missing-on-open directories
+  and failures after opening, along with operation-local completed-link reuse.
+
+The intended benefit is that a maintainer can understand traversal separately
+from entry eligibility and identity preparation. This is a readability change;
+it does not select a different search algorithm or claim a speedup.
+
+## Facade-To-Engine Transport
+
+### Target Representations
+
+The root keeps its public `ImportDefaults`, `PlacementMode`, `Options`, `Input`,
+and `LayeredOptions` types. Explicit conversions remain appropriate at this
+boundary; consumers must not acquire engine type identity, fields, or methods.
+
+Inside the engine, consolidate the equivalent `ImportDefaults` transport and
+private `importDefaults` records into one presence-aware input-defaults
+representation. Carry that representation through the ordinary and layered
+options, normalized request, and acquisition policy instead of translating the
+same five fields between equivalent engine records.
+
+Keep `resolvedImportPolicy` distinct: it represents concrete per-declaration
+choices, including selector-specific rules, rather than another copy of optional
+load-wide input. Likewise, registration compilation and the normalized loading
+request have real responsibilities and are not replacement targets.
+
+Make placement and defaults straightforward to construct on the existing engine
+option/input records. Remove decorator-style bridges made unnecessary by that
+construction surface. Exporting fields or types within `internal/engine` where
+needed by the root converter does not make them part of ConfigKit's supported
+public API. Keep explicit enum conversion and preserve unknown values until the
+existing option validation can report them; conversion must not silently choose
+a default or move validation into facade setters.
+
+### Ownership And Copying
+
+Make the actual mutable ownership boundaries identifiable. Public load calls
+continue to snapshot pointed-to default booleans and the pattern slice before
+source acquisition, as specified by the [public
+declaration](../../lifecycle.go). Preserve omitted, explicit false, nil pattern,
+and non-nil empty pattern as distinct inputs where they are distinct today.
+
+Review the entire handoff, not just the duplicate type definitions: root
+conversion in [conversion.go](../../conversion.go), ordinary construction in
+[lifecycle.go](../../lifecycle.go), layered construction in
+[layered.go](../../layered.go), engine options and normalization, acquisition
+policy construction, and declaration resolution in
+[import_spec.go](../../internal/engine/import_spec.go).
+
+Remove snapshots that merely recopy already-owned read-only defaults as they
+cross those private representations. Keep copies needed when establishing
+independent mutable ownership, including any independently owned resolved
+policy. The target is explainable ownership and less redundant work, not an exact
+copy count or removal of copying throughout the library. Shared defaults within
+one load can be read-only by implementation discipline; this creates no new
+caller immutability or concurrency requirement.
+
+Preserve ordinary and layered option preflight before source reads and application
+hooks, independent declaration overrides, transitive/underlay propagation,
+selected-overlay behavior, destination normalization, and invalid-enum
+diagnostics. Keep callback, source, path, secret, and returned-record conversions
+whose ownership or domain meaning is not redundant transport.
+
+The intended benefit is fewer concepts, field mappings, and construction paths
+to maintain. Reduced copying is useful but does not establish an end-to-end
+performance guarantee.
+
+## Documentation Ownership
+
+Internal documentation describes the resulting discovery and transport
+boundaries once implemented. Public declarations continue to own the existing
+snapshot and type contracts; this effort selects no new public behavior.
+The corresponding C03/F07 future-work entries serve as navigation to this
+roadmap rather than parallel target or progress inventories. Unrelated future
+candidates remain separate. The implementation plan owns the documentation
+updates and retirement of replaced private representations and helpers.
+
+## Validation And Completion
+
+Follow the [testing policy](../policy/testing.md) and [development
+guide](../development.md#baseline-validation). Reuse the existing glob selection,
+directory/link failure, import-default, placement, public facade, and dependency
+boundary coverage. Add a focused behavioral case only where a concrete
+extraction or ownership risk lacks protection.
+
+For discovery, verify literal versus wildcard filters, host/native identity,
+contained and escaping links, optional absence versus read/close failures,
+ordering, exclusions, regular-file checks, and existing resource accounting. For
+transport, verify nil/empty and explicit-false distinctions, mutation of caller
+default storage after the load snapshot through a supported callback, invalid
+options before input consumption, both loading APIs, and transitive/underlay
+defaults. Existing conversion-focused tests may change with private construction;
+preserve the useful ownership behavior rather than fossilizing a helper's shape.
+
+Completion requires an easier-to-follow selector, one engine input-defaults
+representation, fewer redundant mapping/construction sites, justified remaining
+copies, and preserved public behavior under the relevant checks. Review source
+and ownership flow as well as passing tests; moving complexity into a helper or
+renaming a duplicate record alone is insufficient.
+
+No benchmark campaign, complexity score, allocation threshold, exact filesystem
+call trace, new resource limit, or additional release gate is required. This work
+adds no dependencies, caches, background activity, or new consumer controls, and
+does not reopen other audit findings or deferred performance experiments. Routine
+helper signatures, names, and file placement belong to the implementation plan;
+no significant product or architectural decision remains open for this scope.

docs/roadmap/roadmap.md

@@ -0,0 +1,276 @@
+# Audit Remediation Roadmap
+
+Status: target accepted; completed execution is recorded in the historical plan
+at `bb5540b:docs/roadmap/implementation.md`.
+Prepared: 2026-10-25. Planning baseline: `e1e94ee`.
+
+This roadmap describes the target state and related work for the 29 entries
+selected in the [consolidated audit review](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#settled-scope-decisions).
+It is intended for maintainers reviewing the remediation's scope and architecture.
+
+## Document Ownership
+
+- [Archived audit](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md) records finding IDs, original evidence, detailed remedies and
+  acceptance cases, and the settled scope decisions. Its evidence has not been
+  independently reproduced by writing this roadmap.
+- This document owns the combined target state, work areas, and design intent.
+  Finding IDs below provide traceability without repeating the full audit inventory.
+- The historical implementation plan at
+  `bb5540b:docs/roadmap/implementation.md` records concrete stages, private
+  implementation choices, edits, validation commands, and completion evidence.
+
+The [configuration contract](../config.md), public Go declarations, and
+[architecture policy](../policy/architecture.md) remain the owners of behavior
+and guarantees. Update their applicable implementation documentation when work
+lands; this planning document does not redefine their contracts.
+
+## Current Baseline And Targeted End State
+
+The existing loading model remains the foundation. Public loading APIs delegate
+to the shared engine lifecycle; acquisition retains source scopes for composition,
+then precedence, inheritance, decoding, CLI overrides, application resolution,
+secret lookup, and validation produce a detached result with inspection metadata.
+The [component overview](../internal/overview.md) describes the current owners.
+
+The audits identify failures at particular traversal, lifetime, error, and
+accounting boundaries, alongside redundant representations and repeated work.
+The target is to repair those boundaries and simplify their implementation while
+preserving ConfigKit's public API and configuration model.
+
+| Current gap | Targeted outcome |
+| --- | --- |
+| Recursive descriptors can fail to terminate, failed dynamic builds can leave partial state, and slice aliases can be mistaken for cycles. | Traversal terminates for the reported cases, failed construction leaves reusable state valid, and supported acyclic values remain accepted. |
+| Callback and owned-root cleanup can be skipped during panic unwinding; directory errors lose their failure phase. | Cleanup follows the lifetime of each owned resource, and selected-directory failures retain the information needed for correct error handling. |
+| Finite symlink dependencies can be expanded repeatedly. | Completed dependencies are reused within a resolution while active cycles remain detectable. |
+| Projection reservations and published history can escape existing accounting. | Existing budgets govern the relevant allocations and complete retained document graph. |
+| Collection-policy work repeatedly scans or copies unrelated state. | Policy preparation, application, and retained evidence have clear ownership, with work localized to affected policy. |
+| Acquisition has redundant paths and difficult handoffs; several private representations have no production consumer. | Production paths and their tests share the active protocol, with cohesive handoffs and unused representations removed. |
+| Some lookups repeat full scans, and some tests assert incidental representation or miss meaningful mutations. | Local indexing avoids the identified repeated work, and tests protect observable contracts at useful boundaries. |
+
+This is a bounded remediation effort, not a redesign of configuration semantics.
+Correcting erroneous acceptance, rejection, cleanup, or diagnostics in the
+reported cases restores existing contracts. It does not select new loading APIs,
+configuration options, limits, background work, or persistent caches.
+
+## Selected Work
+
+The following areas account for all 29 selected entries. Detailed examples and
+acceptance cases remain with their linked audit group.
+
+### 1. Safe Typed-Value Traversal And Descriptor Reuse
+
+Selected entries: **A01, A02, A03, A04, A05** in
+[schema and typed values](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#a-schema-and-typed-values).
+
+Make pointer-only descriptor traversal terminate in decoding and path resolution,
+while preserving supported null values and ordinary recursive structures. Make
+dynamic descriptor construction transactional: a failed build must not leave
+partial descriptors available to subsequent operations. Staging successful
+additions or rolling back the complete failed addition set are implementation
+choices; both must handle recursive references and preserve prior valid entries.
+
+Correct active-path cycle detection to distinguish finite slice views sharing
+storage from actual cycles. Keep detachment and nil/empty behavior intact across
+the callers that share this traversal machinery.
+
+Enforce existing projection budgets before large container reservations. Include
+map-key materialization and overflow-safe lower-bound checks in the review; a
+check after the reservation does not address the reported failure. This work
+does not introduce a general heap limit or reject large containers independently
+of existing budgets.
+
+Dynamic-type eligibility reuse depends on safe descriptor construction and retains
+completed successful checks within the value operation. Keep failure diagnostics
+specific to their current path and distinguish completed results from traversal visits.
+This removes repeated descriptor-graph walks without a cross-load cache.
+
+### 2. Callback, Filesystem, And Dependency Lifetimes
+
+Selected entries: **B01, B02** in [callback lifetime](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#b-callback-and-filesystem-lifetime)
+and **C01, C02** in [filesystem traversal](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#c-import-selection-and-link-resolution).
+
+Tie callback-context deactivation to the callback invocation with deferred cleanup
+so it runs on normal return and panic unwinding. Independently give the owned
+import root a deferred cleanup fallback. Preserve normal close-error handling,
+caller ownership, and propagation of the original callback panic. Fixing either
+state lifetime or handle lifetime does not fix the other.
+
+Preserve the phase of a directory failure. A genuinely absent directory before
+successful opening remains subject to existing permissiveness rules; a later
+read or close failure must not be treated as absence. Apply this distinction to
+both supported directory-reading paths. Strengthen the shared filesystem tests
+to check each handle's closure and structured, non-disclosing diagnostics,
+including partial reads and failures masked by another successful directory.
+Do not change regular-file close-error policy as a side effect.
+
+Avoid repeatedly expanding completed finite symlink dependencies in both host
+and native filesystem resolution. Use operation-local memoization of completed
+targets, or an equivalent dependency evaluation, with keys that respect the
+resolution namespace. Keep active-path cycle detection separate from completed
+work. Preserve containment, supported finite reuse, and existing host-path
+semantics. The fix must work for directory and glob selection without a new
+arbitrary traversal cap or changed symlink defaults.
+
+### 3. Retained History And Combination Evidence
+
+Selected entries: **D01, D02, D03, D04** in
+[history and retained state](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#d-retained-history-and-unused-evidence).
+
+Remove the unconsumed override-result pre-CLI copy and the unqueried published
+structural-input copy. Preserve the separate pre-CLI state used by active replay
+and the structural input still used before publication for metadata and secret
+classification. Remove unused combination-evidence members and the work needed
+solely to produce them, retaining projections, mappings, origins, and snapshots
+that have active consumers.
+
+With retained roots clarified, enforce the existing history allowance against
+the complete graph published through inspection, including replay and inheritance
+state and publication-created copies. Count shared nodes once by pointer identity;
+distinct retained copies still consume the allowance. Account for required copies
+before creating them where practical. Removing an unused tree alone does not
+resolve the accounting omissions.
+
+Keep required before/after evidence and existing inspection behavior. Validate
+the complete publication path with small controlled allowances, including the
+interaction of CLI, resolution replay, and inheritance, rather than relying only
+on tests of an isolated counting helper.
+
+### 4. Collection-Policy Preparation And Application
+
+Selected entries: **E01, E02, E03** in
+[collection policy and projection](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#e-collection-processing-and-efficiency).
+
+Distinguish the existing policy state used for lookup, changes introduced by an
+incoming tier, and evidence retained by the assembly. Prepare incoming claims and
+declarations against read-only lower state, then apply the owned changes at the
+appropriate combination boundary. Avoid copying every unrelated policy for
+small tiers and scanning all policies for each independent replacement.
+
+Use a semantic-path index, localized updates, or equivalent bounded lookup to
+find affected policy. Preserve retirement timing, policy established by the
+incoming contribution, pending declarations, persistent explicit modes, and
+nested record ownership. Copying remains appropriate wherever mutable ownership
+requires it. Removing only one duplicate clone or shortening a function does not
+address the identified growth patterns.
+
+Clarify preparation, evidence application, and finalization responsibilities.
+Share straightforward history-descriptor translation where useful; evaluate
+shared record-selection logic while keeping overlay and inheritance origin and
+mode lifetimes distinct. Retain the common sequence-combination owner. This work
+does not call for a generic merging framework.
+
+### 5. Active Acquisition Handoffs And Private Cleanup
+
+Selected entries: **F01, F02, F03, F04, F05, F06** in
+[acquisition, cleanup, and test quality](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#f-remaining-maintenance-and-test-improvements).
+
+Remove the test-only eager acquisition mode so acquisition retains scopes through
+the active production protocol. Tests that need composition should invoke the
+actual composition boundary explicitly. Retire the parallel test-only import
+parser and route relevant fixtures through the active directive parser or the
+appropriate declaration boundary, retaining malformed-input coverage.
+
+Clarify active acquisition and coordinator handoffs around cohesive tasks:
+parsed-input directive/version handling, standalone placement, and repeated tier
+application/projection. Use explicit inputs and results consistent with existing
+state owners. Extracting helpers that continue to share surrounding mutable state
+does not establish a clearer handoff. Preserve ordinary/layered differences,
+overlay and diff behavior, source identity, callback timing, diagnostics, and
+accounting scopes.
+
+Remove the disconnected root record-clone family, uncalled engine helpers, and
+unread private transport/result members identified in the audit. Reconfirm
+references on the implementation baseline, including function values and relevant
+build configurations. Preserve live conversion paths and validation side effects;
+the audit's absence of field reads is not a general rule for deleting state.
+Replace meaningful ownership coverage through the public boundary before removing
+tests that only exercise disconnected helpers.
+
+### 6. Local Lookup Improvements
+
+Selected entries: **E04, E05** in
+[collection policy and projection](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#e-collection-processing-and-efficiency)
+and **F08** in [secret-failure lookup](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#f08-index-secret-failure-location-queries-lazily).
+
+Reuse a local document lookup or equivalent traversal for tier-history projection
+instead of rescanning wide mappings for each sequence. Preserve authored
+coordinates, sparse projections, and evidence ownership.
+
+Restore compound identity fields using indexed mapping positions and membership
+checks rather than rescanning child and merged mappings for every field. Preserve
+child-authored values and locations, mapping order, and detached nodes.
+
+Build secret-failure location lookup state lazily when failures need it. Preserve
+all diagnostics, nearest-location fallbacks, and confidentiality; successful
+loads should not incur an unnecessary index merely to support failure reporting.
+These changes address specific repeated work and do not promise a measured
+startup improvement.
+
+### 7. Tests That Protect The Intended Behavior
+
+Selected entries: **F09, F10, F11, F12** in
+[acquisition, cleanup, and test quality](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/453da684d7c7d3843a5f505f5c468b4c889dc6e2/docs/roadmap/audit.md#f-remaining-maintenance-and-test-improvements),
+in addition to the regression coverage attached to each reported defect.
+
+Cover historical-path enumeration through the public API, including removed
+paths, queryability, ordering, and detached returned slices. Use those assertions
+to replace private representation checks where they address the same risk.
+
+Strengthen fuzz detachment checks by mutating existing descendants or in-length
+elements, so spare capacity cannot hide aliases. Replace exact filesystem stat
+call sequences with observable candidate ordering and meaningful failure
+selection, retaining relevant assertions such as not opening excluded inputs.
+
+Consolidate facade scaffolding around public construction, loading, errors,
+ownership, and lifecycle behavior. Preserve distinct integration risks and
+reader-byte snapshot coverage. Existing coverage percentages or a desired test
+count are not acceptance criteria for this work.
+
+## Implementation Plan
+
+The historical implementation plan at `bb5540b:docs/roadmap/implementation.md`
+defines the sequential, Luna-scoped stages and their dependencies, including
+coordinated changes to descriptor state, history ownership/accounting, collection
+policy, acquisition, and replacement public coverage. It also records validation
+commands and the complete finding-to-stage map. The audit remains the owner of
+original evidence and detailed acceptance cases.
+
+## Validation And Completion
+
+Validation follows the audit's detailed acceptance cases and the
+[testing policy](../policy/testing.md). For fatal or hanging regressions, use bounded
+subprocesses; for budget cases, use small controlled limits. For symlink expansion
+and repeated lookup work, prefer bounded operation evidence and algorithm review
+over timing thresholds. Use real filesystem fixtures where filesystem semantics
+matter and narrow fault injection for read/close failures. State platform limits.
+
+Applicable development checks come from the [development guide](../development.md),
+including appropriate race coverage for changed ownership and lifetime boundaries.
+The historical implementation plan owns their execution. Existing audit results
+are baseline evidence, not validation of the eventual changes. No performance
+benchmark campaign is required by this roadmap, and no new benchmark results are
+claimed.
+
+Completion means all 29 selected findings have implementation and acceptance
+evidence, the affected public behavior remains consistent with its contracts,
+and canonical internal documentation describes the resulting owners. The
+historical implementation plan records execution status without duplicate
+progress inventories. Retire completed planning records under the normal
+documentation lifecycle.
+
+## Scope Boundaries And Decision Status
+
+All three audit scope questions are settled. No significant architectural or
+product decision remains open for this roadmap. The historical implementation
+plan records private design choices and sequencing, subject to the acceptance
+cases above.
+If that work exposes a need to change a public contract, bring the specific
+decision back for review rather than treating it as an implicit part of remediation.
+
+The glob preparation refactor (C03) and wider facade transport simplification
+(F07) have since been selected in the
+[internal-boundary roadmap](internal-boundaries.md). Other deferred features,
+clone-traversal redesign, and broader performance experiments remain outside
+this effort. Keep
+existing ownership boundaries and loading APIs; no ADR or architecture-policy
+change is proposed by this roadmap.

example_test.go

@@ -19,6 +19,28 @@ func ExampleLoad() {
 	// Output: {true 3} <nil>
 }
 
+func ExampleLoad_sequenceDeclaration() {
+	type Item struct {
+		ID string `yaml:"id"`
+	}
+	type Config struct {
+		Items []Item `yaml:"items"`
+	}
+	result, err := configkit.Load[Config](configkit.Bytes("catalog.yml", []byte(`configkit:
+  sequences:
+    - destination: [items]
+      key: id
+items:
+  - id: prompt
+`)))
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	fmt.Println(result.Config.Items[0].ID)
+	// Output: prompt
+}
+
 func ExampleNoSource() {
 	type Config struct {
 		Workers int `yaml:"workers"`
@@ -30,6 +52,32 @@ func ExampleNoSource() {
 	// Output: 4 0 <nil>
 }
 
+func ExampleDiff() {
+	type Config struct {
+		Workers int `yaml:"workers"`
+	}
+	before, err := configkit.Load[Config](configkit.Mapping("before", map[string]any{"workers": 2}))
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	after, err := configkit.Load[Config](configkit.Mapping("after", map[string]any{"workers": 4}))
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	differences, err := configkit.Diff(before, after)
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	for _, difference := range differences {
+		fmt.Printf("%s: %s -> %s\n", difference.Path.String(),
+			strings.TrimSpace(difference.Left.YAML), strings.TrimSpace(difference.Right.YAML))
+	}
+	// Output: /workers: 2 -> 4
+}
+
 func ExampleValidation_Error() {
 	type Config struct {
 		First  string `yaml:"first"`
@@ -128,8 +176,15 @@ func ExampleOverride() {
 		})
 	}
 	result, err := configkit.Load(configkit.NoSource(), configkit.Options[Config]{
-		Definition: configkit.Definition[Config]{Defaults: func() Config { return Config{Workers: 4} }},
-		Overrides:  overrides,
+		Definition: configkit.Definition[Config]{
+			Defaults: func() Config { return Config{Workers: 4} },
+			Resolve: func(configuration *Config, resolution *configkit.Resolution) error {
+				configkit.Default(resolution, configkit.Keys("workers"), &configuration.Workers, 100,
+					"preserve an explicitly supplied worker count")
+				return nil
+			},
+		},
+		Overrides: overrides,
 	})
 	fmt.Println(result.Config.Workers, result.Sources()[0].Label, err)
 	// Output: 0 --workers <nil>

