Prepare the v0.14.0 release

AGENTS.md

@@ -1 +1,4 @@
 Please review `docs/development.md` for initial orientation in this repository and follow its task-specific reading guide.
+
+Review and follow `docs/policy/agents.md` before recommending, planning, or
+implementing changes to this repository.

README.md

@@ -6,7 +6,7 @@ OpenAI-compatible model endpoints, and returns generated output with validation
 metadata.
 
 The application uses
-[Promptkit v0.9.0](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/)
+[Promptkit v0.11.0](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/)
 for prompt, profile, schema, preparation, generation, and validation behavior.
 Scriptorium owns executable configuration, CLI and HTTP mapping, process
 behavior, output presentation, and HTTP artifact-containment policy.
@@ -32,20 +32,26 @@ a model. For complete invocation and output behavior, see the
 
 - [CLI reference](docs/cli.md)
 - [Configuration reference](docs/config.md)
+- [Application configuration inspection](docs/cli.md#application-configuration-inspection)
 - [HTTP API reference](docs/api.md)
 - [Operations guide](docs/operations.md)
 - [Consumer integration overview](docs/consumers/api.md)
 - [Migration from the former Go package](docs/consumers/migrating-to-promptkit.md)
 - [Subprocess integration](docs/integrations/subprocess.md)
 - [Architecture policy](docs/policy/architecture.md)
-- [Promptkit framework formats](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md)
-- [Promptkit Go consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md)
+- [Promptkit framework formats](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md)
+- [Promptkit Go consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md)
 
 ## Examples
 
 - [Minimal configuration](examples/config.yml) and
-  [complete configuration](examples/config.full.yml)
+  [complete configuration](examples/config.full.yml), plus a
+  [composed import/overlay example](examples/config.composed.yml)
 - [Prompt definitions](examples/prompts/), [execution profiles](examples/profiles/),
   [schemas](examples/schemas/), and [synthetic input fixtures](examples/fixtures/)
 - [Render script](examples/render-markdown-summary.sh)
 - [HTTP request](examples/http-run.json)
+
+Scriptorium is licensed under the [BSD 3-Clause License](LICENSE).
+
+[Open Source attribution notices](docs/oss/attribution.md) are available.

docs/api.md

@@ -20,7 +20,7 @@ of their `Content-Type`; successful JSON responses use
 
 The configured request-body limit includes inline artifact bodies. The artifact
 limit applies to HTTP `file` inputs. The response limit applies to the encoded
-response, including the artifact body and optional raw output. A limit of zero
+response, including the artifact body, optional raw output, and optional execution report. A limit of zero
 disables that limit.
 
 A request body over its limit returns `413 request_too_large`; an oversized
@@ -50,6 +50,9 @@ copyable shape. At the HTTP adapter boundary, the smallest valid shape is:
 | `vars` | no | Object mapping template-variable names to strings. |
 | `model` | no | Runtime model-override object. |
 | `include_raw_output` | no | Include `raw_model_output` when true. |
+| `include_execution_report` | no | Include a terminal `execution` report when true and Promptkit supplies a terminal observation. Omitted, `null`, or false leaves reporting off. |
+| `transient_retries` | no | Nonnegative additional transient attempt count for this run. Omitted or `null` inherits the configured engine value and then Promptkit's default; `0` disables additional attempts. |
+| `run_timeout_seconds` | no | Optional signed 64-bit whole-second limit for engine execution. Omitted, `null`, or `0` adds no deadline; positive values