examples/consumers/destinations.go

@@ -0,0 +1,154 @@
+package consumers
+
+import (
+	"errors"
+	"fmt"
+	"io/fs"
+	"path"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+// CollectionBundleConfig is a representative application-owned collection
+// shape. Collection and prompt version labels remain ordinary domain fields.
+type CollectionBundleConfig struct {
+	Version     string                          `yaml:"version"`
+	Collections map[string]CollectionDefinition `yaml:"collections"`
+}
+
+type CollectionDefinition struct {
+	Name        string                       `yaml:"name"`
+	Version     string                       `yaml:"version"`
+	Description string                       `yaml:"description"`
+	Prompts     map[string]CollectionPrompt  `yaml:"prompts"`
+	Profiles    map[string]CollectionProfile `yaml:"profiles"`
+}
+
+type CollectionPrompt struct {
+	Version        string              `yaml:"version"`
+	DefaultProfile string              `yaml:"default_profile"`
+	ContentFile    string              `yaml:"content_file"`
+	ResourceText   string              `yaml:"resource_text"`
+	Messages       []CollectionMessage `yaml:"messages"`
+}
+
+type CollectionMessage struct {
+	Role    string `yaml:"role"`
+	Content string `yaml:"content"`
+}
+
+type CollectionProfile struct {
+	Format      string  `yaml:"format"`
+	Temperature float64 `yaml:"temperature"`
+}
+
+// LoadCollectionBundle demonstrates one layered collection load. Its main
+// document owns the application schema version; the mounted entry declares
+// relative destinations for collection metadata, prompt/profile peers, and
+// their nested imports. The operator version override remains a separate tier.
+func LoadCollectionBundle() (*configkit.Result[CollectionBundleConfig], error) {
+	files, err := fs.Sub(bundledFiles, "testdata/collection-bundle")
+	if err != nil {
+		return nil, err
+	}
+	bundle := configkit.NewFilesystem("collection bundle", files)
+	collectionDestination := configkit.Keys("collections", "weatherreporter")
+	promptDestination := collectionDestination.Key("prompts")
+	return configkit.LoadLayered(
+		configkit.Input{Source: configkit.Bytes("application main", []byte("version: 1\n"))},
+		configkit.LayeredOptions[CollectionBundleConfig]{
+			Definition: configkit.Definition[CollectionBundleConfig]{
+				Version: &configkit.VersionRule{Path: configkit.Keys("version"), Check: func(version configkit.Version, _ *configkit.Validation) error {
+					if !version.Present || version.Text != "1" {
+						return errors.New("unsupported application schema version")
+					}
+					return nil
+				}},
+				Resolve:  resolveCollectionResources,
+				Validate: validateCollectionBundle,
+			},
+			Underlays: []configkit.Underlay{
+				{Inputs: []configkit.Input{{
+					Source: bundle.File("collections/weather/collection.yml"), Imports: true, Destination: collectionDestination,
+				}}},
+				{Inputs: []configkit.Input{{
+					Source: bundle.File("operators/weather-prompt-version.yml"), Destination: promptDestination,
+				}}},
+			},
+		},
+	)
+}
+
+func resolveCollectionResources(configuration *CollectionBundleConfig, resolution *configkit.Resolution) error {
+	for collectionID, collection := range configuration.Collections {
+		for promptID, prompt := range collection.Prompts {
+			if prompt.ContentFile == "" {
+				continue
+			}
+			contentPath := configkit.Keys("collections", collectionID, "prompts", promptID, "content_file")
+			info, found := resolution.InputInfo(contentPath)
+			if !found || len(info.Origin.Locations) != 1 {
+				return fmt.Errorf("prompt resource has no single declaring source")
+			}
+			resourceSource, found := resolution.Source(info.Origin.Locations[0].SourceID)
+			if !found {
+				return fmt.Errorf("prompt resource source is unavailable")
+			}
+			context, found := resolution.SourceContext(resourceSource.ID)
+			if !found {
+				return fmt.Errorf("prompt resource context is unavailable")
+			}
+			data, err := fs.ReadFile(context.FS, path.Join(context.BaseDir, prompt.ContentFile))
+			if err != nil {
+				return fmt.Errorf("read prompt resource: %w", err)
+			}
+			prompt.ResourceText = string(data)
+			collection.Prompts[promptID] = prompt
+			resolution.Record(
+				configkit.Keys("collections", collectionID, "prompts", promptID, "resource_text"),
+				configkit.ChangeNormalization, "read source-relative prompt content", contentPath,
+			)
+		}
+		configuration.Collections[collectionID] = collection
+	}
+	return nil
+}
+
+func validateCollectionBundle(configuration *CollectionBundleConfig, validation *configkit.Validation) {
+	for collectionID, collection := range configuration.Collections {
+		for promptID, prompt := range collection.Prompts {
+			if prompt.DefaultProfile == "" {
+				continue
+			}
+			if _, found := collection.Profiles[prompt.DefaultProfile]; !found {
+				validation.Error(configkit.Keys("collections", collectionID, "prompts", promptID, "default_profile"), "must name a profile in this collection")
+			}
+		}
+	}
+}
+
+// ReadCollectionPromptContent follows the effective content field's physical
+// source, including when it was inherited from a peer prompt.
+func ReadCollectionPromptContent(result *configkit.Result[CollectionBundleConfig], collectionID, promptID string) (string, error) {
+	if result == nil {
+		return "", errors.New("collection result is nil")
+	}
+	contentPath := configkit.Keys("collections", collectionID, "prompts", promptID, "content_file")
+	info, found := result.Metadata().Lookup(contentPath)
+	if !found || len(info.Origin.Locations) != 1 {
+		return "", errors.New("prompt resource has no single declaring source")
+	}
+	context, found := result.SourceContext(info.Origin.Locations[0].SourceID)
+	if !found {
+		return "", errors.New("prompt resource context is unavailable")
+	}
+	prompt, found := result.Config.Collections[collectionID].Prompts[promptID]
+	if !found {
+		return "", errors.New("prompt is unavailable")
+	}
+	data, err := fs.ReadFile(context.FS, path.Join(context.BaseDir, prompt.ContentFile))
+	if err != nil {
+		return "", err
+	}
+	return string(data), nil
+}

examples/consumers/destinations_test.go

@@ -0,0 +1,137 @@
+package consumers_test
+
+import (
+	"errors"
+	"fmt"
+	"testing"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+	"gitea.maximumdirect.net/eric/ConfigKit/examples/consumers"
+)
+
+func Example_inputDestination() {
+	mounted, err := configkit.LoadLayered[map[string]any](configkit.Input{
+		Source: configkit.Mapping("weather prompt", map[string]any{
+			"summarize.alerts": map[string]any{"version": "1"},
+		}),
+		Destination: configkit.Keys("collections", "weather.reporter", "prompts"),
+	})
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	root, err := configkit.LoadLayered[map[string]any](configkit.Input{
+		Source: configkit.Bytes("standalone", []byte("name: standalone\n")),
+	})
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	_, found := mounted.Config["collections"]
+	fmt.Println(found, root.Config["name"])
+	// Output: true standalone
+}
+
+func TestCollectionBundleAssemblesMountedPeersAndResources(t *testing.T) {
+	result, err := consumers.LoadCollectionBundle()
+	if err != nil {
+		t.Fatalf("LoadCollectionBundle: %v", err)
+	}
+	collection := result.Config.Collections["weatherreporter"]
+	if result.Config.Version != "1" || collection.Name != "Weather Reporter" || collection.Version != "2.4.0" {
+		t.Fatalf("main and collection versions = %q, %#v", result.Config.Version, collection)
+	}
+	brief := collection.Prompts["brief"]
+	custom := collection.Prompts["custom"]
+	if brief.Version != "2" || brief.DefaultProfile != "compact" || brief.ContentFile != "welcome.txt" || len(brief.Messages) != 1 {
+		t.Fatalf("version-only override replaced lower prompt fields: %#v", brief)
+	}
+	if custom.Version != "2" || custom.DefaultProfile != "weather" || custom.ContentFile != brief.ContentFile ||
+		len(custom.Messages) != 1 || custom.ResourceText != "Welcome to the weather collection.\n" {
+		t.Fatalf("inherited prompt = %#v", custom)
+	}
+	weatherProfile := collection.Profiles["weather"]
+	if weatherProfile.Format != "text" || weatherProfile.Temperature != 0.4 {
+		t.Fatalf("inherited profile = %#v", weatherProfile)
+	}
+	resource, err := consumers.ReadCollectionPromptContent(result, "weatherreporter", "custom")
+	if err != nil || resource != custom.ResourceText {
+		t.Fatalf("inherited prompt resource = %q, %v", resource, err)
+	}
+
+	versionInfo, ok := result.Metadata().Lookup(configkit.Keys("version"))
+	if !ok || !versionInfo.Explicit || versionInfo.Origin.Layer != configkit.LayerBase || len(versionInfo.Origin.Locations) != 1 {
+		t.Fatalf("main-owned version metadata = %#v, %v", versionInfo, ok)
+	}
+	collectionVersionPath := configkit.Keys("collections", "weatherreporter", "version")
+	collectionVersion, ok := result.Metadata().Lookup(collectionVersionPath)
+	if !ok || !collectionVersion.Explicit || collectionVersion.Origin.Layer != configkit.LayerUnderlay || len(collectionVersion.Origin.Locations) != 1 {
+		t.Fatalf("ordinary collection version metadata = %#v, %v", collectionVersion, ok)
+	}
+	var metadataSource configkit.SourceInfo
+	var collectionEntrySource configkit.SourceInfo
+	var versionSource configkit.SourceInfo
+	var parentPromptSource configkit.SourceInfo
+	var promptIndexSource configkit.SourceInfo
+	var profileIndexSource configkit.SourceInfo
+	var promptVersionSource configkit.SourceInfo
+	for _, source := range result.Sources() {
+		switch source.Path {
+		case "collections/weather/collection.yml":
+			collectionEntrySource = source
+		case "collections/weather/metadata.yml":
+			metadataSource = source
+		case "collections/weather/version.yml":
+			versionSource = source
+		case "collections/weather/prompts/index.yml":
+			promptIndexSource = source
+		case "collections/weather/prompts/brief.yml":
+			parentPromptSource = source
+		case "collections/weather/profiles/index.yml":
+			profileIndexSource = source
+		case "operators/weather-prompt-version.yml":
+			promptVersionSource = source
+		}
+	}
+	collectionVersionLocation := collectionVersion.Origin.Locations[0]
+	if collectionEntrySource.ID != "source-2" || collectionEntrySource.Destination.String() != "/collections/weatherreporter" ||
+		metadataSource.ID == "" || metadataSource.Destination.String() != "/collections/weatherreporter" ||
+		metadataSource.ImportedFrom == nil || metadataSource.ImportedFrom.SourceID != collectionEntrySource.ID ||
+		versionSource.ID == "" || collectionVersionLocation.SourceID != versionSource.ID || versionSource.Destination.String() != "/collections/weatherreporter" ||
+		versionSource.ImportedFrom == nil || versionSource.ImportedFrom.SourceID != metadataSource.ID || collectionVersionLocation.Line != 1 {
+		t.Fatalf("nested collection version provenance = entry %#v, metadata %#v, version %#v, location %#v", collectionEntrySource, metadataSource, versionSource, collectionVersionLocation)
+	}
+	if promptIndexSource.ID == "" || promptIndexSource.Destination.String() != "/collections/weatherreporter/prompts" ||
+		promptIndexSource.ImportedFrom == nil || promptIndexSource.ImportedFrom.SourceID != collectionEntrySource.ID ||
+		profileIndexSource.ID == "" || profileIndexSource.Destination.String() != "/collections/weatherreporter/profiles" ||
+		profileIndexSource.ImportedFrom == nil || profileIndexSource.ImportedFrom.SourceID != collectionEntrySource.ID {
+		t.Fatalf("nested prompt/profile placement = prompt %#v profile %#v", promptIndexSource, profileIndexSource)
+	}
+	if promptVersionSource.ID == "" || promptVersionSource.Destination.String() != "/collections/weatherreporter/prompts" || promptVersionSource.UnderlayIndex != 2 {
+		t.Fatalf("prompt version source = %#v", promptVersionSource)
+	}
+	contentPath := configkit.Keys("collections", "weatherreporter", "prompts", "custom", "content_file")
+	contentInfo, ok := result.Metadata().Lookup(contentPath)
+	if !ok || !contentInfo.Explicit || len(contentInfo.Origin.Inheritance) != 1 || len(contentInfo.Origin.Locations) != 1 {
+		t.Fatalf("inherited content field metadata = %#v, %v", contentInfo, ok)
+	}
+	contentLocation := contentInfo.Origin.Locations[0]
+	if parentPromptSource.ID == "" || contentLocation.SourceID != parentPromptSource.ID || contentLocation.Line != 4 ||
+		parentPromptSource.Destination.String() != "/collections/weatherreporter/prompts" {
+		t.Fatalf("inherited resource source coordinates = %#v, source %#v", contentLocation, parentPromptSource)
+	}
+}
+
+func TestCollectionPeerPromptConflictsRemainStrict(t *testing.T) {
+	destination := configkit.Keys("collections", "weatherreporter", "prompts")
+	_, err := configkit.LoadLayered[map[string]any](configkit.Input{Source: configkit.NoSource()}, configkit.LayeredOptions[map[string]any]{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{
+			{Source: configkit.Bytes("peer-a", []byte("brief: {version: '1'}\n")), Destination: destination},
+			{Source: configkit.Bytes("peer-b", []byte("brief: {version: '1'}\n")), Destination: destination},
+		}}},
+	})
+	var loadError *configkit.LoadError
+	if !errors.As(err, &loadError) || len(loadError.Diagnostics()) == 0 || loadError.Diagnostics()[0].Code != configkit.CodeComposeConflict {
+		t.Fatalf("duplicate peer prompt error = %v", err)
+	}
+}

examples/consumers/diff.go

@@ -0,0 +1,136 @@
+package consumers
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+// FileComparisonConfig is a small application-owned shape for comparing
+// caller-selected configuration files.
+type FileComparisonConfig struct {
+	Name    string         `yaml:"name"`
+	Entries map[string]any `yaml:"entries"`
+}
+
+// CompareFiles loads two caller-selected files with the same application
+// definition, then formats only the safe records returned by configkit.Diff.
+func CompareFiles(leftPath, rightPath string) (string, error) {
+	options := configkit.Options[FileComparisonConfig]{
+		Definition: configkit.Definition[FileComparisonConfig]{
+			Defaults: func() FileComparisonConfig {
+				return FileComparisonConfig{Entries: map[string]any{}}
+			},
+		},
+	}
+	left, err := configkit.Load(configkit.File(leftPath), options)
+	if err != nil {
+		return "", fmt.Errorf("load left configuration: %w", err)
+	}
+	right, err := configkit.Load(configkit.File(rightPath), options)
+	if err != nil {
+		return "", fmt.Errorf("load right configuration: %w", err)
+	}
+	differences, err := configkit.Diff(left, right)
+	if err != nil {
+		return "", err
+	}
+	return FormatDifferences(differences), nil
+}
+
+// RolloutConfig models a service endpoint whose URL is derived by the
+// application resolver and whose credential reference is classified.
+type RolloutConfig struct {
+	Service RolloutService `yaml:"service"`
+}
+
+// RolloutService contains endpoint settings and a classified credential.
+type RolloutService struct {
+	Endpoint   RolloutEndpoint   `yaml:"endpoint"`
+	Credential RolloutCredential `yaml:"credential"`
+}
+
+// RolloutEndpoint is the effective destination used by the application.
+type RolloutEndpoint struct {
+	Host string `yaml:"host"`
+	Port int    `yaml:"port"`
+	URL  string `yaml:"url"`
+}
+
+// RolloutCredential holds only a typed secret reference in configuration.
+type RolloutCredential struct {
+	Token configkit.Secret `yaml:"token" configkit:"secret"`
+}
+
+// LoadRollout loads a base file, optionally with one selected configuration
+// overlay. Resolution derives the effective URL after the overlay is applied.
+func LoadRollout(basePath, overlayName string) (*configkit.Result[RolloutConfig], error) {
+	return configkit.Load(configkit.File(basePath), configkit.Options[RolloutConfig]{
+		Overlay: overlayName,
+		Definition: configkit.Definition[RolloutConfig]{
+			Resolve: resolveRollout,
+		},
+	})
+}
+
+// CompareRolloutOverlay compares the effective base configuration with that
+// same file plus one selected overlay. It presents only safe diff records.
+func CompareRolloutOverlay(basePath, overlayName string) (string, error) {
+	base, err := LoadRollout(basePath, "")
+	if err != nil {
+		return "", fmt.Errorf("load base configuration: %w", err)
+	}
+	withOverlay, err := LoadRollout(basePath, overlayName)
+	if err != nil {
+		return "", fmt.Errorf("load overlay configuration: %w", err)
+	}
+	differences, err := configkit.Diff(base, withOverlay)
+	if err != nil {
+		return "", err
+	}
+	return FormatDifferences(differences), nil
+}
+
+func resolveRollout(configuration *RolloutConfig, resolution *configkit.Resolution) error {
+	endpoint := &configuration.Service.Endpoint
+	if endpoint.Host == "" || endpoint.Port <= 0 {
+		return nil
+	}
+	url := fmt.Sprintf("https://%s:%d", endpoint.Host, endpoint.Port)
+	if endpoint.URL != url {
+		endpoint.URL = url
+		resolution.Record(configkit.Keys("service", "endpoint", "url"),
+			configkit.ChangeNormalization, "derived endpoint URL",
+			configkit.Keys("service", "endpoint", "host"),
+			configkit.Keys("service", "endpoint", "port"))
+	}
+	return nil
+}
+
+// FormatDifferences is consumer-owned presentation. It distinguishes absence
+// through Present and reports withheld changes through HiddenChanged without
+// interpreting redaction marker text.
+func FormatDifferences(differences []configkit.Difference) string {
+	if len(differences) == 0 {
+		return "No effective changes.\n"
+	}
+	var output strings.Builder
+	for _, difference := range differences {
+		fmt.Fprintf(&output, "%s: left=%s right=%s", difference.Path.String(),
+			formatSnapshot(difference.Left), formatSnapshot(difference.Right))
+		if difference.HiddenChanged {
+			output.WriteString(" hidden-change")
+		}
+		output.WriteByte('\n')
+	}
+	return output.String()
+}
+
+func formatSnapshot(snapshot configkit.ValueSnapshot) string {
+	if !snapshot.Present {
+		return "<absent>"
+	}
+	return strconv.Quote(snapshot.YAML)
+}

examples/consumers/diff_test.go

@@ -0,0 +1,187 @@
+package consumers_test
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+	"gitea.maximumdirect.net/eric/ConfigKit/examples/consumers"
+)
+
+func ExampleCompareFiles() {
+	directory, err := os.MkdirTemp("", "configkit-diff-files-")
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	defer os.RemoveAll(directory)
+	leftPath := filepath.Join(directory, "before.yml")
+	rightPath := filepath.Join(directory, "after.yml")
+	if err := os.WriteFile(leftPath, []byte("name: before\nentries: {}\n"), 0o600); err != nil {
+		fmt.Println(err)
+		return
+	}
+	if err := os.WriteFile(rightPath, []byte("name: after\nentries:\n  added: null\n"), 0o600); err != nil {
+		fmt.Println(err)
+		return
+	}
+	output, err := consumers.CompareFiles(leftPath, rightPath)
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	fmt.Print(output)
+	// Output:
+	// /entries/added: left=<absent> right="null\n"
+	// /name: left="before\n" right="after\n"
+}
+
+func ExampleCompareRolloutOverlay() {
+	const baseToken = "CONFIGKIT_DIFF_EXAMPLE_BASE_TOKEN"
+	const overlayToken = "CONFIGKIT_DIFF_EXAMPLE_OVERLAY_TOKEN"
+	defer preserveEnvironment(baseToken)()
+	defer preserveEnvironment(overlayToken)()
+	_ = os.Setenv(baseToken, "disposable-example-base-credential")
+	_ = os.Setenv(overlayToken, "disposable-example-overlay-credential")
+
+	directory, err := os.MkdirTemp("", "configkit-diff-overlay-")
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	defer os.RemoveAll(directory)
+	basePath := filepath.Join(directory, "config.yml")
+	overlayPath := filepath.Join(directory, "overlays", "canary.yml")
+	if err := os.MkdirAll(filepath.Dir(overlayPath), 0o700); err != nil {
+		fmt.Println(err)
+		return
+	}
+	base := "service:\n  endpoint:\n    host: base.example\n    port: 8443\n  credential:\n    token: " + baseToken + "\n"
+	overlay := "service:\n  endpoint:\n    host: canary.example\n  credential:\n    token: " + overlayToken + "\n"
+	if err := os.WriteFile(basePath, []byte(base), 0o600); err != nil {
+		fmt.Println(err)
+		return
+	}
+	if err := os.WriteFile(overlayPath, []byte(overlay), 0o600); err != nil {
+		fmt.Println(err)
+		return
+	}
+	output, err := consumers.CompareRolloutOverlay(basePath, "canary")
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	fmt.Print(output)
+	// Output:
+	// /service/credential/token: left="'[REDACTED]'\n" right="'[REDACTED]'\n" hidden-change
+	// /service/endpoint/host: left="base.example\n" right="canary.example\n"
+	// /service/endpoint/url: left="https://base.example:8443\n" right="https://canary.example:8443\n"
+}
+
+func TestCompareFilesUsesCallerSelectedSourcesAndSafePresence(t *testing.T) {
+	directory := t.TempDir()
+	leftPath := filepath.Join(directory, "before.yml")
+	rightPath := filepath.Join(directory, "after.yml")
+	writeConfiguration(t, leftPath, "name: before\nentries: {}\n")
+	writeConfiguration(t, rightPath, "name: after\nentries:\n  added: null\n")
+
+	output, err := consumers.CompareFiles(leftPath, rightPath)
+	if err != nil {
+		t.Fatal(err)
+	}
+	want := "/entries/added: left=<absent> right=\"null\\n\"\n" +
+		"/name: left=\"before\\n\" right=\"after\\n\"\n"
+	if output != want {
+		t.Fatalf("comparison output = %q, want %q", output, want)
+	}
+	if again, err := consumers.CompareFiles(leftPath, rightPath); err != nil || again != output {
+		t.Fatalf("repeat output = %q, %v", again, err)
+	}
+}
+
+func TestCompareFilesReturnsEachLoadError(t *testing.T) {
+	directory := t.TempDir()
+	leftPath := filepath.Join(directory, "before.yml")
+	rightPath := filepath.Join(directory, "after.yml")
+	writeConfiguration(t, leftPath, "name: before\n")
+	writeConfiguration(t, rightPath, "unknown: invalid\n")
+
+	output, err := consumers.CompareFiles(leftPath, rightPath)
+	if err == nil || output != "" || !strings.Contains(err.Error(), "load right configuration") {
+		t.Fatalf("right load failure = %q, %v", output, err)
+	}
+	if _, err := consumers.CompareFiles(filepath.Join(directory, "missing.yml"), rightPath); err == nil || !strings.Contains(err.Error(), "load left configuration") {
+		t.Fatalf("left load failure = %v", err)
+	}
+}
+
+func TestCompareRolloutOverlayShowsResolvedSafeChangesAndRetainsSiblings(t *testing.T) {
+	t.Setenv("CONFIGKIT_DIFF_BASE_TOKEN", "disposable-base-credential")
+	t.Setenv("CONFIGKIT_DIFF_OVERLAY_TOKEN", "disposable-overlay-credential")
+	directory := t.TempDir()
+	basePath := filepath.Join(directory, "config.yml")
+	overlayPath := filepath.Join(directory, "overlays", "canary.yml")
+	if err := os.MkdirAll(filepath.Dir(overlayPath), 0o700); err != nil {
+		t.Fatal(err)
+	}
+	writeConfiguration(t, basePath, "service:\n  endpoint:\n    host: base.example\n    port: 8443\n  credential:\n    token: CONFIGKIT_DIFF_BASE_TOKEN\n")
+	writeConfiguration(t, overlayPath, "service:\n  endpoint:\n    host: canary.example\n  credential:\n    token: CONFIGKIT_DIFF_OVERLAY_TOKEN\n")
+
+	base, err := consumers.LoadRollout(basePath, "")
+	if err != nil {
+		t.Fatal(err)
+	}
+	withOverlay, err := consumers.LoadRollout(basePath, "canary")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if base.Config.Service.Endpoint.Port != 8443 || withOverlay.Config.Service.Endpoint.Port != 8443 {
+		t.Fatalf("lower-priority port was not retained: base=%#v overlay=%#v", base.Config.Service.Endpoint, withOverlay.Config.Service.Endpoint)
+	}
+	urlInfo, found := withOverlay.Metadata().Lookup(configkit.Keys("service", "endpoint", "url"))
+	if !found || urlInfo.Origin.Layer != configkit.LayerResolution || withOverlay.Config.Service.Endpoint.URL != "https://canary.example:8443" {
+		t.Fatalf("resolved endpoint = %#v, metadata=%#v, %v", withOverlay.Config.Service.Endpoint, urlInfo, found)
+	}
+
+	differences, err := configkit.Diff(base, withOverlay)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(differences) != 3 {
+		t.Fatalf("differences = %#v", differences)
+	}
+	wantPaths := []string{"/service/credential/token", "/service/endpoint/host", "/service/endpoint/url"}
+	for index, difference := range differences {
+		if got := difference.Path.String(); got != wantPaths[index] {
+			t.Errorf("path[%d] = %q, want %q", index, got, wantPaths[index])
+		}
+	}
+	secret := differences[0]
+	if !secret.Left.Redacted || !secret.Right.Redacted || !secret.HiddenChanged || secret.Left.YAML != "'[REDACTED]'\n" || secret.Right.YAML != "'[REDACTED]'\n" {
+		t.Fatalf("classified credential difference = %#v", secret)
+	}
+	if strings.Contains(consumers.FormatDifferences(differences), "disposable-") {
+		t.Fatal("formatted differences exposed credential contents")
+	}
+
+	output, err := consumers.CompareRolloutOverlay(basePath, "canary")
+	if err != nil || output != consumers.FormatDifferences(differences) {
+		t.Fatalf("formatted base/overlay comparison = %q, %v", output, err)
+	}
+}
+
+func TestFormatDifferencesUsesRecordMembershipForNoChange(t *testing.T) {
+	if got := consumers.FormatDifferences(nil); got != "No effective changes.\n" {
+		t.Fatalf("no-change output = %q", got)
+	}
+}
+
+func writeConfiguration(t *testing.T, filename, contents string) {
+	t.Helper()
+	if err := os.WriteFile(filename, []byte(contents), 0o600); err != nil {
+		t.Fatal(err)
+	}
+}

examples/consumers/discovery_test.go

@@ -0,0 +1,45 @@
+package consumers_test
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+type workingDirectoryPresetConfig struct {
+	Name string `yaml:"name"`
+}
+
+func Example_workingDirectoryPreset() {
+	original, err := os.Getwd()
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	defer func() { _ = os.Chdir(original) }()
+	root, err := os.MkdirTemp("", "configkit-working-search-")
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	defer os.RemoveAll(root)
+	if err := os.Chdir(root); err != nil {
+		fmt.Println(err)
+		return
+	}
+	if err := os.WriteFile(filepath.Join(root, "service.yml"), []byte("name: example\n"), 0o600); err != nil {
+		fmt.Println(err)
+		return
+	}
+	result, err := configkit.Load[workingDirectoryPresetConfig](
+		configkit.SearchConfig("service", configkit.WorkingDirectory()),
+	)
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	fmt.Println(result.Config.Name, filepath.Base(result.Sources()[0].Path))
+	// Output: example service.yml
+}

examples/consumers/layered.go

@@ -9,7 +9,7 @@ import (
 	configkit "gitea.maximumdirect.net/eric/ConfigKit"
 )
 
-//go:embed testdata/native-source testdata/layered-bundle
+//go:embed testdata/native-source testdata/layered-bundle testdata/collection-bundle testdata/ordinary-imports testdata/source-version
 var bundledFiles embed.FS
 
 // BundledConfig is a small configuration loaded directly from an embedded file.
@@ -18,6 +18,30 @@ type BundledConfig struct {
 	Enabled bool   `yaml:"enabled"`
 }
 
+type PromptBundleConfig struct {
+	Name        string                  `yaml:"name"`
+	Version     string                  `yaml:"version"`
+	Description string                  `yaml:"description"`
+	Prompts     map[string]BundlePrompt `yaml:"prompts"`
+}
+
+type BundlePrompt struct {
+	Summary string `yaml:"summary"`
+}
+
+// LoadPromptBundle demonstrates ordinary Load with opt-in imports whose
+// destinations are declared relative to the entry document.
+func LoadPromptBundle() (*configkit.Result[PromptBundleConfig], error) {
+	files, err := fs.Sub(bundledFiles, "testdata/ordinary-imports")
+	if err != nil {
+		return nil, err
+	}
+	return configkit.Load[PromptBundleConfig](
+		configkit.NewFilesystem("prompt bundle", files).File("config.yml"),
+		configkit.Options[PromptBundleConfig]{Imports: true},
+	)
+}
+
 // LoadBundled demonstrates a native filesystem source without layered options.
 func LoadBundled() (*configkit.Result[BundledConfig], error) {
 	files, err := fs.Sub(bundledFiles, "testdata/native-source")

examples/consumers/layered_test.go

@@ -20,6 +20,34 @@ func Example_nativeFilesystem() {
 	// Output: embedded false
 }
 
+func Example_rootRelativeImports() {
+	result, err := consumers.LoadPromptBundle()
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	source := result.Sources()[2]
+	fmt.Println(result.Config.Name, result.Config.Version, result.Config.Prompts["brief"].Summary, source.Destination.String(), err)
+	// Output: Root import example 1.0 A root-loaded prompt. /prompts <nil>
+}
+
+func TestPromptBundleUsesRelativeImportDestinations(t *testing.T) {
+	result, err := consumers.LoadPromptBundle()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Config.Name != "Root import example" || result.Config.Version != "1.0" || result.Config.Prompts["brief"].Summary != "A root-loaded prompt." {
+		t.Fatalf("root import result = %#v", result.Config)
+	}
+	sources := result.Sources()
+	if len(sources) != 3 || sources[0].Destination.String() != configkit.Keys().String() ||
+		sources[1].Path != "metadata.yml" || sources[1].Destination.String() != configkit.Keys().String() ||
+		sources[2].Path != "prompts/brief.yml" || sources[2].Destination.String() != "/prompts" ||
+		sources[2].ImportedFrom == nil || sources[2].ImportedFrom.SourceID != sources[0].ID {
+		t.Fatalf("root import sources = %#v", sources)
+	}
+}
+
 func Example_layeredCatalog() {
 	result, err := consumers.LoadCatalog(fixture("layered-main", "config.yml"))
 	if err != nil {
@@ -32,6 +60,58 @@ func Example_layeredCatalog() {
 	// Output: production false 0 bundle example Welcome from the bundle. <nil>
 }
 
+func Example_sourceVersionAdoption() {
+	result, err := consumers.LoadVersionedService(configkit.SourceVersionModeIfPresent)
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	checks, redacted := 0, true
+	for _, source := range result.Sources() {
+		if check := source.VersionCheck; check != nil {
+			checks++
+			redacted = redacted && check.Redacted && check.Version.Text == ""
+		}
+	}
+	required, requiredErr := consumers.LoadVersionedService(configkit.SourceVersionModeRequired)
+	fmt.Println(result.Config.Service.Version, result.Config.Service.Metadata.Release,
+		checks, redacted, required != nil && requiredErr == nil)
+	// Output: service-v2 application-2026.10 2 true true
+}
+
+func TestVersionedServiceModesAndMountedProtection(t *testing.T) {
+	for _, mode := range []configkit.SourceVersionMode{
+		configkit.SourceVersionModeIfPresent,
+		configkit.SourceVersionModeRequired,
+	} {
+		result, err := consumers.LoadVersionedService(mode)
+		if err != nil || result == nil || result.Config.Service.Version != "service-v2" ||
+			result.Config.Service.Name != "example service" || result.Config.Service.Endpoint != "https://service.example.invalid" ||
+			result.Config.Service.Metadata.Release != "application-2026.10" {
+			t.Fatalf("versioned service mode %d = %#v, %v", mode, result, err)
+		}
+		sources := result.Sources()
+		if len(sources) != 2 {
+			t.Fatalf("versioned service sources = %#v", sources)
+		}
+		for _, source := range sources {
+			check := source.VersionCheck
+			if check == nil || check.Path.String() != "/version" || !check.Version.Present ||
+				!check.Redacted || check.Version.Text != "" {
+				t.Fatalf("safe source version evidence = %#v", sources)
+			}
+		}
+	}
+	_, err := consumers.LoadVersionedService(configkit.SourceVersionModeIfPresent,
+		configkit.Override{Path: configkit.Keys("service", "version"), Value: "cli-v3", Flag: "--service-version"})
+	var loadError *configkit.LoadError
+	if !errors.As(err, &loadError) || len(loadError.Diagnostics()) != 1 ||
+		loadError.Diagnostics()[0].Code != configkit.CodeOverrideVersion ||
+		loadError.Diagnostics()[0].Path.String() != "/service/version" {
+		t.Fatalf("mounted main version override error = %v", err)
+	}
+}
+
 func TestCatalogRetainsPresenceAndSourceContext(t *testing.T) {
 	result, err := consumers.LoadCatalog(fixture("layered-main", "config.yml"))
 	if err != nil {

examples/consumers/prompt_catalog.go

@@ -0,0 +1,67 @@
+package consumers
+
+import (
+	"strings"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+// PromptCatalogConfig is an application-owned prompt collection shape.
+type PromptCatalogConfig struct {
+	Prompts []PromptRecord `yaml:"prompts"`
+}
+
+// PromptRecord keeps its identity and message meaning in the consumer.
+type PromptRecord struct {
+	ID       string   `yaml:"id"`
+	Name     string   `yaml:"name"`
+	Messages []string `yaml:"messages"`
+	Resource string   `yaml:"resource"`
+}
+
+// LoadPromptCatalog discovers explicitly registered prompt records in the main
+// tier and replaces the lower-priority default collection as a whole.
+func LoadPromptCatalog(mainPath string) (*configkit.Result[PromptCatalogConfig], error) {
+	return configkit.LoadLayered(configkit.Input{
+		Source:  configkit.File(mainPath),
+		Imports: true,
+	}, configkit.LayeredOptions[PromptCatalogConfig]{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{{
+			Source: configkit.Bytes("default prompts", []byte("prompts:\n  - id: default\n    name: Default prompt\n    messages: [Default message]\n")),
+		}}}},
+		ImportDefaults: configkit.ImportDefaults{
+			Pattern: []string{"curated*.yml", "generated-*.yml"},
+		},
+		Definition: promptCatalogDefinition(),
+	})
+}
+
+// LoadRequiredPromptCatalog keeps the common one-file import shape small.
+func LoadRequiredPromptCatalog(mainPath string) (*configkit.Result[PromptCatalogConfig], error) {
+	return configkit.Load(configkit.File(mainPath), configkit.Options[PromptCatalogConfig]{
+		Imports: true, Definition: promptCatalogDefinition(),
+	})
+}
+
+func promptCatalogDefinition() configkit.Definition[PromptCatalogConfig] {
+	return configkit.Definition[PromptCatalogConfig]{
+		SequenceIdentities: []configkit.SequenceIdentity{
+			configkit.SequenceIdentityAt(configkit.Keys("prompts"), "id"),
+		},
+		Validate: validatePromptCatalog,
+	}
+}
+
+func validatePromptCatalog(configuration *PromptCatalogConfig, validation *configkit.Validation) {
+	for index, prompt := range configuration.Prompts {
+		if strings.TrimSpace(prompt.Name) == "" {
+			validation.Error(configkit.Keys("prompts").Index(index).Key("name"), "must not be empty")
+		}
+		if len(prompt.Messages) == 0 {
+			validation.Error(configkit.Keys("prompts").Index(index).Key("messages"), "must contain at least one message")
+		}
+	}
+	if len(configuration.Prompts) == 0 {
+		validation.Error(configkit.Keys("prompts"), "must contain at least one prompt")
+	}
+}

examples/consumers/prompt_catalog_test.go

@@ -0,0 +1,135 @@
+package consumers_test
+
+import (
+	"errors"
+	"fmt"
+	"io/fs"
+	pathpkg "path"
+	"path/filepath"
+	"testing"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+	"gitea.maximumdirect.net/eric/ConfigKit/examples/consumers"
+)
+
+func Example_identifiedPromptCatalog() {
+	result, err := consumers.LoadPromptCatalog(fixture("prompt-catalog", "config.yml"))
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	ids := make([]string, 0, len(result.Config.Prompts))
+	for _, prompt := range result.Config.Prompts {
+		ids = append(ids, prompt.ID)
+	}
+	emptySearches := 0
+	for _, source := range result.Sources() {
+		for _, selection := range source.ImportSelections {
+			if selection.Selector == configkit.ImportSelectorDirectory && selection.Outcome == configkit.ImportSelectionNone {
+				emptySearches++
+			}
+		}
+	}
+	fmt.Println(ids, emptySearches, err)
+	// Output: [weather curated welcome main] 1 <nil>
+}
+
+func Example_requiredPromptCatalog() {
+	result, err := consumers.LoadRequiredPromptCatalog(fixture("prompt-catalog", "simple.yml"))
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	fmt.Println(result.Config.Prompts[0].ID, err)
+	// Output: required <nil>
+}
+
+func TestPromptCatalogImportsRegisteredItemsAndReplacesDefaults(t *testing.T) {
+	result, err := consumers.LoadPromptCatalog(fixture("prompt-catalog", "config.yml"))
+	if err != nil {
+		var loadError *configkit.LoadError
+		if errors.As(err, &loadError) {
+			t.Fatalf("load prompt catalog: %v (%#v)", err, loadError.Diagnostics())
+		}
+		t.Fatal(err)
+	}
+	if len(result.Config.Prompts) != 4 {
+		t.Fatalf("prompt records = %#v", result.Config.Prompts)
+	}
+	ids := []string{"weather", "curated", "welcome", "main"}
+	for index, id := range ids {
+		if result.Config.Prompts[index].ID != id {
+			t.Fatalf("prompt %d ID = %q, want %q", index, result.Config.Prompts[index].ID, id)
+		}
+	}
+	for _, prompt := range result.Config.Prompts {
+		if prompt.ID == "default" {
+			t.Fatalf("lower-priority default survived main sequence replacement: %#v", result.Config.Prompts)
+		}
+		if prompt.ID == "ignored" {
+			t.Fatalf("excluded generated prompt was selected: %#v", result.Config.Prompts)
+		}
+	}
+
+	var itemImports, helperImports int
+	for _, source := range result.Sources() {
+		switch filepath.Base(source.Path) {
+		case "curated.yml", "weather-feature.yml", "generated-welcome.yml":
+			if source.PlacementMode != configkit.PlacementModeSequenceItem || source.Destination.String() != "/prompts" {
+				t.Fatalf("item source placement = %#v", source)
+			}
+			itemImports++
+		case "curated-id.yml":
+			if source.PlacementMode != configkit.PlacementModeMapping || source.ImportedFrom == nil {
+				t.Fatalf("helper source = %#v", source)
+			}
+			helperImports++
+		}
+	}
+	if itemImports != 3 || helperImports != 1 {
+		t.Fatalf("item imports = %d, helper imports = %d, sources = %#v", itemImports, helperImports, result.Sources())
+	}
+	if result.Config.Prompts[1].ID != "curated" {
+		t.Fatalf("helper did not provide prompt ID: %#v", result.Config.Prompts[1])
+	}
+}
+
+func TestPromptCatalogMountsResourceFromScheduledPromptOrigin(t *testing.T) {
+	result, err := consumers.LoadPromptCatalog(fixture("prompt-catalog", "config.yml"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Config.Prompts[2].ID != "welcome" || result.Config.Prompts[2].Resource != "welcome.txt" {
+		t.Fatalf("scheduled resource prompt = %#v", result.Config.Prompts)
+	}
+	resourcePath := configkit.Keys("prompts").Index(2).Key("resource")
+	info, ok := result.Metadata().Lookup(resourcePath)
+	if !ok || len(info.Origin.Locations) != 1 {
+		t.Fatalf("resource origin = %#v, %v", info, ok)
+	}
+	context, ok := result.SourceContext(info.Origin.Locations[0].SourceID)
+	if !ok || context.FS == nil || filepath.Base(context.BaseDir) != "prompts" {
+		t.Fatalf("resource source context = %#v, %v", context, ok)
+	}
+	contents, err := fs.ReadFile(context.FS, pathpkg.Join(context.BaseDir, result.Config.Prompts[2].Resource))
+	if err != nil || string(contents) != "Welcome resource text.\n" {
+		t.Fatalf("mounted resource = %q, %v", contents, err)
+	}
+}
+
+func TestRequiredPromptCatalogUsesOneRequiredFile(t *testing.T) {
+	result, err := consumers.LoadRequiredPromptCatalog(fixture("prompt-catalog", "simple.yml"))
+	if err != nil || len(result.Config.Prompts) != 1 || result.Config.Prompts[0].ID != "required" {
+		t.Fatalf("simple required prompt = %#v, %v", result, err)
+	}
+}
+
+func TestPromptCatalogLeavesDomainValidationWithConsumer(t *testing.T) {
+	_, err := consumers.LoadPromptCatalog(fixture("prompt-catalog", "invalid.yml"))
+	var loadError *configkit.LoadError
+	if !errors.As(err, &loadError) || len(loadError.Diagnostics()) != 1 ||
+		loadError.Diagnostics()[0].Code != configkit.CodeValidateError ||
+		loadError.Diagnostics()[0].Path.String() != "/prompts/0/messages" {
+		t.Fatalf("consumer domain validation error = %v", err)
+	}
+}

examples/consumers/sequence_catalog.go

@@ -0,0 +1,145 @@
+package consumers
+
+import (
+	"errors"
+	"fmt"
+	"io/fs"
+	"os"
+	"path"
+	"path/filepath"
+	"strings"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+// SequenceCatalogConfig is a source-backed prompt catalog with an application
+// supplied run label and a value derived from that label during resolution.
+type SequenceCatalogConfig struct {
+	RunLabel      string                 `yaml:"run_label"`
+	ResolvedLabel string                 `yaml:"resolved_label"`
+	Prompts       []SequenceCatalogEntry `yaml:"prompts"`
+}
+
+// SequenceCatalogEntry keeps prompt meaning, resources, and credentials in the
+// consuming application while ConfigKit handles collection assembly.
+type SequenceCatalogEntry struct {
+	ID          string           `yaml:"id"`
+	Name        string           `yaml:"name"`
+	Description string           `yaml:"description"`
+	Messages    []string         `yaml:"messages"`
+	Resource    string           `yaml:"resource"`
+	Credential  configkit.Secret `yaml:"credential"`
+}
+
+// LoadSequenceCatalog searches its caller-selected main path before a custom
+// fallback directory, assembles imported fragments over a file-backed underlay,
+// applies one selected collection overlay, then resolves and validates once.
+func LoadSequenceCatalog(mainPath, overlayName, runLabel string) (*configkit.Result[SequenceCatalogConfig], error) {
+	absoluteMain, err := filepath.Abs(mainPath)
+	if err != nil {
+		return nil, err
+	}
+	root := filepath.Dir(absoluteMain)
+	filesystem := configkit.NewFilesystem("sequence catalog", os.DirFS(root))
+	filesystemContext := &configkit.FilesystemContext{BaseDir: ".", Root: ".", FS: &filesystem}
+	mainSearch := configkit.SearchConfig("", configkit.Searches(
+		configkit.Paths(absoluteMain),
+		configkit.SearchDirectory(filepath.Join(root, "fallback"), configkit.DirectorySearchOptions{
+			Filenames: []configkit.FilenameGroup{configkit.CustomFilenames(filepath.Base(absoluteMain))},
+		}),
+	), configkit.SearchOptions{Required: true})
+	definition := configkit.Definition[SequenceCatalogConfig]{
+		SequenceIdentities: []configkit.SequenceIdentity{
+			configkit.SequenceIdentityAt(configkit.Keys("prompts"), "id"),
+		},
+		Resolve: func(configuration *SequenceCatalogConfig, resolution *configkit.Resolution) error {
+			configuration.ResolvedLabel = "for:" + configuration.RunLabel
+			resolution.Record(configkit.Keys("resolved_label"), configkit.ChangeNormalization,
+				"derive the resolved label from the selected run label", configkit.Keys("run_label"))
+			return nil
+		},
+		Validate: validateSequenceCatalog,
+	}
+	return configkit.LoadLayered(configkit.Input{
+		Source:     mainSearch,
+		Imports:    true,
+		Filesystem: &configkit.FilesystemContext{Root: root},
+	}, configkit.LayeredOptions[SequenceCatalogConfig]{
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{{
+			Source: filesystem.File("underlay.yml"),
+		}}}},
+		Overlay:           overlayName,
+		OverlayFilesystem: filesystemContext,
+		Overrides: []configkit.Override{{
+			Path: configkit.Keys("run_label"), Value: runLabel, Flag: "--run-label",
+		}},
+		Definition: definition,
+	})
+}
+
+// ReadSequenceCatalogResource follows the final prompt field's source context,
+// including when record inheritance retained a resource from an underlay.
+func ReadSequenceCatalogResource(result *configkit.Result[SequenceCatalogConfig], promptID string) (string, error) {
+	if result == nil {
+		return "", errors.New("sequence catalog result is nil")
+	}
+	for index, prompt := range result.Config.Prompts {
+		if prompt.ID != promptID {
+			continue
+		}
+		resourcePath := configkit.Keys("prompts").Index(index).Key("resource")
+		info, found := result.Metadata().Lookup(resourcePath)
+		if !found || len(info.Origin.Locations) != 1 {
+			return "", errors.New("prompt resource has no single declaring source")
+		}
+		context, found := result.SourceContext(info.Origin.Locations[0].SourceID)
+		if !found {
+			return "", errors.New("prompt resource source context is unavailable")
+		}
+		data, err := fs.ReadFile(context.FS, path.Join(context.BaseDir, prompt.Resource))
+		if err != nil {
+			return "", fmt.Errorf("read prompt resource: %w", err)
+		}
+		return string(data), nil
+	}
+	return "", fmt.Errorf("prompt %q is unavailable", promptID)
+}
+
+func validateSequenceCatalog(configuration *SequenceCatalogConfig, validation *configkit.Validation) {
+	for index, prompt := range configuration.Prompts {
+		if strings.TrimSpace(prompt.Name) == "" {
+			validation.Error(configkit.Keys("prompts").Index(index).Key("name"), "must not be empty")
+		}
+		if len(prompt.Messages) == 0 {
+			validation.Error(configkit.Keys("prompts").Index(index).Key("messages"), "must contain at least one message")
+		}
+	}
+	if len(configuration.Prompts) == 0 {
+		validation.Error(configkit.Keys("prompts"), "must contain at least one prompt")
+	}
+}
+
+// NamespacedMapCatalogConfig demonstrates typed application maps whose
+// independent prompt sequences use root and dynamic namespace registrations.
+type NamespacedMapCatalogConfig struct {
+	Prompts     []map[string]any          `yaml:"prompts"`
+	Collections map[string]map[string]any `yaml:"collections"`
+}
+
+// LoadNamespacedMapCatalog applies the same scalar identity constraint to one
+// root collection and each namespace's independently scoped collection.
+func LoadNamespacedMapCatalog(source configkit.Source) (*configkit.Result[NamespacedMapCatalogConfig], error) {
+	selector := configkit.SequenceSelector{
+		configkit.SequenceLiteralKey("collections"),
+		configkit.SequenceAnyMappingKey(),
+		configkit.SequenceLiteralKey("prompts"),
+	}
+	return configkit.Load(source, configkit.Options[NamespacedMapCatalogConfig]{
+		Definition: configkit.Definition[NamespacedMapCatalogConfig]{
+			SequenceIdentities: []configkit.SequenceIdentity{
+				configkit.SequenceIdentityAt(configkit.Keys("prompts"), "id"),
+				configkit.SequenceIdentityMatching(selector, "id"),
+			},
+		},
+	})
+}

examples/consumers/sequence_catalog_test.go

@@ -0,0 +1,239 @@
+package consumers_test
+
+import (
+	"bytes"
+	"errors"
+	"fmt"
+	"os"
+	"strings"
+	"testing"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+	"gitea.maximumdirect.net/eric/ConfigKit/examples/consumers"
+)
+
+func Example_sequenceCatalog() {
+	const oldSecret = "CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET"
+	const newSecret = "CONFIGKIT_SEQUENCE_CATALOG_NEW_SECRET"
+	restoreOld, restoreNew := preserveEnvironment(oldSecret), preserveEnvironment(newSecret)
+	defer restoreOld()
+	defer restoreNew()
+	_ = os.Setenv(oldSecret, "disposable-old-example-credential")
+	_ = os.Setenv(newSecret, "disposable-new-example-credential")
+
+	result, err := consumers.LoadSequenceCatalog(fixture("sequence-catalog", "config.yml"), "upsert", "preview")
+	if err != nil {
+		fmt.Println(err)
+		return
+	}
+	ids := make([]string, 0, len(result.Config.Prompts))
+	for _, prompt := range result.Config.Prompts {
+		ids = append(ids, prompt.ID)
+	}
+	shared := configkit.Keys("prompts").Index(1).Key("name")
+	info, _ := result.Metadata().Lookup(shared)
+	resource, resourceErr := consumers.ReadSequenceCatalogResource(result, "concise")
+	fmt.Println(ids, result.Config.ResolvedLabel, info.Explicit, len(result.Metadata().History(shared)), resource, resourceErr == nil, err)
+	// Output: [foundation shared welcome concise extra] for:preview true 2 Foundation resource text.
+	//  true <nil>
+}
+
+func TestSequenceCatalogSelectedOverlayOperations(t *testing.T) {
+	const oldSecret = "CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET"
+	const newSecret = "CONFIGKIT_SEQUENCE_CATALOG_NEW_SECRET"
+	t.Setenv(oldSecret, "sequence-catalog-old-canary")
+	t.Setenv(newSecret, "sequence-catalog-new-canary")
+
+	tests := []struct {
+		name    string
+		wantIDs []string
+		shared  *consumers.SequenceCatalogEntry
+	}{
+		{name: "append", wantIDs: []string{"foundation", "shared", "welcome", "concise", "extra"}},
+		{name: "upsert", wantIDs: []string{"foundation", "shared", "welcome", "concise", "extra"}, shared: &consumers.SequenceCatalogEntry{ID: "shared", Name: "Shared replacement", Messages: []string{"Replacement message"}, Credential: configkit.SecretEnv(newSecret)}},
+		{name: "replace", wantIDs: []string{"replacement", "last"}},
+	}
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			result, err := consumers.LoadSequenceCatalog(fixture("sequence-catalog", "config.yml"), test.name, "test")
+			if err != nil {
+				var loadError *configkit.LoadError
+				if errors.As(err, &loadError) {
+					t.Fatalf("load selected overlay: %v (%#v)", err, loadError.Diagnostics())
+				}
+				t.Fatal(err)
+			}
+			gotIDs := make([]string, 0, len(result.Config.Prompts))
+			for _, prompt := range result.Config.Prompts {
+				gotIDs = append(gotIDs, prompt.ID)
+			}
+			if strings.Join(gotIDs, ",") != strings.Join(test.wantIDs, ",") {
+				t.Fatalf("prompt order = %v, want %v", gotIDs, test.wantIDs)
+			}
+			if test.shared == nil {
+				return
+			}
+			shared := result.Config.Prompts[1]
+			if shared.ID != test.shared.ID || shared.Name != test.shared.Name || len(shared.Messages) != 1 ||
+				shared.Messages[0] != test.shared.Messages[0] || shared.Description != "" || shared.Resource != "" || shared.Credential.EnvName() != newSecret {
+				t.Fatalf("upsert did not replace the complete matched record: %#v", shared)
+			}
+		})
+	}
+}
+
+func TestSequenceCatalogInspectionTracksResourcesAndSecretHistory(t *testing.T) {
+	const oldSecret = "CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET"
+	const newSecret = "CONFIGKIT_SEQUENCE_CATALOG_NEW_SECRET"
+	const oldCanary = "sequence-catalog-old-canary"
+	const newCanary = "sequence-catalog-new-canary"
+	t.Setenv(oldSecret, oldCanary)
+	t.Setenv(newSecret, newCanary)
+
+	result, err := consumers.LoadSequenceCatalog(fixture("sequence-catalog", "config.yml"), "upsert", "history")
+	if err != nil {
+		var loadError *configkit.LoadError
+		if errors.As(err, &loadError) {
+			t.Fatalf("load sequence catalog: %v (%#v)", err, loadError.Diagnostics())
+		}
+		t.Fatal(err)
+	}
+	resourcePath := configkit.Keys("prompts").Index(3).Key("resource")
+	resourceInfo, found := result.Metadata().Lookup(resourcePath)
+	if !found || len(resourceInfo.Origin.Locations) != 1 || len(resourceInfo.Origin.Inheritance) != 1 {
+		t.Fatalf("inherited resource inspection = %#v, %v", resourceInfo, found)
+	}
+	resource, err := consumers.ReadSequenceCatalogResource(result, "concise")
+	if err != nil || resource != "Foundation resource text.\n" {
+		t.Fatalf("resource from inherited source context = %q, %v", resource, err)
+	}
+	if len(result.Metadata().History(configkit.Keys("resolved_label"))) == 0 {
+		t.Fatal("recorded CLI-driven resolution was absent from final inspection")
+	}
+	credentialPath := configkit.Keys("prompts").Index(1).Key("credential")
+	credentialHistory := result.Metadata().History(credentialPath)
+	if len(credentialHistory) == 0 {
+		t.Fatal("upsert did not retain the replaced secret reference in history")
+	}
+	var explanation bytes.Buffer
+	if err := result.WriteExplanation(&explanation, credentialPath); err != nil {
+		t.Fatal(err)
+	}
+	if strings.Contains(explanation.String(), oldCanary) || strings.Contains(explanation.String(), newCanary) {
+		t.Fatalf("secret explanation exposed a credential value: %s", explanation.String())
+	}
+}
+
+func TestNamespacedTypedMapIdentityScopesAreIndependent(t *testing.T) {
+	source := configkit.Bytes("namespaced maps", []byte(`
+prompts:
+  - id: shared
+    text: root prompt
+collections:
+  team:
+    prompts:
+      - id: shared
+        text: team prompt
+  operations:
+    prompts:
+      - id: shared
+        text: operations prompt
+`))
+	result, err := consumers.LoadNamespacedMapCatalog(source)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if result.Config.Prompts[0]["text"] != "root prompt" ||
+		result.Config.Collections["team"]["prompts"].([]any)[0].(map[string]any)["text"] != "team prompt" ||
+		result.Config.Collections["operations"]["prompts"].([]any)[0].(map[string]any)["text"] != "operations prompt" {
+		t.Fatalf("typed map collections = %#v", result.Config)
+	}
+}
+
+func TestNamespacedTypedMapIdentityFindsCollectionsIntroducedByInheritance(t *testing.T) {
+	result, err := consumers.LoadNamespacedMapCatalog(configkit.Bytes("inherited namespace", []byte(`
+collections:
+  source:
+    prompts:
+      - id: shared
+        text: inherited prompt
+  copied:
+    extends: source
+`)))
+	if err != nil {
+		t.Fatal(err)
+	}
+	copied := result.Config.Collections["copied"]["prompts"].([]any)
+	if len(copied) != 1 || copied[0].(map[string]any)["text"] != "inherited prompt" {
+		t.Fatalf("inherited typed-map collection = %#v", result.Config.Collections["copied"])
+	}
+}
+
+func Example_yamlDeclaredIdentityOnly() {
+	result, err := configkit.Load(configkit.Bytes("YAML identity only", []byte(`
+configkit:
+  sequences:
+    - destination: [prompts]
+      key: id
+prompts:
+  - id: initial
+`)), configkit.Options[map[string]any]{
+		Overrides: []configkit.Override{{
+			Path: configkit.Keys("prompts"),
+			Value: []any{
+				map[string]any{"id": "same"},
+				map[string]any{"id": "same"},
+			},
+			Flag: "--prompts",
+		}},
+	})
+	fmt.Println(len(result.Config["prompts"].([]any)), err)
+	// Output: 2 <nil>
+}
+
+func TestYAMLDeclaredIdentityDoesNotRegisterFinalTypedConstraint(t *testing.T) {
+	result, err := configkit.Load(configkit.Bytes("YAML identity only", []byte(`
+configkit:
+  sequences:
+    - destination: [prompts]
+      key: id
+prompts:
+  - id: initial
+`)), configkit.Options[map[string]any]{
+		Overrides: []configkit.Override{{
+			Path: configkit.Keys("prompts"),
+			Value: []any{
+				map[string]any{"id": "same"},
+				map[string]any{"id": "same"},
+			},
+			Flag: "--prompts",
+		}},
+	})
+	if err != nil {
+		t.Fatalf("YAML-only identity incorrectly became an application constraint: %v", err)
+	}
+	prompts := result.Config["prompts"].([]any)
+	if len(prompts) != 2 || prompts[0].(map[string]any)["id"] != prompts[1].(map[string]any)["id"] {
+		t.Fatalf("CLI result = %#v", result.Config)
+	}
+}
+
+func TestNamespacedTypedMapRejectsDuplicatesWithinOneCollection(t *testing.T) {
+	_, err := consumers.LoadNamespacedMapCatalog(configkit.Bytes("duplicate namespace", []byte(`
+collections:
+  team:
+    prompts:
+      - id: duplicate
+      - id: duplicate
+`)))
+	var loadError *configkit.LoadError
+	if !errors.As(err, &loadError) || len(loadError.Diagnostics()) != 1 ||
+		loadError.Diagnostics()[0].Code != configkit.CodeComposeConflict ||
+		loadError.Diagnostics()[0].Path.String() != "/collections/team/prompts" {
+		diagnostics := []configkit.Diagnostic(nil)
+		if loadError != nil {
+			diagnostics = loadError.Diagnostics()
+		}
+		t.Fatalf("duplicate namespaced identity error = %v (%#v)", err, diagnostics)
+	}
+}

examples/consumers/source_version.go

@@ -0,0 +1,64 @@
+package consumers
+
+import (
+	"errors"
+	"io/fs"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+// VersionedServiceConfig demonstrates a mounted main input whose source-local
+// schema version is retained at the effective service path.
+type VersionedServiceConfig struct {
+	Service struct {
+		Version  string `yaml:"version"`
+		Name     string `yaml:"name"`
+		Endpoint string `yaml:"endpoint"`
+		Metadata struct {
+			Release string `yaml:"release"`
+		} `yaml:"metadata"`
+	} `yaml:"service"`
+}
+
+// LoadVersionedService checks the main and underlay declarations with one
+// application policy. The local rule path is "version" for each source; the
+// main input's destination places its protected value at /service/version.
+// Optional CLI overrides are passed through to ConfigKit for normal protection.
+func LoadVersionedService(mode configkit.SourceVersionMode, overrides ...configkit.Override) (*configkit.Result[VersionedServiceConfig], error) {
+	files, err := fs.Sub(bundledFiles, "testdata/source-version")
+	if err != nil {
+		return nil, err
+	}
+	main, err := fs.ReadFile(files, "main.yml")
+	if err != nil {
+		return nil, err
+	}
+	underlay, err := fs.ReadFile(files, "underlay.yml")
+	if err != nil {
+		return nil, err
+	}
+	definition := configkit.Definition[VersionedServiceConfig]{
+		Version: &configkit.VersionRule{
+			Path: configkit.Keys("version"),
+			Check: func(version configkit.Version, _ *configkit.Validation) error {
+				if !version.Present || version.Null || version.Text == "" {
+					return errors.New("a nonempty source version is required")
+				}
+				return nil
+			},
+		},
+		Secret: func(path configkit.Path) bool {
+			return path.String() == "/service/version"
+		},
+	}
+	return configkit.LoadLayered(configkit.Input{
+		Source: configkit.Bytes("service main", main), Destination: configkit.Keys("service"),
+	}, configkit.LayeredOptions[VersionedServiceConfig]{
+		Definition:        definition,
+		SourceVersionMode: mode,
+		Overrides:         overrides,
+		Underlays: []configkit.Underlay{{Inputs: []configkit.Input{{
+			Source: configkit.Bytes("service defaults", underlay), Destination: configkit.Keys("service"),
+		}}}},
+	})
+}

examples/consumers/testdata/collection-bundle/collections/weather/collection.yml

@@ -0,0 +1,9 @@
+configkit:
+  imports:
+    - path: metadata.yml
+      destination: []
+    - path: prompts/index.yml
+      destination: [prompts]
+    - path: profiles/index.yml
+      destination: [profiles]
+name: Weather Reporter

examples/consumers/testdata/collection-bundle/collections/weather/metadata.yml

@@ -0,0 +1,4 @@
+configkit:
+  imports:
+    - path: version.yml
+description: Imported collection metadata.

examples/consumers/testdata/collection-bundle/collections/weather/profiles/compact.yml

@@ -0,0 +1,3 @@
+compact:
+  format: text
+  temperature: 0.2

examples/consumers/testdata/collection-bundle/collections/weather/profiles/index.yml

@@ -0,0 +1,4 @@
+configkit:
+  imports:
+    - path: compact.yml
+    - path: weather.yml

examples/consumers/testdata/collection-bundle/collections/weather/profiles/weather.yml

@@ -0,0 +1,3 @@
+weather:
+  extends: compact
+  temperature: 0.4

examples/consumers/testdata/collection-bundle/collections/weather/prompts/brief.yml

@@ -0,0 +1,7 @@
+brief:
+  version: "1"
+  default_profile: compact
+  content_file: welcome.txt
+  messages:
+    - role: user
+      content: Write a concise weather summary.

examples/consumers/testdata/collection-bundle/collections/weather/prompts/custom.yml

@@ -0,0 +1,3 @@
+custom:
+  extends: brief
+  default_profile: weather

examples/consumers/testdata/collection-bundle/collections/weather/prompts/index.yml

@@ -0,0 +1,4 @@
+configkit:
+  imports:
+    - path: brief.yml
+    - path: custom.yml

examples/consumers/testdata/collection-bundle/collections/weather/prompts/welcome.txt

@@ -0,0 +1 @@
+Welcome to the weather collection.

examples/consumers/testdata/collection-bundle/collections/weather/version.yml

@@ -0,0 +1 @@
+version: "2.4.0"

examples/consumers/testdata/collection-bundle/operators/weather-prompt-version.yml

@@ -0,0 +1,2 @@
+brief:
+  version: "2"

examples/consumers/testdata/ordinary-imports/config.yml

@@ -0,0 +1,6 @@
+configkit:
+  imports:
+    - path: metadata.yml
+    - path: prompts/brief.yml
+      destination: [prompts]
+name: Root import example

examples/consumers/testdata/ordinary-imports/metadata.yml

@@ -0,0 +1,2 @@
+version: "1.0"
+description: Metadata imported into the root document.

examples/consumers/testdata/ordinary-imports/prompts/brief.yml

@@ -0,0 +1,2 @@
+brief:
+  summary: A root-loaded prompt.

examples/consumers/testdata/prompt-catalog/config.yml

@@ -0,0 +1,19 @@
+configkit:
+  sequences:
+    - destination: [prompts]
+      key: id
+      mode: replace
+  imports:
+    - dir: prompts
+      exclude: [generated-ignored.yml]
+      destination: [prompts]
+      placement: sequence-item
+    - dir: featured
+      pattern: [weather-*.yml]
+      destination: [prompts]
+      placement: sequence-item
+    - dir: optional
+prompts:
+  - id: main
+    name: Main authored prompt
+    messages: [Use the main prompt.]

examples/consumers/testdata/prompt-catalog/featured/weather-feature.yml

@@ -0,0 +1,4 @@
+id: weather
+name: Weather forecast
+messages:
+  - Describe today's forecast.

examples/consumers/testdata/prompt-catalog/invalid.yml

@@ -0,0 +1,9 @@
+configkit:
+  sequences:
+    - destination: [prompts]
+      key: id
+      mode: replace
+prompts:
+  - id: invalid
+    name: Invalid prompt
+    messages: []

examples/consumers/testdata/prompt-catalog/optional/ignored.txt

@@ -0,0 +1 @@
+This file does not match the application import defaults.

examples/consumers/testdata/prompt-catalog/prompts/curated.yml

@@ -0,0 +1,6 @@
+configkit:
+  imports:
+    - path: helpers/curated-id.yml
+name: Curated summary
+messages:
+  - Summarize the supplied content.

examples/consumers/testdata/prompt-catalog/prompts/generated-ignored.yml

@@ -0,0 +1,3 @@
+id: ignored
+name: Excluded generated prompt
+messages: [This record is excluded.]

examples/consumers/testdata/prompt-catalog/prompts/generated-welcome.yml

@@ -0,0 +1,5 @@
+id: welcome
+name: Welcome
+messages:
+  - Welcome the user.
+resource: welcome.txt

examples/consumers/testdata/prompt-catalog/prompts/helpers/curated-id.yml

@@ -0,0 +1 @@
+id: curated

examples/consumers/testdata/prompt-catalog/prompts/required.yml

@@ -0,0 +1,3 @@
+id: required
+name: Required prompt
+messages: [Use the required prompt.]

examples/consumers/testdata/prompt-catalog/prompts/welcome.txt

@@ -0,0 +1 @@
+Welcome resource text.

examples/consumers/testdata/prompt-catalog/simple.yml

@@ -0,0 +1,7 @@
+configkit:
+  imports:
+    - path: prompts/required.yml
+      required: true
+      destination: [prompts]
+      placement: sequence-item
+prompts: []

examples/consumers/testdata/sequence-catalog/config.yml

@@ -0,0 +1,7 @@
+run_label: source-default
+configkit:
+  imports:
+    - dir: fragments
+      pattern: '*.yml'
+      destination: [prompts]
+      placement: sequence-item

examples/consumers/testdata/sequence-catalog/foundation.txt

@@ -0,0 +1 @@
+Foundation resource text.

examples/consumers/testdata/sequence-catalog/fragments/01-welcome.yml

@@ -0,0 +1,4 @@
+id: welcome
+name: Welcome
+messages: [Welcome message]
+credential: CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET

examples/consumers/testdata/sequence-catalog/fragments/02-concise.yml

@@ -0,0 +1,3 @@
+id: concise
+extends: {id: foundation}
+name: Concise

examples/consumers/testdata/sequence-catalog/overlays/append.yml

@@ -0,0 +1,10 @@
+configkit:
+  sequences:
+    - destination: [prompts]
+      key: id
+      mode: append
+prompts:
+  - id: extra
+    name: Extra
+    messages: [Extra message]
+    credential: CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET

examples/consumers/testdata/sequence-catalog/overlays/replace.yml

@@ -0,0 +1,14 @@
+configkit:
+  sequences:
+    - destination: [prompts]
+      key: id
+      mode: replace
+prompts:
+  - id: replacement
+    name: Replacement
+    messages: [Replacement message]
+    credential: CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET
+  - id: last
+    name: Last
+    messages: [Last message]
+    credential: CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET

examples/consumers/testdata/sequence-catalog/overlays/upsert.yml

@@ -0,0 +1,14 @@
+configkit:
+  sequences:
+    - destination: [prompts]
+      key: id
+      mode: upsert
+prompts:
+  - id: shared
+    name: Shared replacement
+    messages: [Replacement message]
+    credential: CONFIGKIT_SEQUENCE_CATALOG_NEW_SECRET
+  - id: extra
+    name: Extra
+    messages: [Extra message]
+    credential: CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET

examples/consumers/testdata/sequence-catalog/underlay.yml

@@ -0,0 +1,18 @@
+configkit:
+  sequences:
+    - destination: [prompts]
+      key: id
+      mode: append
+prompts:
+  - id: foundation
+    name: Foundation
+    description: Shared foundation instruction
+    messages: [Foundation message]
+    resource: foundation.txt
+    credential: CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET
+  - id: shared
+    name: Shared draft
+    description: This field must not survive a whole-record replacement
+    messages: [Draft message]
+    resource: shared.txt
+    credential: CONFIGKIT_SEQUENCE_CATALOG_OLD_SECRET

examples/consumers/testdata/source-version/main.yml

@@ -0,0 +1,4 @@
+version: service-v2
+name: example service
+metadata:
+  release: application-2026.10

examples/consumers/testdata/source-version/underlay.yml

@@ -0,0 +1,2 @@
+version: catalog-v1
+endpoint: https://service.example.invalid

facade_compatibility_test.go

@@ -5,7 +5,6 @@ import (
 	"errors"
 	"fmt"
 	"io"
-	"reflect"
 	"testing"
 
 	configkit "gitea.maximumdirect.net/eric/ConfigKit"
@@ -27,23 +26,6 @@ type facadeCompatibilityConfig struct {
 func TestFacadeCompatibility(t *testing.T) {
 	t.Parallel()
 
-	const packagePath = "gitea.maximumdirect.net/eric/ConfigKit"
-	for _, candidate := range []struct {
-		name  string
-		type_ reflect.Type
-	}{
-		{"Path", reflect.TypeFor[configkit.Path]()},
-		{"Source", reflect.TypeFor[configkit.Source]()},
-		{"Secret", reflect.TypeFor[configkit.Secret]()},
-		{"Definition", reflect.TypeFor[configkit.Definition[facadeCompatibilityConfig]]()},
-		{"Options", reflect.TypeFor[configkit.Options[facadeCompatibilityConfig]]()},
-		{"Result", reflect.TypeFor[configkit.Result[facadeCompatibilityConfig]]()},
-	} {
-		if candidate.type_.PkgPath() != packagePath {
-			t.Errorf("%s belongs to %q, want %q", candidate.name, candidate.type_.PkgPath(), packagePath)
-		}
-	}
-
 	constructed := configkit.Options[facadeCompatibilityConfig]{
 		Definition: configkit.Definition[facadeCompatibilityConfig]{},
 		Filesystem: &configkit.FilesystemContext{BaseDir: "configs", Root: "configs"},
@@ -56,11 +38,6 @@ func TestFacadeCompatibility(t *testing.T) {
 	if got := constructed.Overrides[0].Path.String(); got != "/value" {
 		t.Fatalf("constructed override path = %q, want /value", got)
 	}
-	publicResult := configkit.Result[facadeCompatibilityConfig]{Config: facadeCompatibilityConfig{Value: "constructed"}}
-	if publicResult.Config.Value != "constructed" {
-		t.Fatal("constructed result lost its public configuration")
-	}
-
 	result, err := configkit.Load(configkit.Bytes("facade.yml", []byte("value: input\n")), configkit.Options[facadeCompatibilityConfig]{
 		Definition: configkit.Definition[facadeCompatibilityConfig]{
 			Resolve: func(configuration *facadeCompatibilityConfig, resolution *configkit.Resolution) error {

facade_destination_test.go

@@ -0,0 +1,705 @@
+package configkit_test
+
+import (
+	"bytes"
+	"errors"
+	"fmt"
+	"io"
+	"io/fs"
+	"os"
+	"path"
+	"path/filepath"
+	"reflect"
+	"strings"
+	"testing"
+	"testing/fstest"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+func TestFacadeInputDestinationsAcrossSourceKinds(t *testing.T) {
+	directory := t.TempDir()
+	hostPath := filepath.Join(directory, "host.yml")
+	if err := os.WriteFile(hostPath, []byte("value: host\n"), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	native := configkit.NewFilesystem("native", fstest.MapFS{
+		"main.yml": &fstest.MapFile{Data: []byte("value: native\n")},
+	})
+	const literalValuePath = "/collections/weather.reporter/profiles/a~1b/~0/0"
+	destination := configkit.Keys("collections", "weather.reporter", "profiles", "a/b", "~", "0")
+	for _, test := range []struct {
+		name   string
+		source configkit.Source
+		want   string
+		kind   configkit.SourceKind
+	}{
+		{name: "host file", source: configkit.File(hostPath), want: "host", kind: configkit.SourceKindFile},
+		{name: "native filesystem", source: native.File("main.yml"), want: "native", kind: configkit.SourceKindFilesystem},
+		{name: "bytes", source: configkit.Bytes("bytes", []byte("value: bytes\n")), want: "bytes", kind: configkit.SourceKindBytes},
+		{name: "reader", source: configkit.Reader("reader", strings.NewReader("value: reader\n")), want: "reader", kind: configkit.SourceKindReader},
+		{name: "mapping", source: configkit.Mapping("mapping", map[string]any{"value": "mapping"}), want: "mapping", kind: configkit.SourceKindMapping},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			result, err := configkit.LoadLayered[map[string]any](configkit.Input{Source: test.source, Destination: destination})
+			if err != nil {
+				t.Fatalf("LoadLayered: %v", err)
+			}
+			if got := facadeDestinationValue(t, result.Config, "collections", "weather.reporter", "profiles", "a/b", "~", "0", "value"); got != test.want {
+				t.Fatalf("placed value = %#v, want %q", got, test.want)
+			}
+			sources := result.Sources()
+			if len(sources) != 1 || sources[0].Kind != test.kind || sources[0].Destination.String() != destination.String() || destination.String() != literalValuePath {
+				t.Fatalf("source placement = %#v, destination = %q", sources, destination.String())
+			}
+		})
+	}
+}
+
+func TestFacadeDestinationRootEmptyMappingNoSourceAndInvalidPaths(t *testing.T) {
+	for _, test := range []struct {
+		name        string
+		destination configkit.Path
+	}{
+		{name: "omitted destination"},
+		{name: "explicit empty destination", destination: configkit.Path{}},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			result, err := configkit.LoadLayered[map[string]any](configkit.Input{Source: configkit.Bytes("root", []byte("value: root\n")), Destination: test.destination})
+			if err != nil {
+				t.Fatalf("root load: %v", err)
+			}
+			if result.Config["value"] != "root" || len(result.Sources()) != 1 || result.Sources()[0].Destination.String() != "" {
+				t.Fatalf("root placement = %#v, sources %#v", result.Config, result.Sources())
+			}
+		})
+	}
+
+	emptyPath := configkit.Keys("collections", "empty")
+	empty, err := configkit.LoadLayered[map[string]any](configkit.Input{
+		Source: configkit.Mapping("empty", map[string]any{}), Destination: emptyPath,
+	})
+	if err != nil {
+		t.Fatalf("empty mapping load: %v", err)
+	}
+	if value := facadeDestinationMap(t, empty.Config, "collections")["empty"]; value == nil {
+		t.Fatal("empty mapping did not materialize its destination")
+	}
+	if info, ok := empty.Metadata().Lookup(emptyPath); !ok || !info.Explicit {
+		t.Fatalf("empty destination metadata = %#v, %v", info, ok)
+	}
+
+	absentPath := configkit.Keys("collections", "absent")
+	absent, err := configkit.LoadLayered[map[string]any](configkit.Input{Source: configkit.NoSource(), Destination: absentPath})
+	if err != nil {
+		t.Fatalf("NoSource with a destination: %v", err)
+	}
+	if len(absent.Sources()) != 0 {
+		t.Fatalf("NoSource created source records: %#v", absent.Sources())
+	}
+	if _, ok := absent.Metadata().Lookup(absentPath); ok {
+		t.Fatal("NoSource materialized its destination")
+	}
+
+	reader := &destinationCountingReader{}
+	for _, test := range []struct {
+		name        string
+		source      configkit.Source
+		destination configkit.Path
+		wantReads   int
+	}{
+		{name: "index segment", source: configkit.Reader("unread", reader), destination: configkit.Keys("catalog").Index(0)},
+		{name: "reserved configkit key", source: configkit.NoSource(), destination: configkit.Keys("configkit")},
+		{name: "reserved extends key", source: configkit.NoSource(), destination: configkit.Keys("catalog", "extends")},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			_, err := configkit.LoadLayered[map[string]any](configkit.Input{Source: test.source, Destination: test.destination})
+			loadError, ok := err.(*configkit.LoadError)
+			if !ok || len(loadError.Diagnostics()) == 0 || loadError.Diagnostics()[0].Code != configkit.CodeOptionsInvalid {
+				t.Fatalf("invalid destination error = %#v", err)
+			}
+			if reader.reads != test.wantReads {
+				t.Fatalf("reader calls = %d, want %d", reader.reads, test.wantReads)
+			}
+		})
+	}
+}
+
+func TestFacadeImportedSourcesRetainInputDestination(t *testing.T) {
+	directory := t.TempDir()
+	mainPath := filepath.Join(directory, "main.yml")
+	fragmentPath := filepath.Join(directory, "fragment.yml")
+	if err := os.WriteFile(mainPath, []byte("configkit: {imports: [{path: fragment.yml}]}\nmain: present\n"), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.WriteFile(fragmentPath, []byte("imported: present\n"), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	destination := configkit.Keys("collections", "weatherreporter")
+	result, err := configkit.LoadLayered[map[string]any](configkit.Input{
+		Source: configkit.File(mainPath), Imports: true, Destination: destination,
+	})
+	if err != nil {
+		t.Fatalf("mounted import load: %v", err)
+	}
+	collection := facadeDestinationMap(t, result.Config, "collections", "weatherreporter")
+	if collection["main"] != "present" || collection["imported"] != "present" {
+		t.Fatalf("mounted imported values = %#v", collection)
+	}
+	sources := result.Sources()
+	if len(sources) != 2 || sources[0].Destination.String() != destination.String() || sources[1].Destination.String() != destination.String() {
+		t.Fatalf("mounted import source destinations = %#v", sources)
+	}
+	if sources[1].ImportedFrom == nil {
+		t.Fatalf("imported source lacks its authored directive: %#v", sources[1])
+	}
+}
+
+func TestFacadeRelativeImportDestinationsForRootAndMountedLoads(t *testing.T) {
+	directory := t.TempDir()
+	write := func(relative, contents string) string {
+		t.Helper()
+		filename := filepath.Join(directory, filepath.FromSlash(relative))
+		if err := os.MkdirAll(filepath.Dir(filename), 0o755); err != nil {
+			t.Fatal(err)
+		}
+		if err := os.WriteFile(filename, []byte(contents), 0o600); err != nil {
+			t.Fatal(err)
+		}
+		return filename
+	}
+
+	mainPath := write("main.yml", `configkit:
+  imports:
+    - path: metadata.yml
+      destination: [metadata]
+    - glob: prompts/*.yml
+      destination: [prompts]
+    - path: entries/main.yml
+      destination: [profiles]
+    - path: sibling.yml
+      destination: [sibling]
+root: main
+`)
+	write("metadata.yml", "name: imported\n")
+	write("prompts/a.yml", "a: {name: alpha}\n")
+	write("prompts/b.yml", "b: {name: beta}\n")
+	write("entries/main.yml", `configkit:
+  imports:
+    - path: nested/details.yml
+      destination: [details]
+name: profile
+`)
+	write("entries/nested/details.yml", "summary: nested\n")
+	write("sibling.yml", "name: separate\n")
+
+	root, err := configkit.Load(
+		configkit.File(mainPath), configkit.Options[map[string]any]{Imports: true},
+	)
+	if err != nil {
+		if loadError, ok := err.(*configkit.LoadError); ok {
+			t.Fatalf("ordinary root Load: %v, diagnostics %#v", err, loadError.Diagnostics())
+		}
+		t.Fatalf("ordinary root Load: %v", err)
+	}
+	mountedDestination := configkit.Keys("collections", "weatherreporter")
+	mounted, err := configkit.LoadLayered[map[string]any](configkit.Input{
+		Source: configkit.File(mainPath), Imports: true, Destination: mountedDestination,
+	})
+	if err != nil {
+		t.Fatalf("mounted layered Load: %v", err)
+	}
+
+	for _, test := range []struct {
+		name   string
+		result *configkit.Result[map[string]any]
+		prefix []string
+	}{
+		{name: "ordinary root", result: root},
+		{name: "mounted entry", result: mounted, prefix: []string{"collections", "weatherreporter"}},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			at := func(keys ...string) []string {
+				return append(append([]string(nil), test.prefix...), keys...)
+			}
+			if got := facadeDestinationValue(t, test.result.Config, at("root")...); got != "main" {
+				t.Fatalf("root value = %#v", got)
+			}
+			if got := facadeDestinationValue(t, test.result.Config, at("metadata", "name")...); got != "imported" {
+				t.Fatalf("metadata value = %#v", got)
+			}
+			if got := facadeDestinationValue(t, test.result.Config, at("prompts", "a", "name")...); got != "alpha" {
+				t.Fatalf("first glob value = %#v", got)
+			}
+			if got := facadeDestinationValue(t, test.result.Config, at("prompts", "b", "name")...); got != "beta" {
+				t.Fatalf("second glob value = %#v", got)
+			}
+			if got := facadeDestinationValue(t, test.result.Config, at("profiles", "name")...); got != "profile" {
+				t.Fatalf("entry value = %#v", got)
+			}
+			if got := facadeDestinationValue(t, test.result.Config, at("profiles", "details", "summary")...); got != "nested" {
+				t.Fatalf("nested import value = %#v", got)
+			}
+			if got := facadeDestinationValue(t, test.result.Config, at("sibling", "name")...); got != "separate" {
+				t.Fatalf("subsequent sibling value = %#v", got)
+			}
+
+			wantPaths := []string{
+				"main.yml", "metadata.yml", "a.yml", "b.yml", "main.yml", "details.yml", "sibling.yml",
+			}
+			wantPhysicalPaths := []string{
+				mainPath,
+				filepath.Join(directory, "metadata.yml"),
+				filepath.Join(directory, "prompts", "a.yml"),
+				filepath.Join(directory, "prompts", "b.yml"),
+				filepath.Join(directory, "entries", "main.yml"),
+				filepath.Join(directory, "entries", "nested", "details.yml"),
+				filepath.Join(directory, "sibling.yml"),
+			}
+			wantImportParents := []string{"", "source-1", "source-1", "source-1", "source-1", "source-5", "source-1"}
+			wantDestinations := []configkit.Path{configkit.Keys(test.prefix...)}
+			for _, suffix := range [][]string{{"metadata"}, {"prompts"}, {"prompts"}, {"profiles"}, {"profiles", "details"}, {"sibling"}} {
+				destination := append([]string(nil), test.prefix...)
+				for _, segment := range suffix {
+					destination = append(destination, segment)
+				}
+				wantDestinations = append(wantDestinations, configkit.Keys(destination...))
+			}
+			sources := test.result.Sources()
+			if len(sources) != len(wantPaths) {
+				t.Fatalf("sources = %#v, want %d", sources, len(wantPaths))
+			}
+			for index, source := range sources {
+				if filepath.Base(source.Path) != wantPaths[index] || source.Path != wantPhysicalPaths[index] || source.Destination.String() != wantDestinations[index].String() {
+					t.Errorf("source %d path/destination = (%q, %q), want physical path %q and destination %q", index, source.Path, source.Destination.String(), wantPhysicalPaths[index], wantDestinations[index].String())
+				}
+				if source.ID != fmt.Sprintf("source-%d", index+1) || source.Kind != configkit.SourceKindFile {
+					t.Errorf("source %d identity/kind = (%q, %q)", index, source.ID, source.Kind)
+				}
+				if wantImportParents[index] == "" {
+					if source.ImportedFrom != nil {
+						t.Errorf("entry source unexpectedly has an import location: %#v", source.ImportedFrom)
+					}
+				} else if source.ImportedFrom == nil || source.ImportedFrom.SourceID != wantImportParents[index] || source.ImportedFrom.Line == 0 || source.ImportedFrom.Column == 0 {
+					t.Errorf("source %d import location = %#v, want declaration in %s", index, source.ImportedFrom, wantImportParents[index])
+				}
+			}
+		})
+	}
+}
+
+func TestFacadeSameFragmentMountsAtSeparateCollectionDestinations(t *testing.T) {
+	weatherDestination := configkit.Keys("collections", "weather", "prompts")
+	homeDestination := configkit.Keys("collections", "homewatch", "prompts")
+	result, err := configkit.LoadLayered(
+		configkit.Input{Source: configkit.NoSource()},
+		configkit.LayeredOptions[map[string]any]{Underlays: []configkit.Underlay{{Inputs: []configkit.Input{
+			{Source: configkit.Bytes("weather", []byte("summarize.alerts: {title: weather}\n")), Destination: weatherDestination},
+			{Source: configkit.Bytes("homewatch", []byte("summarize.alerts: {title: home}\n")), Destination: homeDestination},
+		}}}},
+	)
+	if err != nil {
+		t.Fatalf("LoadLayered: %v", err)
+	}
+	for _, test := range []struct {
+		path []string
+		want string
+	}{
+		{path: []string{"collections", "weather", "prompts", "summarize.alerts", "title"}, want: "weather"},
+		{path: []string{"collections", "homewatch", "prompts", "summarize.alerts", "title"}, want: "home"},
+	} {
+		if got := facadeDestinationValue(t, result.Config, test.path...); got != test.want {
+			t.Errorf("value at %v = %#v, want %q", test.path, got, test.want)
+		}
+	}
+	sources := result.Sources()
+	if len(sources) != 2 || sources[0].Destination.String() != weatherDestination.String() || sources[1].Destination.String() != homeDestination.String() {
+		t.Fatalf("independent collection source placements = %#v", sources)
+	}
+}
+
+type destinationCountingReader struct{ reads int }
+
+func (reader *destinationCountingReader) Read([]byte) (int, error) {
+	reader.reads++
+	return 0, io.EOF
+}
+
+func TestFacadeMountedMainAndUnderlayInspectionResourcesAndDiff(t *testing.T) {
+	t.Parallel()
+
+	filesystem := configkit.NewFilesystem("resources", fstest.MapFS{
+		"assets/summary.md": &fstest.MapFile{Data: []byte("source-relative summary")},
+	})
+	context := &configkit.FilesystemContext{Root: ".", BaseDir: "assets", FS: &filesystem}
+	destination := configkit.Keys("catalog", "entries")
+	tokenPath := configkit.Keys("catalog", "entries", "item", "token")
+	titlePath := configkit.Keys("catalog", "entries", "item", "title")
+	var callbacks struct{ resolve, validate int }
+	load := func(token string) *configkit.Result[map[string]any] {
+		t.Helper()
+		main := configkit.Input{
+			Source:      configkit.Bytes("main", []byte("item:\n  title: main\n  resource: summary.md\n  token: "+token+"\n")),
+			Destination: destination, Filesystem: context,
+		}
+		underlay := configkit.Input{
+			Source:      configkit.Bytes("underlay", []byte("item:\n  title: lower\n  retained: lower-value\n  token: LOWER-TOKEN-CANARY\n")),
+			Destination: destination,
+		}
+		result, err := configkit.LoadLayered(main, configkit.LayeredOptions[map[string]any]{
+			Underlays: []configkit.Underlay{{Inputs: []configkit.Input{underlay}}},
+			Definition: configkit.Definition[map[string]any]{
+				Resolve: func(configuration *map[string]any, resolution *configkit.Resolution) error {
+					callbacks.resolve++
+					resourcePath := configkit.Keys("catalog", "entries", "item", "resource")
+					input, ok := resolution.InputInfo(resourcePath)
+					if !ok || len(input.Origin.Locations) != 1 {
+						return fmt.Errorf("mounted resource input has no single source origin")
+					}
+					source, ok := resolution.Source(input.Origin.Locations[0].SourceID)
+					if !ok || source.Destination.String() != destination.String() {
+						return fmt.Errorf("resolution source did not retain its destination")
+					}
+					resourceContext, ok := resolution.SourceContext(source.ID)
+					if !ok {
+						return fmt.Errorf("resolution source context is unavailable")
+					}
+					data, err := fs.ReadFile(resourceContext.FS, path.Join(resourceContext.BaseDir, "summary.md"))
+					if err != nil {
+						return fmt.Errorf("read source-relative resource: %w", err)
+					}
+					entry := facadeDestinationMap(t, facadeDestinationMap(t, facadeDestinationMap(t, *configuration, "catalog"), "entries"), "item")
+					entry["resource_text"] = string(data)
+					resolution.Record(configkit.Keys("catalog", "entries", "item", "resource_text"), configkit.ChangeNormalization, "read source-relative summary", resourcePath)
+					return nil
+				},
+				Validate: func(*map[string]any, *configkit.Validation) { callbacks.validate++ },
+				Secret:   func(candidate configkit.Path) bool { return candidate.String() == tokenPath.String() },
+			},
+		})
+		if err != nil {
+			t.Fatalf("mounted load: %v", err)
+		}
+		return result
+	}
+
+	first := load("FIRST-TOKEN-CANARY")
+	second := load("SECOND-TOKEN-CANARY")
+	if callbacks.resolve != 2 || callbacks.validate != 2 {
+		t.Fatalf("lifecycle callbacks = %#v", callbacks)
+	}
+	for _, result := range []*configkit.Result[map[string]any]{first, second} {
+		entry := facadeDestinationMap(t, facadeDestinationMap(t, facadeDestinationMap(t, result.Config, "catalog"), "entries"), "item")
+		if entry["title"] != "main" || entry["retained"] != "lower-value" || entry["resource_text"] != "source-relative summary" {
+			t.Fatalf("mounted main/underlay configuration = %#v", entry)
+		}
+		sources := result.Sources()
+		if len(sources) != 2 || sources[0].Destination.String() != destination.String() || sources[1].Destination.String() != destination.String() {
+			t.Fatalf("public source destinations = %#v", sources)
+		}
+		resource, ok := result.SourceContext(sources[0].ID)
+		if !ok || resource.BaseDir != "assets" {
+			t.Fatalf("completed source context = %#v, %v", resource, ok)
+		}
+		data, err := fs.ReadFile(resource.FS, path.Join(resource.BaseDir, "summary.md"))
+		if err != nil || string(data) != "source-relative summary" {
+			t.Fatalf("completed relative resource = %q, %v", data, err)
+		}
+		if info, ok := result.Metadata().Lookup(titlePath); !ok || info.Origin.Layer != configkit.LayerBase {
+			t.Fatalf("mounted title metadata = %#v, %v", info, ok)
+		}
+		if history := result.Metadata().History(titlePath); len(history) < 2 {
+			t.Fatalf("mounted title history = %#v", history)
+		}
+		if explanation, ok := result.Explain(tokenPath); !ok || !explanation.Effective.Redacted {
+			t.Fatalf("mounted secret explanation = %#v, %v", explanation, ok)
+		}
+		var effective, explanation bytes.Buffer
+		if err := result.WriteEffective(&effective); err != nil {
+			t.Fatal(err)
+		}
+		if err := result.WriteExplanation(&explanation, tokenPath); err != nil {
+			t.Fatal(err)
+		}
+		output := effective.String() + explanation.String()
+		for _, canary := range []string{"FIRST-TOKEN-CANARY", "SECOND-TOKEN-CANARY", "LOWER-TOKEN-CANARY"} {
+			if strings.Contains(output, canary) {
+				t.Fatalf("public inspection leaked %q: %s", canary, output)
+			}
+		}
+	}
+
+	// Mutating published records and query copies must not alter the retained result.
+	returnedSources := first.Sources()
+	returnedSources[0].Destination = configkit.Keys("mutated")
+	if got := first.Sources()[0].Destination.String(); got != destination.String() {
+		t.Fatalf("source destination changed after caller mutation: %q", got)
+	}
+	info, ok := first.Metadata().Lookup(titlePath)
+	if !ok {
+		t.Fatal("mounted metadata lookup failed")
+	}
+	info.Origin.Locations[0].SourceID = "mutated"
+	if fresh, ok := first.Metadata().Lookup(titlePath); !ok || fresh.Origin.Locations[0].SourceID == "mutated" {
+		t.Fatalf("metadata copy mutation escaped: %#v, %v", fresh, ok)
+	}
+	history := first.Metadata().History(titlePath)
+	if len(history) == 0 {
+		t.Fatal("mounted history is empty")
+	}
+	history[0].Path = configkit.Keys("mutated")
+	if fresh := first.Metadata().History(titlePath); len(fresh) == 0 || fresh[0].Path.String() != titlePath.String() {
+		t.Fatalf("history copy mutation escaped: %#v", fresh)
+	}
+
+	differences, err := configkit.Diff(first, second)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(differences) != 1 || differences[0].Path.String() != tokenPath.String() ||
+		!differences[0].Left.Redacted || !differences[0].Right.Redacted ||
+		strings.Contains(differences[0].Left.YAML+differences[0].Right.YAML, "TOKEN-CANARY") {
+		t.Fatalf("public mounted secret diff = %#v", differences)
+	}
+	if callbacks.resolve != 2 || callbacks.validate != 2 {
+		t.Fatalf("inspection reran lifecycle callbacks: %#v", callbacks)
+	}
+}
+
+func TestFacadeFailedLoadExposesDetachedSourceDestination(t *testing.T) {
+	type configuration struct {
+		Tenant struct {
+			Count int `yaml:"count"`
+		} `yaml:"tenant"`
+	}
+	destination := configkit.Keys("tenant")
+	_, err := configkit.LoadLayered[configuration](configkit.Input{
+		Source: configkit.Bytes("invalid", []byte("count: not-an-integer\n")), Destination: destination,
+	})
+	loadError, ok := err.(*configkit.LoadError)
+	if !ok {
+		t.Fatalf("failed load error = %T %v", err, err)
+	}
+	sources := loadError.Sources()
+	if len(sources) != 1 || sources[0].Destination.String() != destination.String() {
+		t.Fatalf("failed source destinations = %#v", sources)
+	}
+	sources[0].Destination = configkit.Keys("caller-change")
+	if fresh := loadError.Sources(); len(fresh) != 1 || fresh[0].Destination.String() != destination.String() {
+		t.Fatalf("failed source destination was not detached: %#v", fresh)
+	}
+}
+
+func TestFacadeRelativeImportFailuresRetainSourceRecordsAndCoordinates(t *testing.T) {
+	directory := t.TempDir()
+	write := func(name, contents string) string {
+		t.Helper()
+		filename := filepath.Join(directory, name)
+		if err := os.MkdirAll(filepath.Dir(filename), 0o755); err != nil {
+			t.Fatal(err)
+		}
+		if err := os.WriteFile(filename, []byte(contents), 0o600); err != nil {
+			t.Fatal(err)
+		}
+		return filename
+	}
+
+	t.Run("imported parse failure retains effective source destination", func(t *testing.T) {
+		mainPath := write("parse/main.yml", "configkit:\n  imports:\n    - path: child.yml\n      destination: [tenant]\n")
+		childPath := write("parse/child.yml", "count: [unterminated\n")
+		_, err := configkit.LoadLayered[map[string]any](configkit.Input{
+			Source: configkit.File(mainPath), Imports: true, Destination: configkit.Keys("mount"),
+		})
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) {
+			t.Fatalf("load error = %T %v", err, err)
+		}
+		diagnostics := loadError.Diagnostics()
+		if len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeYAMLInvalid || diagnostics[0].Primary == nil || diagnostics[0].Primary.SourceID != "source-2" {
+			t.Fatalf("imported parse diagnostic = %#v", diagnostics)
+		}
+		sources := loadError.Sources()
+		if len(sources) != 2 || sources[0].Destination.String() != "/mount" || sources[1].Destination.String() != "/mount/tenant" || sources[1].Path != childPath {
+			t.Fatalf("imported parse failure sources = %#v", sources)
+		}
+	})
+
+	t.Run("directive failure stays source local", func(t *testing.T) {
+		mainPath := write("directive/main.yml", "configkit:\n  imports:\n    - path: parent.yml\n      destination: [tenant]\n")
+		parentPath := write("directive/parent.yml", "configkit:\n  imports:\n    - path: details.yml\n      destination: [3]\n")
+		_, err := configkit.LoadLayered[map[string]any](configkit.Input{
+			Source: configkit.File(mainPath), Imports: true, Destination: configkit.Keys("mount"),
+		})
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) {
+			t.Fatalf("load error = %T %v", err, err)
+		}
+		diagnostics := loadError.Diagnostics()
+		wantPath := configkit.Keys("configkit", "imports").Index(0).Key("destination").Index(0)
+		if len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeImportInvalid || !reflect.DeepEqual(diagnostics[0].Path, wantPath) || diagnostics[0].Primary == nil || diagnostics[0].Primary.SourceID != "source-2" {
+			t.Fatalf("import directive diagnostic = %#v", diagnostics)
+		}
+		sources := loadError.Sources()
+		if len(sources) != 2 || sources[0].Destination.String() != "/mount" || sources[1].Destination.String() != "/mount/tenant" || sources[1].Path != parentPath {
+			t.Fatalf("directive failure sources = %#v", sources)
+		}
+	})
+
+	t.Run("composition failure reports effective path and all sources", func(t *testing.T) {
+		mainPath := write("compose/main.yml", "configkit:\n  imports:\n    - path: left.yml\n      destination: [tenant]\n    - path: right.yml\n      destination: [tenant]\n")
+		write("compose/left.yml", "value: same\n")
+		write("compose/right.yml", "value: same\n")
+		_, err := configkit.LoadLayered[map[string]any](configkit.Input{
+			Source: configkit.File(mainPath), Imports: true, Destination: configkit.Keys("mount"),
+		})
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) {
+			t.Fatalf("load error = %T %v", err, err)
+		}
+		diagnostics := loadError.Diagnostics()
+		if len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeComposeConflict || diagnostics[0].Path.String() != "/mount/tenant/value" {
+			t.Fatalf("composition diagnostic = %#v", diagnostics)
+		}
+		sources := loadError.Sources()
+		if len(sources) != 3 || sources[0].Destination.String() != "/mount" || sources[1].Destination.String() != "/mount/tenant" || sources[2].Destination.String() != "/mount/tenant" {
+			t.Fatalf("composition failure sources = %#v", sources)
+		}
+	})
+
+	t.Run("version failure reports effective path and imported source", func(t *testing.T) {
+		mainPath := write("version/main.yml", "configkit:\n  imports:\n    - path: child.yml\n      destination: [tenant]\n")
+		childPath := write("version/child.yml", "version: forbidden\n")
+		_, err := configkit.LoadLayered[map[string]any](configkit.Input{
+			Source: configkit.File(mainPath), Imports: true, Destination: configkit.Keys("mount"),
+		}, configkit.LayeredOptions[map[string]any]{
+			Definition: configkit.Definition[map[string]any]{Version: &configkit.VersionRule{
+				Path:  configkit.Keys("mount", "tenant", "version"),
+				Check: func(configkit.Version, *configkit.Validation) error { return nil },
+			}},
+		})
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) {
+			t.Fatalf("load error = %T %v", err, err)
+		}
+		diagnostics := loadError.Diagnostics()
+		if len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeImportVersion || diagnostics[0].Path.String() != "/mount/tenant/version" || diagnostics[0].Primary == nil || diagnostics[0].Primary.SourceID != "source-2" {
+			t.Fatalf("version diagnostic = %#v", diagnostics)
+		}
+		sources := loadError.Sources()
+		if len(sources) != 2 || sources[0].Destination.String() != "/mount" || sources[1].Destination.String() != "/mount/tenant" || sources[1].Path != childPath {
+			t.Fatalf("version failure sources = %#v", sources)
+		}
+	})
+
+	t.Run("typed validation reports effective path and authored location", func(t *testing.T) {
+		type configuration struct {
+			Mount struct {
+				Tenant struct {
+					Count int `yaml:"count"`
+				} `yaml:"tenant"`
+			} `yaml:"mount"`
+		}
+		mainPath := write("typed/main.yml", "configkit:\n  imports:\n    - path: child.yml\n      destination: [tenant]\n")
+		childPath := write("typed/child.yml", "count: invalid\n")
+		_, err := configkit.LoadLayered[configuration](configkit.Input{
+			Source: configkit.File(mainPath), Imports: true, Destination: configkit.Keys("mount"),
+		})
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) {
+			t.Fatalf("load error = %T %v", err, err)
+		}
+		diagnostics := loadError.Diagnostics()
+		if len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeDecodeType || diagnostics[0].Path.String() != "/mount/tenant/count" || diagnostics[0].Primary == nil || diagnostics[0].Primary.SourceID != "source-2" || diagnostics[0].Primary.Line != 1 {
+			t.Fatalf("typed validation diagnostic = %#v", diagnostics)
+		}
+		sources := loadError.Sources()
+		if len(sources) != 2 || sources[0].Destination.String() != "/mount" || sources[1].Destination.String() != "/mount/tenant" || sources[1].Path != childPath {
+			t.Fatalf("typed validation sources = %#v", sources)
+		}
+	})
+}
+
+func TestFacadeDestinationMappingFailurePaths(t *testing.T) {
+	t.Parallel()
+
+	cycle := map[string]any{}
+	cycle["self"] = cycle
+	for _, fixture := range []struct {
+		name   string
+		values map[string]any
+		path   func(configkit.Path) configkit.Path
+	}{
+		{
+			name:   "unsupported nested sequence value",
+			values: map[string]any{"a/b": []any{map[string]any{"~": func() {}}}},
+			path:   func(base configkit.Path) configkit.Path { return base.Key("a/b").Index(0).Key("~") },
+		},
+		{
+			name:   "cyclic value",
+			values: map[string]any{"item": cycle},
+			path:   func(base configkit.Path) configkit.Path { return base.Key("item").Key("self") },
+		},
+		{
+			name: "constructor failure keeps source coordinates",
+			path: func(configkit.Path) configkit.Path { return configkit.Path{} },
+		},
+	} {
+		t.Run(fixture.name, func(t *testing.T) {
+			for _, role := range []string{"root", "mounted main", "mounted underlay"} {
+				t.Run(role, func(t *testing.T) {
+					var destination configkit.Path
+					if role != "root" {
+						destination = configkit.Keys("catalog", "entries")
+					}
+					wantPath := fixture.path(destination)
+					input := configkit.Input{Source: configkit.Mapping("invalid mapping", fixture.values), Destination: destination}
+					var result *configkit.Result[map[string]any]
+					var err error
+					if role == "mounted underlay" {
+						result, err = configkit.LoadLayered(configkit.Input{Source: configkit.NoSource()}, configkit.LayeredOptions[map[string]any]{
+							Underlays: []configkit.Underlay{{Inputs: []configkit.Input{input}}},
+						})
+					} else {
+						result, err = configkit.LoadLayered[map[string]any](input)
+					}
+					var loadError *configkit.LoadError
+					if result != nil || !errors.As(err, &loadError) {
+						t.Fatalf("failed load = %#v, %v", result, err)
+					}
+					diagnostics := loadError.Diagnostics()
+					if len(diagnostics) != 1 {
+						t.Fatalf("diagnostics = %#v", diagnostics)
+					}
+					diagnostic := diagnostics[0]
+					if diagnostic.Code != configkit.CodeSourceInvalid || !reflect.DeepEqual(diagnostic.Path, wantPath) {
+						t.Fatalf("diagnostic = %#v; want source.invalid at %q", diagnostic, wantPath.String())
+					}
+					if diagnostic.Primary == nil || *diagnostic.Primary != (configkit.Location{SourceID: "source-1"}) || len(diagnostic.Related) != 0 {
+						t.Fatalf("mapping source locations = %#v, %#v", diagnostic.Primary, diagnostic.Related)
+					}
+					sources := loadError.Sources()
+					if len(sources) != 1 || sources[0].ID != "source-1" || sources[0].Destination.String() != destination.String() {
+						t.Fatalf("error sources = %#v", sources)
+					}
+				})
+			}
+		})
+	}
+}
+
+func facadeDestinationMap(t *testing.T, value map[string]any, keys ...string) map[string]any {
+	t.Helper()
+	for _, key := range keys {
+		child, ok := value[key].(map[string]any)
+		if !ok {
+			t.Fatalf("%q = %#v, want mapping", key, value[key])
+		}
+		value = child
+	}
+	return value
+}
+
+func facadeDestinationValue(t *testing.T, value map[string]any, keys ...string) any {
+	t.Helper()
+	return facadeDestinationMap(t, value, keys[:len(keys)-1]...)[keys[len(keys)-1]]
+}

facade_diff_test.go

@@ -0,0 +1,327 @@
+package configkit_test
+
+import (
+	"errors"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+type facadeDiffConfig struct {
+	Value   string         `yaml:"value"`
+	Count   int            `yaml:"count"`
+	Items   []string       `yaml:"items"`
+	Dynamic map[string]any `yaml:"dynamic"`
+}
+
+type facadeDiffSecretConfig struct {
+	Secret configkit.Secret `yaml:"secret"`
+}
+
+func TestFacadeDiffKeepsPublicSequenceChangesWithHistoricalDescendants(t *testing.T) {
+	type configuration struct {
+		Items []any `yaml:"items"`
+	}
+	load := func(public string) *configkit.Result[configuration] {
+		t.Helper()
+		result, err := configkit.Load(configkit.Bytes("historical-sequence.yml", []byte("items: [{secret: disposable-canary}, "+public+"]\n")),
+			configkit.Options[configuration]{Definition: configkit.Definition[configuration]{
+				Secret: func(path configkit.Path) bool { return path.String() == "/items/0/secret" },
+				Resolve: func(current *configuration, resolution *configkit.Resolution) error {
+					current.Items[0] = "public scalar"
+					resolution.Record(configkit.Keys("items"), configkit.ChangeNormalization, "normalize first item")
+					return nil
+				},
+			}})
+		if err != nil {
+			t.Fatal(err)
+		}
+		return result
+	}
+	left, right := load("1"), load("2")
+	history, ok := left.Explain(configkit.Keys("items").Index(0).Key("secret"))
+	if !ok || history.Effective.Present || len(history.History) == 0 {
+		t.Fatalf("removed historical secret = %#v, known=%v", history, ok)
+	}
+	maskedHistory := false
+	for _, change := range history.History {
+		maskedHistory = maskedHistory || change.Before.Redacted || change.After.Redacted
+	}
+	if !maskedHistory {
+		t.Fatal("historical sensitivity was not retained")
+	}
+	differences, err := configkit.Diff(left, right)
+	if err != nil || len(differences) != 1 || differences[0].Path.String() != "/items" {
+		t.Fatalf("sequence difference = %#v, %v", differences, err)
+	}
+	difference := differences[0]
+	if !difference.Left.Present || !difference.Right.Present || difference.Left.Redacted || difference.Right.Redacted || difference.HiddenChanged {
+		t.Fatalf("public sequence change was concealed: %#v", difference)
+	}
+	if difference.Left.YAML != "- public scalar\n- 1\n" || difference.Right.YAML != "- public scalar\n- 2\n" {
+		t.Fatalf("public sequence snapshots = %#v", difference)
+	}
+}
+
+func TestFacadeDiffKeepsPublicScalarAcrossHistoricalAbsence(t *testing.T) {
+	type configuration struct {
+		Dynamic map[string]any `yaml:"dynamic"`
+	}
+	options := configkit.Options[configuration]{Definition: configkit.Definition[configuration]{
+		Secret: func(path configkit.Path) bool { return path.String() == "/dynamic/entry/token" },
+		Resolve: func(current *configuration, resolution *configkit.Resolution) error {
+			delete(current.Dynamic, "entry")
+			resolution.Record(configkit.Keys("dynamic", "entry"), configkit.ChangeNormalization, "remove entry")
+			return nil
+		},
+	}}
+	absent, err := configkit.Load(configkit.Bytes("removed-entry.yml", []byte("dynamic: {entry: {token: disposable-canary}}\n")), options)
+	if err != nil {
+		t.Fatal(err)
+	}
+	options.Definition.Resolve = nil
+	present, err := configkit.Load(configkit.Bytes("public-entry.yml", []byte("dynamic: {entry: public}\n")), options)
+	if err != nil {
+		t.Fatal(err)
+	}
+	for _, test := range []struct {
+		name        string
+		left, right *configkit.Result[configuration]
+		leftPresent bool
+	}{
+		{name: "addition", left: absent, right: present},
+		{name: "removal", left: present, right: absent, leftPresent: true},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			differences, err := configkit.Diff(test.left, test.right)
+			if err != nil || len(differences) != 1 || differences[0].Path.String() != "/dynamic/entry" {
+				t.Fatalf("entry difference = %#v, %v", differences, err)
+			}
+			difference := differences[0]
+			if difference.Left.Present != test.leftPresent || difference.Right.Present == test.leftPresent ||
+				difference.Left.Redacted || difference.Right.Redacted || difference.HiddenChanged {
+				t.Fatalf("public presence change was concealed: %#v", difference)
+			}
+			wantLeft, wantRight := "", "public\n"
+			if test.leftPresent {
+				wantLeft, wantRight = wantRight, wantLeft
+			}
+			if difference.Left.YAML != wantLeft || difference.Right.YAML != wantRight {
+				t.Fatalf("entry snapshots = %#v", difference)
+			}
+		})
+	}
+}
+
+func TestFacadeDiffReturnsDetachedCanonicalRecords(t *testing.T) {
+	left, err := configkit.Load[facadeDiffConfig](configkit.Bytes("left.yml", []byte("value: before\ncount: 0\nitems: [one, two]\ndynamic: {}\n")))
+	if err != nil {
+		t.Fatalf("load left: %v", err)
+	}
+	right, err := configkit.Load[facadeDiffConfig](configkit.Bytes("right.yml", []byte("value: after\ncount: 0\nitems: [one, three]\ndynamic:\n  added: null\n")))
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	differences, err := configkit.Diff(left, right)
+	if err != nil {
+		t.Fatal(err)
+	}
+	wantPaths := []string{"/dynamic/added", "/items", "/value"}
+	if len(differences) != len(wantPaths) {
+		t.Fatalf("differences = %#v", differences)
+	}
+	for index, difference := range differences {
+		if got := difference.Path.String(); got != wantPaths[index] {
+			t.Errorf("path[%d] = %q, want %q", index, got, wantPaths[index])
+		}
+	}
+	if differences[0].Left.Present || !differences[0].Right.Present || differences[0].Right.YAML != "null\n" {
+		t.Errorf("null/absence snapshots = %#v / %#v", differences[0].Left, differences[0].Right)
+	}
+	if differences[1].Left.YAML != "- one\n- two\n" || differences[1].Right.YAML != "- one\n- three\n" {
+		t.Errorf("sequence snapshots = %#v / %#v", differences[1].Left, differences[1].Right)
+	}
+
+	left.Config.Value = "caller mutation"
+	left.Config.Items[0] = "caller mutation"
+	right.Config.Dynamic["added"] = "caller mutation"
+	differences[2].Left.YAML = "record mutation"
+	differences[2].Path = configkit.Keys("rewritten")
+
+	again, err := configkit.Diff(left, right)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if again[2].Path.String() != "/value" || again[2].Left.YAML != "before\n" || again[1].Left.YAML != "- one\n- two\n" {
+		t.Fatalf("repeat diff changed = %#v", again)
+	}
+	explanation, ok := left.Explain(configkit.Keys("value"))
+	if !ok || explanation.Effective.YAML != "before\n" {
+		t.Fatalf("frozen explanation = %#v, %v", explanation, ok)
+	}
+}
+
+func TestFacadeDiffRejectsUnavailableResults(t *testing.T) {
+	type config struct {
+		Value string `yaml:"value"`
+	}
+	loaded, err := configkit.Load[config](configkit.Bytes("valid.yml", []byte("value: valid\n")))
+	if err != nil {
+		t.Fatal(err)
+	}
+	var nilResult *configkit.Result[config]
+	zeroResult := &configkit.Result[config]{Config: config{Value: "must not be inspected"}}
+	for name, left := range map[string]*configkit.Result[config]{"nil": nilResult, "zero": zeroResult} {
+		t.Run(name, func(t *testing.T) {
+			if records, err := configkit.Diff(left, loaded); err == nil || records != nil {
+				t.Fatalf("Diff = %#v, %v; want nil records and error", records, err)
+			}
+		})
+	}
+	if records, err := configkit.Diff(loaded, nilResult); err == nil || records != nil {
+		t.Fatalf("right nil Diff = %#v, %v", records, err)
+	}
+}
+
+func TestFacadeDiffIgnoresSourceAndResolutionHistory(t *testing.T) {
+	type config struct {
+		Count int `yaml:"count"`
+	}
+	defaults := configkit.Options[config]{Definition: configkit.Definition[config]{
+		Defaults: func() config { return config{Count: 7} },
+	}}
+	defaulted, err := configkit.Load[config](configkit.NoSource(), defaults)
+	if err != nil {
+		t.Fatal(err)
+	}
+	explicit, err := configkit.Load[config](configkit.Bytes("explicit.yml", []byte("count: 7\n")))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if differences, err := configkit.Diff(defaulted, explicit); err != nil || len(differences) != 0 {
+		t.Fatalf("default/source difference = %#v, %v", differences, err)
+	}
+
+	zeroDefault, err := configkit.Load[config](configkit.NoSource(), configkit.Options[config]{Definition: configkit.Definition[config]{Defaults: func() config { return config{} }}})
+	if err != nil {
+		t.Fatal(err)
+	}
+	explicitZero, err := configkit.Load[config](configkit.Bytes("explicit-zero.yml", []byte("count: 0\n")))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if differences, err := configkit.Diff(zeroDefault, explicitZero); err != nil || len(differences) != 0 {
+		t.Fatalf("explicit/defaulted zero difference = %#v, %v", differences, err)
+	}
+
+	resolve := func(value string) configkit.Options[config] {
+		return configkit.Options[config]{Definition: configkit.Definition[config]{Resolve: func(current *config, resolution *configkit.Resolution) error {
+			current.Count = len(value) + 8
+			resolution.Record(configkit.Keys("count"), configkit.ChangeNormalization, "set effective count")
+			return nil
+		}}}
+	}
+	resolvedA, err := configkit.Load[config](configkit.Bytes("a.yml", []byte("{}\n")), resolve("a"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	resolvedB, err := configkit.Load[config](configkit.Bytes("b.yml", []byte("{}\n")), resolve("b"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if differences, err := configkit.Diff(resolvedA, resolvedB); err != nil || len(differences) != 0 {
+		t.Fatalf("equal resolved values = %#v, %v", differences, err)
+	}
+}
+
+func TestFacadeDiffUsesFrozenFilesAndRunsCallbacksOnlyDuringLoad(t *testing.T) {
+	type config struct {
+		Value string `yaml:"value"`
+	}
+	calls := 0
+	options := configkit.Options[config]{Definition: configkit.Definition[config]{Defaults: func() config {
+		calls++
+		return config{}
+	}}}
+	dir := t.TempDir()
+	paths := []string{filepath.Join(dir, "left.yml"), filepath.Join(dir, "right.yml")}
+	for index, contents := range []string{"value: left\n", "value: right\n"} {
+		if err := os.WriteFile(paths[index], []byte(contents), 0o600); err != nil {
+			t.Fatal(err)
+		}
+	}
+	left, err := configkit.Load[config](configkit.File(paths[0]), options)
+	if err != nil {
+		t.Fatal(err)
+	}
+	right, err := configkit.Load[config](configkit.File(paths[1]), options)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if calls != 2 {
+		t.Fatalf("defaults calls after loads = %d", calls)
+	}
+	for _, path := range paths {
+		if err := os.Remove(path); err != nil {
+			t.Fatal(err)
+		}
+	}
+	differences, err := configkit.Diff(left, right)
+	if err != nil || len(differences) != 1 || differences[0].Path.String() != "/value" {
+		t.Fatalf("Diff after removing files = %#v, %v", differences, err)
+	}
+	if calls != 2 {
+		t.Fatalf("defaults reran during Diff: %d calls", calls)
+	}
+}
+
+func TestFacadeDiffCombinesSecretSensitivityAndLoadErrors(t *testing.T) {
+	t.Setenv("FACADE_DIFF_SECRET", "credential-one")
+	secretOptions := configkit.Options[facadeDiffSecretConfig]{Definition: configkit.Definition[facadeDiffSecretConfig]{Secret: func(path configkit.Path) bool {
+		return path.String() == "/secret"
+	}}}
+	left, err := configkit.Load[facadeDiffSecretConfig](configkit.Bytes("left.yml", []byte("secret: FACADE_DIFF_SECRET\n")), secretOptions)
+	if err != nil {
+		t.Fatal(err)
+	}
+	t.Setenv("FACADE_DIFF_SECRET", "credential-two")
+	right, err := configkit.Load[facadeDiffSecretConfig](configkit.Bytes("right.yml", []byte("secret: FACADE_DIFF_SECRET\n")), secretOptions)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if differences, err := configkit.Diff(left, right); err != nil || len(differences) != 0 {
+		t.Fatalf("credential rotation = %#v, %v", differences, err)
+	}
+	t.Setenv("OTHER_FACADE_DIFF_SECRET", "other-credential")
+	otherReference, err := configkit.Load[facadeDiffSecretConfig](configkit.Bytes("other.yml", []byte("secret: OTHER_FACADE_DIFF_SECRET\n")), secretOptions)
+	if err != nil {
+		t.Fatal(err)
+	}
+	differences, err := configkit.Diff(left, otherReference)
+	if err != nil || len(differences) != 1 || !differences[0].HiddenChanged || !differences[0].Left.Redacted || !differences[0].Right.Redacted {
+		t.Fatalf("reference change = %#v, %v", differences, err)
+	}
+	if strings.Contains(differences[0].Left.YAML+differences[0].Right.YAML, "credential") {
+		t.Fatalf("credential leaked: %#v", differences)
+	}
+
+	invalidLeft, leftErr := configkit.Load[facadeDiffSecretConfig](configkit.Bytes("invalid-left.yml", []byte("unknown: true\n")))
+	if invalidLeft != nil || leftErr == nil {
+		t.Fatalf("invalid left load = %#v, %v", invalidLeft, leftErr)
+	}
+	var loadError *configkit.LoadError
+	if !errors.As(leftErr, &loadError) {
+		t.Fatalf("left load error = %T, want *configkit.LoadError", leftErr)
+	}
+	invalidRight, rightErr := configkit.Load[facadeDiffSecretConfig](configkit.Bytes("invalid-right.yml", []byte("unknown: true\n")))
+	if invalidRight != nil || rightErr == nil {
+		t.Fatalf("invalid right load = %#v, %v", invalidRight, rightErr)
+	}
+	if !errors.As(rightErr, &loadError) {
+		t.Fatalf("right load error = %T, want *configkit.LoadError", rightErr)
+	}
+}

facade_discovery_test.go

@@ -0,0 +1,316 @@
+package configkit
+
+import (
+	"errors"
+	"os"
+	"path/filepath"
+	"reflect"
+	"strings"
+	"testing"
+	"testing/fstest"
+)
+
+type facadeDiscoveryConfig struct {
+	Value    string            `yaml:"value"`
+	Settings map[string]string `yaml:"settings"`
+}
+
+type facadeDiscoveryLifecycleConfig struct {
+	Value    string `yaml:"value"`
+	Resolved string `yaml:"resolved"`
+}
+
+type facadeDiscoveryPlacementConfig struct {
+	Items []facadeDiscoveryPlacementItem `yaml:"items"`
+}
+
+type facadeDiscoveryPlacementItem struct {
+	Name string `yaml:"name"`
+}
+
+func TestFacadeSearchPresetUsesCurrentWorkingDirectory(t *testing.T) {
+	original, err := os.Getwd()
+	if err != nil {
+		t.Fatal(err)
+	}
+	root := t.TempDir()
+	if err := os.Chdir(root); err != nil {
+		t.Fatal(err)
+	}
+	t.Cleanup(func() {
+		if err := os.Chdir(original); err != nil {
+			t.Errorf("restore working directory: %v", err)
+		}
+	})
+	writeFacadeImportFile(t, root, "myapp.yml", "value: working\n")
+
+	result, err := Load[facadeDiscoveryConfig](SearchConfig("myapp", WorkingDirectory()))
+	if err != nil || result.Config.Value != "working" {
+		t.Fatalf("working-directory search = %#v, %v", result, err)
+	}
+	if sources := result.Sources(); len(sources) != 1 || sources[0].Path != canonicalFacadePath(t, filepath.Join(root, "myapp.yml")) {
+		t.Fatalf("working-directory sources = %#v", sources)
+	}
+}
+
+func TestFacadeSearchCustomGenericNamesAndEmptyFilenameSelection(t *testing.T) {
+	t.Parallel()
+
+	root := t.TempDir()
+	filename := writeFacadeImportFile(t, root, "config.yml", "value: found\n")
+	custom := SearchDirectory(root, DirectorySearchOptions{
+		Filenames: []FilenameGroup{CustomFilenames("config.yml")},
+	})
+	result, err := Load[facadeDiscoveryConfig](SearchConfig("", custom))
+	if err != nil || result.Config.Value != "found" {
+		t.Fatalf("custom generic-only search with empty slug = %#v, %v", result, err)
+	}
+	if sources := result.Sources(); len(sources) != 1 || sources[0].Path != canonicalFacadePath(t, filename) {
+		t.Fatalf("custom generic-only sources = %#v", sources)
+	}
+
+	defaults := func() facadeDiscoveryConfig { return facadeDiscoveryConfig{Value: "default"} }
+	empty := SearchDirectory(root, DirectorySearchOptions{Filenames: []FilenameGroup{}})
+	absent, err := Load(SearchConfig("myapp", empty), Options[facadeDiscoveryConfig]{
+		Definition: Definition[facadeDiscoveryConfig]{Defaults: defaults},
+	})
+	if err != nil || absent.Config.Value != "default" || len(absent.Sources()) != 0 {
+		t.Fatalf("explicit empty filename selection = %#v, %v", absent, err)
+	}
+	omitted, err := Load[facadeDiscoveryConfig](SearchConfig("myapp", SearchDirectory(root)))
+	if err != nil || omitted.Config.Value != "found" {
+		t.Fatalf("omitted filename selection did not use defaults = %#v, %v", omitted, err)
+	}
+}
+
+func TestFacadeSearchOrderingRequirednessAndSnapshots(t *testing.T) {
+	t.Parallel()
+
+	root := t.TempDir()
+	selectedPath := writeFacadeImportFile(t, root, "config.yml", "value: selected\n")
+	laterPath := writeFacadeImportFile(t, root, "later.yml", "value: later\n")
+	paths := []string{filepath.Join(root, "missing.yml")}
+	names := []string{"config.yml"}
+	groups := []FilenameGroup{CustomFilenames(names...)}
+	directoryOptions := []DirectorySearchOptions{{Filenames: groups}}
+	search := Searches(Paths(paths...), SearchDirectory(root, directoryOptions...), Paths(laterPath))
+	source := SearchConfig("", search, SearchOptions{Required: true})
+
+	paths[0] = laterPath
+	names[0] = "later.yml"
+	groups[0] = CustomFilenames("later.yml")
+	directoryOptions[0].Filenames[0] = CustomFilenames("later.yml")
+	search = Paths(laterPath)
+
+	result, err := Load[facadeDiscoveryConfig](source)
+	if err != nil || result.Config.Value != "selected" {
+		t.Fatalf("ordered required search after caller mutation = %#v, %v", result, err)
+	}
+	if sources := result.Sources(); len(sources) != 1 || sources[0].Path != canonicalFacadePath(t, selectedPath) {
+		t.Fatalf("ordered search sources = %#v", sources)
+	}
+
+	_, err = Load[facadeDiscoveryConfig](SearchConfig("", Paths(filepath.Join(root, "missing.yml")), SearchOptions{Required: true}))
+	var loadErr *LoadError
+	if !errors.As(err, &loadErr) {
+		t.Fatalf("required exhausted search error = %v, want *LoadError", err)
+	}
+	foundMissing := false
+	for _, diagnostic := range loadErr.Diagnostics() {
+		foundMissing = foundMissing || diagnostic.Code == CodeSourceMissing
+	}
+	if !foundMissing {
+		t.Fatalf("required exhausted search diagnostics = %#v", loadErr.Diagnostics())
+	}
+}
+
+func TestFacadeSearchWorksForLayeredMainAndUnderlay(t *testing.T) {
+	t.Parallel()
+
+	root := t.TempDir()
+	mainPath := writeFacadeImportFile(t, root, "main.yml", "settings: {main: value}\n")
+	underlayPath := writeFacadeImportFile(t, root, "underlay/base.yml", "settings: {underlay: value}\n")
+	result, err := LoadLayered(
+		Input{Source: SearchConfig("", Paths(mainPath))},
+		LayeredOptions[facadeDiscoveryConfig]{Underlays: []Underlay{{Inputs: []Input{{
+			Source: SearchConfig("", SearchDirectory(filepath.Dir(underlayPath), DirectorySearchOptions{
+				Filenames: []FilenameGroup{CustomFilenames("base.yml")},
+			})),
+		}}}}},
+	)
+	if err != nil || !reflect.DeepEqual(result.Config.Settings, map[string]string{"main": "value", "underlay": "value"}) {
+		t.Fatalf("layered search = %#v, %v", result, err)
+	}
+	sources := result.Sources()
+	if len(sources) != 2 || sources[0].Path != canonicalFacadePath(t, mainPath) || sources[0].Layer != LayerBase ||
+		sources[1].Path != canonicalFacadePath(t, underlayPath) || sources[1].Layer != LayerUnderlay {
+		t.Fatalf("layered search sources = %#v", sources)
+	}
+}
+
+func TestFacadeSearchSelectedAndAbsentUseRootOnlyContext(t *testing.T) {
+	t.Parallel()
+
+	root := t.TempDir()
+	mainPath := writeFacadeImportFile(t, root, "main.yml", "value: main\n")
+	writeFacadeImportFile(t, root, "overlays/testing.yml", "value: overlay\n")
+	filesystem := &FilesystemContext{Root: root}
+	options := Options[facadeDiscoveryConfig]{Overlay: "testing", Filesystem: filesystem}
+
+	selected, err := Load(SearchConfig("", Paths(mainPath)), options)
+	if err != nil || selected.Config.Value != "overlay" {
+		t.Fatalf("selected search with root-only context = %#v, %v", selected, err)
+	}
+	sources := selected.Sources()
+	if len(sources) != 2 || sources[0].Path != canonicalFacadePath(t, mainPath) || sources[0].ID != "source-1" ||
+		sources[1].ID != "source-2" || sources[1].Layer != LayerOverlay {
+		t.Fatalf("selected search source evidence = %#v", sources)
+	}
+	mainContext, ok := selected.SourceContext("source-1")
+	if !ok || mainContext.BaseDir != "." {
+		t.Fatalf("selected file base context = %#v, %v", mainContext, ok)
+	}
+
+	missing := SearchConfig("", Paths(filepath.Join(root, "missing.yml")))
+	searchedAbsent, err := Load(missing, options)
+	if err != nil {
+		t.Fatalf("exhausted search with root-only overlay context: %v", err)
+	}
+	explicitAbsent, err := Load(NoSource(), options)
+	if err != nil {
+		t.Fatalf("NoSource with root-only overlay context: %v", err)
+	}
+	if !reflect.DeepEqual(searchedAbsent.Config, explicitAbsent.Config) ||
+		!reflect.DeepEqual(searchedAbsent.Sources(), explicitAbsent.Sources()) {
+		t.Fatalf("searched absence differs from NoSource: %#v / %#v", searchedAbsent, explicitAbsent)
+	}
+	left, leftOK := searchedAbsent.Metadata().Lookup(Keys("value"))
+	right, rightOK := explicitAbsent.Metadata().Lookup(Keys("value"))
+	if leftOK != rightOK || !reflect.DeepEqual(left, right) || !leftOK {
+		t.Fatalf("searched absence metadata = %#v, %v; NoSource metadata = %#v, %v", left, leftOK, right, rightOK)
+	}
+	if sources := searchedAbsent.Sources(); len(sources) != 1 || sources[0].ID != "source-1" || sources[0].Layer != LayerOverlay {
+		t.Fatalf("absent search created unexpected source evidence = %#v", sources)
+	}
+}
+
+func TestFacadeSearchRejectsSelectedContextWithoutFallback(t *testing.T) {
+	t.Parallel()
+
+	root := t.TempDir()
+	inside := writeFacadeImportFile(t, root, "inside.yml", "value: fallback\n")
+	outsideRoot := t.TempDir()
+	outside := writeFacadeImportFile(t, outsideRoot, "outside.yml", "value: selected\n")
+	search := SearchConfig("", Paths(outside, inside))
+
+	_, err := Load(search, Options[facadeDiscoveryConfig]{Filesystem: &FilesystemContext{Root: root}})
+	var loadErr *LoadError
+	if !errors.As(err, &loadErr) || !hasFacadeDiagnostic(loadErr, CodeImportOutsideRoot) {
+		t.Fatalf("selected outside-root file error = %v, want containment failure", err)
+	}
+
+	_, err = Load(search, Options[facadeDiscoveryConfig]{Filesystem: &FilesystemContext{BaseDir: root, Root: root}})
+	if !errors.As(err, &loadErr) || !hasFacadeDiagnostic(loadErr, CodeOptionsInvalid) {
+		t.Fatalf("selected file with explicit base error = %v, want option failure", err)
+	}
+
+	native := NewFilesystem("native", fstest.MapFS{})
+	_, err = Load(search, Options[facadeDiscoveryConfig]{Filesystem: &FilesystemContext{FS: &native, BaseDir: ".", Root: "."}})
+	if !errors.As(err, &loadErr) || !hasFacadeDiagnostic(loadErr, CodeOptionsInvalid) {
+		t.Fatalf("selected file with native context error = %v, want option failure", err)
+	}
+}
+
+func TestFacadeSearchRunsCallbacksOnceAndKeepsInspectionWithSelectedFile(t *testing.T) {
+	t.Parallel()
+
+	root := t.TempDir()
+	selectedPath := writeFacadeImportFile(t, root, "selected.yml", "value: selected\n")
+	writeFacadeImportFile(t, root, "later.yml", "value: later\n")
+	defaultsCalls, resolveCalls, validateCalls := 0, 0, 0
+	result, err := Load(SearchConfig("", Paths(selectedPath, filepath.Join(root, "later.yml"))), Options[facadeDiscoveryLifecycleConfig]{
+		Definition: Definition[facadeDiscoveryLifecycleConfig]{
+			Defaults: func() facadeDiscoveryLifecycleConfig {
+				defaultsCalls++
+				return facadeDiscoveryLifecycleConfig{}
+			},
+			Resolve: func(configuration *facadeDiscoveryLifecycleConfig, resolution *Resolution) error {
+				resolveCalls++
+				configuration.Resolved = "from:" + configuration.Value
+				resolution.Record(Keys("resolved"), ChangeNormalization, "derive from selected value", Keys("value"))
+				return nil
+			},
+			Validate: func(_ *facadeDiscoveryLifecycleConfig, _ *Validation) { validateCalls++ },
+		},
+	})
+	if err != nil || result.Config.Value != "selected" || result.Config.Resolved != "from:selected" {
+		t.Fatalf("search lifecycle result = %#v, %v", result, err)
+	}
+	if defaultsCalls != 1 || resolveCalls != 1 || validateCalls != 1 {
+		t.Fatalf("search callbacks ran defaults=%d resolve=%d validate=%d times", defaultsCalls, resolveCalls, validateCalls)
+	}
+	sources := result.Sources()
+	if len(sources) != 1 || sources[0].Path != canonicalFacadePath(t, selectedPath) || sources[0].ID != "source-1" {
+		t.Fatalf("selected source evidence = %#v", sources)
+	}
+	valueInfo, found := result.Metadata().Lookup(Keys("value"))
+	if !found || len(valueInfo.Origin.Locations) != 1 || valueInfo.Origin.Locations[0].SourceID != sources[0].ID {
+		t.Fatalf("selected value metadata = %#v, %v", valueInfo, found)
+	}
+	var rendered strings.Builder
+	if err := result.WriteEffective(&rendered); err != nil || !strings.Contains(rendered.String(), "value: selected") || strings.Contains(rendered.String(), "later") {
+		t.Fatalf("effective output = %q, %v", rendered.String(), err)
+	}
+}
+
+func TestFacadeSearchSequenceItemAndAbsentInputPlacement(t *testing.T) {
+	t.Parallel()
+
+	root := t.TempDir()
+	selectedPath := writeFacadeImportFile(t, root, "item.yml", "name: selected\n")
+	options := Options[facadeDiscoveryPlacementConfig]{
+		Destination:   Keys("items"),
+		PlacementMode: PlacementModeSequenceItem,
+	}
+	selected, err := Load(SearchConfig("", Paths(selectedPath)), options)
+	if err != nil || len(selected.Config.Items) != 1 || selected.Config.Items[0].Name != "selected" {
+		t.Fatalf("selected sequence item = %#v, %v", selected, err)
+	}
+	if sources := selected.Sources(); len(sources) != 1 || sources[0].Destination.String() != "/items" || sources[0].PlacementMode != PlacementModeSequenceItem {
+		t.Fatalf("selected sequence-item source = %#v", sources)
+	}
+
+	defaults := func() facadeDiscoveryPlacementConfig {
+		return facadeDiscoveryPlacementConfig{Items: []facadeDiscoveryPlacementItem{}}
+	}
+	absent, err := Load(SearchConfig("", Paths(filepath.Join(root, "missing.yml"))), Options[facadeDiscoveryPlacementConfig]{
+		Definition:    Definition[facadeDiscoveryPlacementConfig]{Defaults: defaults},
+		Destination:   options.Destination,
+		PlacementMode: options.PlacementMode,
+	})
+	if err != nil || len(absent.Config.Items) != 0 || len(absent.Sources()) != 0 {
+		t.Fatalf("absent search created a sequence item: %#v, %v", absent, err)
+	}
+}
+
+func hasFacadeDiagnostic(err *LoadError, code Code) bool {
+	for _, diagnostic := range err.Diagnostics() {
+		if diagnostic.Code == code {
+			return true
+		}
+	}
+	return false
+}
+
+func canonicalFacadePath(t *testing.T, path string) string {
+	t.Helper()
+	absolute, err := filepath.Abs(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	canonical, err := filepath.EvalSymlinks(absolute)
+	if err != nil {
+		t.Fatal(err)
+	}
+	return canonical
+}

facade_equivalence_test.go

@@ -23,6 +23,9 @@ type equivalentConfig struct {
 
 func TestEquivalentSourceFormsPreserveInputSemantics(t *testing.T) {
 	const yaml = "version: 1\nname: retained\nenabled: false\ncount: 0\nitems: []\noptional: null\n"
+	bytesData := []byte(yaml)
+	bytesSource := configkit.Bytes("memory", bytesData)
+	bytesData[0] = 'X'
 	hostPath := filepath.Join(t.TempDir(), "main.yml")
 	if err := os.WriteFile(hostPath, []byte(yaml), 0o600); err != nil {
 		t.Fatal(err)
@@ -37,7 +40,7 @@ func TestEquivalentSourceFormsPreserveInputSemantics(t *testing.T) {
 	}{
 		{"host", configkit.File(hostPath), nil, configkit.SourceKindFile},
 		{"native", namespace.File("main.yml"), nil, configkit.SourceKindFilesystem},
-		{"bytes", configkit.Bytes("memory", []byte(yaml)), context, configkit.SourceKindBytes},
+		{"bytes", bytesSource, context, configkit.SourceKindBytes},
 		{"reader", configkit.Reader("stream", strings.NewReader(yaml)), context, configkit.SourceKindReader},
 		{"mapping", configkit.Mapping("programmatic", map[string]any{
 			"version": uint64(1), "name": "retained", "enabled": false, "count": 0,
@@ -112,6 +115,15 @@ func TestRepeatedImportsAcrossTiersAndReaderConsumption(t *testing.T) {
 	if got := layeredErrorCode(t, err); got != configkit.CodeSourceReaderUsed {
 		t.Fatalf("reader re-consumption = %s", got)
 	}
+	copiedReader := configkit.Reader("copied-single-use", strings.NewReader("name: retained\n"))
+	readerCopy := copiedReader
+	if _, err := configkit.Load[config](readerCopy, configkit.Options[config]{}); err != nil {
+		t.Fatal(err)
+	}
+	_, err = configkit.Load[config](copiedReader, configkit.Options[config]{})
+	if got := layeredErrorCode(t, err); got != configkit.CodeSourceReaderUsed {
+		t.Fatalf("copied reader re-consumption = %s", got)
+	}
 }
 
 func TestReplacedSecretRemainsAbsentFromSafeOutput(t *testing.T) {

facade_final_sequence_identity_test.go

@@ -0,0 +1,249 @@
+package configkit_test
+
+import (
+	"fmt"
+	"strings"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+func TestFinalSequenceIdentityTreatsLateExtendsFieldsAsPayload(t *testing.T) {
+	for _, phase := range []string{"defaults", "resolution"} {
+		for _, component := range []any{"application-data", nil} {
+			for _, duplicate := range []bool{false, true} {
+				t.Run(fmt.Sprintf("%s/%v/duplicate=%t", phase, component, duplicate), func(t *testing.T) {
+					items := []any{map[string]any{"extends": component}}
+					if duplicate {
+						items = append(items, map[string]any{"extends": component})
+					}
+					definition := configkit.Definition[map[string]any]{
+						SequenceIdentities: []configkit.SequenceIdentity{configkit.SequenceIdentityAt(configkit.Keys("items"), "extends")},
+						Defaults: func() map[string]any {
+							if phase == "defaults" {
+								return map[string]any{"items": items}
+							}
+							return map[string]any{}
+						},
+					}
+					if phase == "resolution" {
+						definition.Resolve = func(value *map[string]any, resolution *configkit.Resolution) error {
+							(*value)["items"] = items
+							resolution.Record(configkit.Keys("items"), configkit.ChangeNormalization, "supply application payload")
+							return nil
+						}
+					}
+					loaded, err := configkit.Load(configkit.NoSource(), configkit.Options[map[string]any]{Definition: definition})
+					if duplicate {
+						loadErr := requireFacadeLoadError(t, err)
+						diagnostics := loadErr.Diagnostics()
+						if loaded != nil || len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeSequenceConflict || diagnostics[0].Path.String() != "/items" {
+							t.Fatalf("duplicate effective payload diagnostics = %#v", diagnostics)
+						}
+						return
+					}
+					if err != nil || loaded == nil {
+						t.Fatalf("ordinary effective identity was interpreted as a directive: %v", err)
+					}
+					value := loaded.Config["items"].([]any)[0].(map[string]any)
+					if actual, present := value["extends"]; !present || actual != component {
+						t.Fatalf("effective payload = %#v", value)
+					}
+				})
+			}
+		}
+	}
+}
+
+func TestFinalSequenceIdentityValidatesLateValues(t *testing.T) {
+	identity := configkit.SequenceIdentityAt(configkit.Keys("items"), "id")
+	for _, source := range []string{"defaults", "CLI", "resolution"} {
+		for _, shape := range []string{"duplicate", "missing", "non-scalar"} {
+			t.Run(source+"/"+shape, func(t *testing.T) {
+				items := finalIdentityRecords(shape)
+				definition := configkit.Definition[map[string]any]{
+					SequenceIdentities: []configkit.SequenceIdentity{identity},
+					Defaults: func() map[string]any {
+						if source == "defaults" {
+							return map[string]any{"items": items}
+						}
+						return map[string]any{}
+					},
+				}
+				options := configkit.Options[map[string]any]{Definition: definition}
+				switch source {
+				case "CLI":
+					options.Overrides = []configkit.Override{{Path: configkit.Keys("items"), Value: items, Flag: "--items"}}
+				case "resolution":
+					options.Definition.Resolve = func(value *map[string]any, resolution *configkit.Resolution) error {
+						(*value)["items"] = items
+						resolution.Record(configkit.Keys("items"), configkit.ChangeNormalization, "create effective records")
+						return nil
+					}
+				}
+				result, err := configkit.Load(configkit.NoSource(), options)
+				if result != nil {
+					t.Fatalf("invalid final identities returned a result: %#v", result.Config)
+				}
+				loadErr := requireFacadeLoadError(t, err)
+				if len(loadErr.Diagnostics()) != 1 {
+					t.Fatalf("diagnostics = %#v", loadErr.Diagnostics())
+				}
+				diagnostic := loadErr.Diagnostics()[0]
+				wantCode, wantPath := configkit.CodeSequenceInvalid, "/items/0/id"
+				if shape == "duplicate" {
+					wantCode, wantPath = configkit.CodeSequenceConflict, "/items"
+				}
+				if diagnostic.Code != wantCode || diagnostic.Path.String() != wantPath {
+					t.Fatalf("diagnostic = %#v, want %s at %s", diagnostic, wantCode, wantPath)
+				}
+				if source == "CLI" && (diagnostic.Primary == nil || diagnostic.Primary.SourceID != "source-1") {
+					t.Fatalf("CLI identity diagnostic lost its source: %#v", diagnostic)
+				}
+			})
+		}
+	}
+}
+
+func TestFinalSequenceIdentityMatchesLateDynamicNamespaces(t *testing.T) {
+	selector := configkit.SequenceSelector{
+		configkit.SequenceLiteralKey("collections"),
+		configkit.SequenceAnyMappingKey(),
+		configkit.SequenceLiteralKey("items"),
+	}
+	for _, source := range []string{"defaults", "CLI", "resolution"} {
+		t.Run(source, func(t *testing.T) {
+			name := map[string]string{"defaults": "team", "CLI": "cloud", "resolution": "orbit"}[source]
+			path := configkit.Keys("collections", name, "items")
+			items := finalIdentityRecords("duplicate")
+			definition := configkit.Definition[map[string]any]{
+				SequenceIdentities: []configkit.SequenceIdentity{
+					configkit.SequenceIdentityMatching(selector, "id"),
+				},
+				Defaults: func() map[string]any {
+					if source == "defaults" {
+						return map[string]any{"collections": map[string]any{name: map[string]any{"items": items}}}
+					}
+					return map[string]any{}
+				},
+			}
+			options := configkit.Options[map[string]any]{Definition: definition}
+			switch source {
+			case "CLI":
+				options.Overrides = []configkit.Override{{Path: path, Value: items, Flag: "--items"}}
+			case "resolution":
+				options.Definition.Resolve = func(value *map[string]any, resolution *configkit.Resolution) error {
+					(*value)["collections"] = map[string]any{name: map[string]any{"items": items}}
+					resolution.Record(configkit.Keys("collections"), configkit.ChangeNormalization, "create namespaced records")
+					return nil
+				}
+			}
+			result, err := configkit.Load(configkit.NoSource(), options)
+			if result != nil {
+				t.Fatalf("invalid dynamic identity returned a result: %#v", result.Config)
+			}
+			loadErr := requireFacadeLoadError(t, err)
+			if diagnostics := loadErr.Diagnostics(); len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeSequenceConflict || diagnostics[0].Path.String() != path.String() {
+				t.Fatalf("dynamic identity diagnostics = %#v", diagnostics)
+			}
+		})
+	}
+}
+
+func TestFinalSequenceIdentityAppliesToLayeredLoads(t *testing.T) {
+	definition := configkit.Definition[map[string]any]{
+		SequenceIdentities: []configkit.SequenceIdentity{configkit.SequenceIdentityAt(configkit.Keys("items"), "id")},
+		Defaults: func() map[string]any {
+			return map[string]any{"items": finalIdentityRecords("duplicate")}
+		},
+	}
+	result, err := configkit.LoadLayered(configkit.Input{Source: configkit.NoSource()}, configkit.LayeredOptions[map[string]any]{Definition: definition})
+	if result != nil {
+		t.Fatalf("invalid layered final identities returned a result: %#v", result.Config)
+	}
+	loadErr := requireFacadeLoadError(t, err)
+	if diagnostics := loadErr.Diagnostics(); len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeSequenceConflict {
+		t.Fatalf("layered identity diagnostics = %#v", diagnostics)
+	}
+}
+
+func TestFinalSequenceIdentityUsesTypedNumbersAndIgnoresUnmatchedShapes(t *testing.T) {
+	definition := configkit.Definition[map[string]any]{
+		SequenceIdentities: []configkit.SequenceIdentity{
+			configkit.SequenceIdentityAt(configkit.Keys("items"), "id"),
+			configkit.SequenceIdentityAt(configkit.Keys("unmatched"), "id"),
+		},
+	}
+	_, err := configkit.Load(configkit.NoSource(), configkit.Options[map[string]any]{
+		Definition: configkit.Definition[map[string]any]{
+			SequenceIdentities: definition.SequenceIdentities,
+			Defaults: func() map[string]any {
+				return map[string]any{"items": []any{
+					map[string]any{"id": int64(1)}, map[string]any{"id": float64(1)},
+				}}
+			},
+		},
+	})
+	loadErr := requireFacadeLoadError(t, err)
+	if diagnostics := loadErr.Diagnostics(); len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeSequenceConflict {
+		t.Fatalf("typed numeric identity diagnostics = %#v", diagnostics)
+	}
+
+	valid, err := configkit.Load(configkit.Mapping("valid", map[string]any{
+		"items": []any{
+			map[string]any{"id": nil}, map[string]any{"id": true},
+			map[string]any{"id": int64(1)}, map[string]any{"id": "one"},
+		},
+		"other": []any{map[string]any{"id": "same"}, map[string]any{"id": "same"}},
+	}), configkit.Options[map[string]any]{Definition: definition})
+	if err != nil || valid == nil {
+		t.Fatalf("valid scalar identities and unmatched duplicate sequence = %#v, %v", valid, err)
+	}
+	if _, err := configkit.Load(configkit.Mapping("other-shapes", map[string]any{"items": nil}), configkit.Options[map[string]any]{Definition: definition}); err != nil {
+		t.Fatalf("null at a registered path was validated as an identity sequence: %v", err)
+	}
+}
+
+func TestFinalSequenceIdentityDiagnosticDoesNotRevealResolvedSecrets(t *testing.T) {
+	const canary = "FINAL-IDENTITY-CREDENTIAL-CANARY"
+	const secretName = "CONFIGKIT_FINAL_IDENTITY_SECRET"
+	t.Setenv(secretName, canary)
+	type item struct {
+		ID    string           `yaml:"id"`
+		Token configkit.Secret `yaml:"token"`
+	}
+	type config struct {
+		Items []item `yaml:"items"`
+	}
+	definition := configkit.Definition[config]{
+		SequenceIdentities: []configkit.SequenceIdentity{configkit.SequenceIdentityAt(configkit.Keys("items"), "id")},
+		Defaults: func() config {
+			return config{Items: []item{
+				{ID: "duplicate", Token: configkit.SecretEnv(secretName)},
+				{ID: "duplicate", Token: configkit.SecretEnv(secretName)},
+			}}
+		},
+	}
+	result, err := configkit.Load(configkit.NoSource(), configkit.Options[config]{Definition: definition})
+	if result != nil {
+		t.Fatalf("duplicate identity returned a result: %#v", result.Config)
+	}
+	if err == nil || strings.Contains(err.Error(), canary) {
+		t.Fatalf("identity diagnostic leaked a resolved secret: %v", err)
+	}
+	loadErr := requireFacadeLoadError(t, err)
+	if diagnostics := loadErr.Diagnostics(); len(diagnostics) != 1 || diagnostics[0].Code != configkit.CodeSequenceConflict {
+		t.Fatalf("secret-bearing identity diagnostics = %#v", diagnostics)
+	}
+}
+
+func finalIdentityRecords(shape string) []any {
+	switch shape {
+	case "duplicate":
+		return []any{map[string]any{"id": "same"}, map[string]any{"id": "same"}}
+	case "missing":
+		return []any{map[string]any{"other": "value"}}
+	default:
+		return []any{map[string]any{"id": map[string]any{"nested": "value"}}}
+	}
+}

facade_historical_paths_test.go

@@ -0,0 +1,62 @@
+package configkit_test
+
+import (
+	"reflect"
+	"testing"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+func TestPublicHistoricalPathsEnumerateContextualShapes(t *testing.T) {
+	loaded, err := configkit.Load[map[string]any](configkit.Bytes("main", []byte(
+		"retired: {child: 1}\nshape: {'0': base}\n'a/b~': {'': 2}\n",
+	)), configkit.Options[map[string]any]{Overrides: []configkit.Override{
+		{Path: configkit.Keys("retired"), Value: nil, Flag: "--retired"},
+		{Path: configkit.Keys("shape"), Value: []any{"replacement"}, Flag: "--shape"},
+	}})
+	if err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+
+	metadata := loaded.Metadata()
+	paths := metadata.HistoricalPaths()
+	got := make([]string, len(paths))
+	for index, path := range paths {
+		got[index] = path.String()
+		if len(metadata.History(path)) == 0 {
+			t.Errorf("enumerated path %s has no history", path)
+		}
+	}
+	want := []string{"", "/a~1b~0", "/a~1b~0/", "/retired", "/retired/child", "/shape", "/shape/0"}
+	if !reflect.DeepEqual(got, want) {
+		t.Fatalf("historical paths = %v, want sorted unique pointers %v", got, want)
+	}
+	if _, ok := metadata.Lookup(paths[4]); ok {
+		t.Fatal("removed descendant remains effective")
+	}
+	if _, ok := metadata.Lookup(paths[6]); !ok {
+		t.Fatal("enumerated numeric path did not follow the effective sequence shape")
+	}
+	if len(metadata.History(paths[6])) == 0 {
+		t.Fatal("enumerated numeric path lost history")
+	}
+	if _, ok := metadata.Lookup(configkit.Keys("shape", "0")); ok {
+		t.Fatal("mapping-key path resolved against the effective sequence")
+	}
+	if len((configkit.Metadata{}).HistoricalPaths()) != 0 {
+		t.Fatal("zero metadata contains historical paths")
+	}
+
+	paths[1] = configkit.Keys("mutated")
+	if again := metadata.HistoricalPaths(); !reflect.DeepEqual(pathStrings(again), want) {
+		t.Fatalf("returned path slice mutated retained metadata: %v", pathStrings(again))
+	}
+}
+
+func pathStrings(paths []configkit.Path) []string {
+	result := make([]string, len(paths))
+	for index, path := range paths {
+		result[index] = path.String()
+	}
+	return result
+}

facade_import_alias_test.go

@@ -0,0 +1,99 @@
+package configkit_test
+
+import (
+	"errors"
+	"io/fs"
+	"os"
+	"path"
+	"path/filepath"
+	"testing"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+func TestFacadeNativeDiscoveryPreservesAliasIdentity(t *testing.T) {
+	root := t.TempDir()
+	for name, contents := range map[string]string{
+		"data/item.yml":        "configkit:\n  imports:\n    - path: helper.yml\nvalue: item\n",
+		"data/helper.yml":      "owner: target\nresource: resource.txt\n",
+		"aliases/helper.yml":   "owner: alias\nresource: resource.txt\n",
+		"data/resource.txt":    "target resource",
+		"aliases/resource.txt": "alias resource",
+	} {
+		filename := filepath.Join(root, filepath.FromSlash(name))
+		if err := os.MkdirAll(filepath.Dir(filename), 0o700); err != nil {
+			t.Fatal(err)
+		}
+		if err := os.WriteFile(filename, []byte(contents), 0o600); err != nil {
+			t.Fatal(err)
+		}
+	}
+	for name, target := range map[string]string{"aliases/item.yml": "../data/item.yml", "linked": "data"} {
+		if err := os.Symlink(target, filepath.Join(root, filepath.FromSlash(name))); err != nil {
+			t.Skipf("symlinks unavailable: %v", err)
+		}
+	}
+	filesystem := configkit.NewFilesystem("aliases", os.DirFS(root))
+	type configuration struct {
+		Value    string `yaml:"value"`
+		Owner    string `yaml:"owner"`
+		Resource string `yaml:"resource"`
+	}
+	for _, test := range []struct {
+		name, declaration, item, helper, owner, resource string
+	}{
+		{"file/path", "path: aliases/item.yml", "aliases/item.yml", "aliases/helper.yml", "alias", "alias resource"},
+		{"file/dir", "dir: aliases\n      pattern: item.yml", "aliases/item.yml", "aliases/helper.yml", "alias", "alias resource"},
+		{"file/glob", "glob: 'aliases/ite*.yml'", "aliases/item.yml", "aliases/helper.yml", "alias", "alias resource"},
+		{"directory/path", "path: linked/item.yml", "linked/item.yml", "linked/helper.yml", "target", "target resource"},
+		{"directory/dir", "dir: linked\n      pattern: item.yml", "linked/item.yml", "linked/helper.yml", "target", "target resource"},
+		{"directory/glob", "glob: 'lin*/ite*.yml'", "linked/item.yml", "linked/helper.yml", "target", "target resource"},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			result, err := configkit.Load(configkit.Bytes("main", []byte("configkit:\n  imports:\n    - "+test.declaration+"\n")), configkit.Options[configuration]{
+				Imports: true, Filesystem: &configkit.FilesystemContext{FS: &filesystem, Root: ".", BaseDir: "."},
+			})
+			if err != nil {
+				t.Fatal(err)
+			}
+			if result.Config.Owner != test.owner || result.Config.Value != "item" {
+				t.Fatalf("configuration = %#v", result.Config)
+			}
+			sources := result.Sources()
+			if len(sources) != 3 || sources[1].Path != test.item || sources[2].Path != test.helper {
+				t.Fatalf("source identities = %#v", sources)
+			}
+			for field, source := range map[string]configkit.SourceInfo{"value": sources[1], "resource": sources[2]} {
+				info, ok := result.Metadata().Lookup(configkit.Keys(field))
+				if !ok || len(info.Origin.Locations) != 1 || info.Origin.Locations[0].SourceID != source.ID {
+					t.Fatalf("%s origin = %#v", field, info)
+				}
+				context, ok := result.SourceContext(source.ID)
+				if !ok || context.BaseDir != path.Dir(source.Path) {
+					t.Fatalf("%s context = %#v", field, context)
+				}
+				contents, err := fs.ReadFile(context.FS, path.Join(context.BaseDir, result.Config.Resource))
+				if err != nil || string(contents) != test.resource {
+					t.Fatalf("%s resource = %q, %v", field, contents, err)
+				}
+			}
+		})
+	}
+	// The same native path is still a repeat when separately included. An alias
+	// to a leaf file remains a different native namespace path, as for path imports.
+	for _, selector := range []string{"path: aliases/item.yml", "dir: aliases\n      pattern: item.yml", "glob: 'aliases/ite*.yml'"} {
+		_, err := configkit.Load(configkit.Bytes("main", []byte("configkit:\n  imports:\n    - "+selector+"\n    - path: aliases/item.yml\n")), configkit.Options[configuration]{
+			Imports: true, Filesystem: &configkit.FilesystemContext{FS: &filesystem, Root: ".", BaseDir: "."},
+		})
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) || len(loadError.Diagnostics()) != 1 || loadError.Diagnostics()[0].Code != configkit.CodeImportRepeated {
+			t.Fatalf("repeated source: %v", err)
+		}
+		_, err = configkit.Load(configkit.Bytes("main", []byte("configkit:\n  imports:\n    - "+selector+"\n    - path: data/item.yml\n")), configkit.Options[configuration]{
+			Imports: true, Filesystem: &configkit.FilesystemContext{FS: &filesystem, Root: ".", BaseDir: "."},
+		})
+		if !errors.As(err, &loadError) || len(loadError.Diagnostics()) == 0 || loadError.Diagnostics()[0].Code != configkit.CodeComposeConflict {
+			t.Fatalf("distinct native alias paths should reach composition: %v", err)
+		}
+	}
+}

facade_import_defaults_test.go

@@ -0,0 +1,299 @@
+package configkit
+
+import (
+	"errors"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+func TestImportDepthChecksSuffixAfterSaturation(t *testing.T) {
+	for _, test := range []struct {
+		depth string
+		valid bool
+	}{
+		{"999999999999999999999999.5", false},
+		{"999999999999999999999999e1", false},
+		{"999999999999999999999999", true},
+		{"-0", true},
+	} {
+		t.Run(test.depth, func(t *testing.T) {
+			main := writeFacadeImportFile(t, t.TempDir(), "main.yml", "value: true\nconfigkit:\n  imports:\n    - dir: absent\n      recursive: true\n      depth: "+test.depth+"\n")
+			result, err := Load[map[string]any](File(main), Options[map[string]any]{Imports: true})
+			if test.valid {
+				if err != nil || result.Config["value"] != true {
+					t.Fatalf("valid depth: %v", err)
+				}
+				return
+			}
+			var loadError *LoadError
+			if !errors.As(err, &loadError) || len(loadError.Diagnostics()) != 1 {
+				t.Fatalf("invalid depth accepted: %v", err)
+			}
+			if diagnostic := loadError.Diagnostics()[0]; diagnostic.Code != CodeImportInvalid || diagnostic.Path.String() != "/configkit/imports/0/depth" {
+				t.Fatalf("invalid depth diagnostic = %#v", diagnostic)
+			}
+		})
+	}
+}
+
+func TestImportDefaultsApplyToTransitiveAndUnderlayImports(t *testing.T) {
+	type config struct {
+		Main        string `yaml:"main"`
+		Transitive  string `yaml:"transitive"`
+		Exception   string `yaml:"exception"`
+		Underlay    string `yaml:"underlay"`
+		IgnoredDeep string `yaml:"ignored_deep"`
+	}
+	directory := t.TempDir()
+	mainPath := writeFacadeImportFile(t, directory, "main.yml", `configkit:
+  imports:
+    - path: parent/parent.yml
+main: main
+`)
+	writeFacadeImportFile(t, directory, "parent/parent.yml", `configkit:
+  imports:
+    - dir: generated
+    - dir: explicit
+      pattern: '*.yaml'
+      recursive: false
+`)
+	transitivePath := writeFacadeImportFile(t, directory, "parent/generated/deep/transitive.cfg", "transitive: selected\n")
+	writeFacadeImportFile(t, directory, "parent/generated/ignored.yaml", "ignored_deep: wrong-pattern\n")
+	explicitPath := writeFacadeImportFile(t, directory, "parent/explicit/exception.yaml", "exception: declaration\n")
+	writeFacadeImportFile(t, directory, "parent/explicit/deep/ignored.yaml", "ignored_deep: wrong-depth\n")
+	writeFacadeImportFile(t, directory, "parent/explicit/ignored.cfg", "ignored_deep: wrong-explicit-pattern\n")
+	underlayPath := writeFacadeImportFile(t, directory, "underlay.yml", `configkit:
+  imports:
+    - dir: underlay-files
+`)
+	underlayImportPath := writeFacadeImportFile(t, directory, "underlay-files/underlay.cfg", "underlay: selected\n")
+	recursive := true
+	loaded, err := LoadLayered(Input{Source: File(mainPath), Imports: true}, LayeredOptions[config]{
+		ImportDefaults: ImportDefaults{Recursive: &recursive, Pattern: []string{"*.cfg"}},
+		Underlays:      []Underlay{{Inputs: []Input{{Source: File(underlayPath), Imports: true}}}},
+	})
+	if err != nil {
+		t.Fatalf("LoadLayered: %v", err)
+	}
+	if loaded.Config.Main != "main" || loaded.Config.Transitive != "selected" || loaded.Config.Exception != "declaration" || loaded.Config.Underlay != "selected" || loaded.Config.IgnoredDeep != "" {
+		t.Fatalf("configuration = %#v", loaded.Config)
+	}
+	paths := make(map[string]bool)
+	for _, source := range loaded.Sources() {
+		paths[source.Path] = true
+	}
+	for _, selected := range []string{transitivePath, explicitPath, underlayImportPath} {
+		if !paths[canonicalFacadeImportPath(t, selected)] {
+			t.Errorf("selected source %q missing from source records", selected)
+		}
+	}
+}
+
+func TestImportDefaultsPreserveEmptyPatternAndRequiredness(t *testing.T) {
+	type config struct {
+		Value string `yaml:"value"`
+	}
+	directory := t.TempDir()
+	mainPath := writeFacadeImportFile(t, directory, "main.yml", `configkit:
+  imports:
+    - dir: snippets
+`)
+	writeFacadeImportFile(t, directory, "snippets/value.cfg", "value: imported\n")
+
+	unrestricted, err := Load(File(mainPath), Options[config]{Imports: true})
+	if err != nil || unrestricted.Config.Value != "imported" {
+		t.Fatalf("unspecified pattern = %#v, %v", unrestricted, err)
+	}
+	empty, err := Load(File(mainPath), Options[config]{Imports: true, ImportDefaults: ImportDefaults{Pattern: []string{}}})
+	if err != nil || empty.Config.Value != "" {
+		t.Fatalf("empty pattern = %#v, %v", empty, err)
+	}
+	emptySources := empty.Sources()
+	if len(emptySources) != 1 || len(emptySources[0].ImportSelections) != 1 || emptySources[0].ImportSelections[0].Path.String() != "/configkit/imports/0" || emptySources[0].ImportSelections[0].Selector != ImportSelectorDirectory || emptySources[0].ImportSelections[0].Outcome != ImportSelectionNone {
+		t.Fatalf("empty-pattern evidence = %#v", emptySources)
+	}
+	required := true
+	_, err = Load(File(mainPath), Options[config]{Imports: true, ImportDefaults: ImportDefaults{Required: &required, Pattern: []string{}}})
+	var loadError *LoadError
+	if !errors.As(err, &loadError) || len(loadError.Diagnostics()) != 1 || loadError.Diagnostics()[0].Code != CodeImportMissing {
+		t.Fatalf("required empty selection error = %T %v", err, err)
+	}
+	if sources := loadError.Sources(); len(sources) != 1 || len(sources[0].ImportSelections) != 0 {
+		t.Fatalf("required empty selection was recorded as optional: %#v", sources)
+	}
+}
+
+func TestImportDefaultsAreSnapshottedBeforeOrdinaryAndLayeredCallbacks(t *testing.T) {
+	type config struct {
+		Version  string `yaml:"version"`
+		Imported string `yaml:"imported"`
+	}
+	for _, layered := range []bool{false, true} {
+		t.Run(map[bool]string{false: "ordinary", true: "layered"}[layered], func(t *testing.T) {
+			directory := t.TempDir()
+			mainPath := writeFacadeImportFile(t, directory, "main.yml", `version: v1
+configkit:
+  imports:
+    - dir: parts
+`)
+			writeFacadeImportFile(t, directory, "parts/nested/value.cfg", "imported: selected\n")
+
+			recursive := true
+			patterns := []string{"*.cfg"}
+			definition := Definition[config]{Version: &VersionRule{
+				Path: Keys("version"),
+				Check: func(version Version, _ *Validation) error {
+					if version != (Version{Present: true, Text: "v1"}) {
+						t.Errorf("version callback received %#v", version)
+					}
+					recursive = false
+					patterns[0] = "*.skip"
+					return nil
+				},
+			}}
+			load := func() (*Result[config], error) {
+				defaults := ImportDefaults{Recursive: &recursive, Pattern: patterns}
+				if layered {
+					return LoadLayered(Input{Source: File(mainPath), Imports: true}, LayeredOptions[config]{
+						Definition: definition, SourceVersionMode: SourceVersionModeIfPresent, ImportDefaults: defaults,
+					})
+				}
+				return Load(File(mainPath), Options[config]{
+					Definition: definition, SourceVersionMode: SourceVersionModeIfPresent, Imports: true, ImportDefaults: defaults,
+				})
+			}
+
+			first, err := load()
+			if err != nil || first.Config.Imported != "selected" {
+				t.Fatalf("load with pre-callback defaults = %#v, %v", first, err)
+			}
+			if recursive || patterns[0] != "*.skip" {
+				t.Fatalf("version callback did not mutate caller defaults: recursive=%v pattern=%q", recursive, patterns[0])
+			}
+
+			second, err := load()
+			if err != nil || second.Config.Imported != "" {
+				t.Fatalf("subsequent load reused earlier defaults: %#v, %v", second, err)
+			}
+		})
+	}
+}
+
+func TestInvalidImportDefaultsPreflightBeforeOrdinaryAndLayeredReads(t *testing.T) {
+	type config struct {
+		Value string `yaml:"value"`
+	}
+	for _, layered := range []bool{false, true} {
+		input := strings.NewReader("value: read\n")
+		underlayInput := strings.NewReader("value: underlay\n")
+		defaultsCalled := false
+		resolveCalled := false
+		definition := Definition[config]{
+			Defaults: func() config { defaultsCalled = true; return config{} },
+			Resolve:  func(*config, *Resolution) error { resolveCalled = true; return nil },
+		}
+		defaults := ImportDefaults{Pattern: []string{"["}}
+		var err error
+		if layered {
+			_, err = LoadLayered(Input{Source: Reader("reader.yml", input)}, LayeredOptions[config]{
+				Definition: definition, ImportDefaults: defaults,
+				Underlays: []Underlay{{Inputs: []Input{{Source: Reader("underlay.yml", underlayInput)}}}},
+			})
+		} else {
+			_, err = Load(Reader("reader.yml", input), Options[config]{Definition: definition, ImportDefaults: defaults})
+		}
+		var loadError *LoadError
+		if !errors.As(err, &loadError) || len(loadError.Diagnostics()) != 1 || loadError.Diagnostics()[0].Code != CodeOptionsInvalid {
+			if loadError != nil {
+				t.Fatalf("layered=%v diagnostics = %#v", layered, loadError.Diagnostics())
+			}
+			t.Fatalf("layered=%v error = %T %v", layered, err, err)
+		}
+		if input.Len() != len("value: read\n") || underlayInput.Len() != len("value: underlay\n") || defaultsCalled || resolveCalled {
+			t.Fatalf("layered=%v consumed input or called application hooks: remaining=%d underlay=%d defaults=%v resolve=%v", layered, input.Len(), underlayInput.Len(), defaultsCalled, resolveCalled)
+		}
+	}
+}
+
+func TestPublicImportSelectionSnapshotsAreDetachedAcrossResultErrorAndResolution(t *testing.T) {
+	type config struct {
+		Value string `yaml:"value"`
+	}
+	directory := t.TempDir()
+	mainPath := writeFacadeImportFile(t, directory, "main.yml", `configkit:
+  imports:
+    - dir: absent
+value: loaded
+`)
+	callbackPath := ""
+	loaded, err := Load(File(mainPath), Options[config]{Imports: true, Definition: Definition[config]{Resolve: func(_ *config, resolution *Resolution) error {
+		first, ok := resolution.Source("source-1")
+		if !ok || len(first.ImportSelections) != 1 {
+			t.Fatalf("resolution source = %#v, %v", first, ok)
+		}
+		first.ImportSelections[0].Path = Keys("callback mutation")
+		second, ok := resolution.Source("source-1")
+		if !ok || second.ImportSelections[0].Path.String() != "/configkit/imports/0" {
+			t.Fatalf("resolution snapshot changed after mutation: %#v", second)
+		}
+		callbackPath = second.ImportSelections[0].Path.String()
+		return nil
+	}}})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if callbackPath != "/configkit/imports/0" {
+		t.Fatalf("callback path = %q", callbackPath)
+	}
+	sources := loaded.Sources()
+	if len(sources) != 1 || len(sources[0].ImportSelections) != 1 || sources[0].ImportSelections[0].Location.SourceID != sources[0].ID || sources[0].ImportSelections[0].Selector != ImportSelectorDirectory {
+		t.Fatalf("result source evidence = %#v", sources)
+	}
+	sources[0].ImportSelections[0].Path = Keys("result mutation")
+	if got := loaded.Sources()[0].ImportSelections[0].Path.String(); got != "/configkit/imports/0" {
+		t.Fatalf("result snapshot changed after mutation: %q", got)
+	}
+
+	failurePath := writeFacadeImportFile(t, directory, "failure.yml", `configkit:
+  imports:
+    - dir: absent
+    - path: required-missing.yml
+      required: true
+`)
+	_, err = Load(File(failurePath), Options[config]{Imports: true})
+	var loadError *LoadError
+	if !errors.As(err, &loadError) {
+		t.Fatalf("failed load error = %T %v", err, err)
+	}
+	errorSources := loadError.Sources()
+	if len(errorSources) != 1 || len(errorSources[0].ImportSelections) != 1 || errorSources[0].ImportSelections[0].Path.String() != "/configkit/imports/0" {
+		t.Fatalf("error source evidence = %#v", errorSources)
+	}
+	errorSources[0].ImportSelections[0].Path = Keys("error mutation")
+	if got := loadError.Sources()[0].ImportSelections[0].Path.String(); got != "/configkit/imports/0" {
+		t.Fatalf("error snapshot changed after mutation: %q", got)
+	}
+}
+
+func writeFacadeImportFile(t *testing.T, root, name, contents string) string {
+	t.Helper()
+	filename := filepath.Join(root, name)
+	if err := os.MkdirAll(filepath.Dir(filename), 0o700); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.WriteFile(filename, []byte(contents), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	return filename
+}
+
+func canonicalFacadeImportPath(t *testing.T, filename string) string {
+	t.Helper()
+	canonical, err := filepath.EvalSymlinks(filename)
+	if err != nil {
+		t.Fatal(err)
+	}
+	return filepath.ToSlash(canonical)
+}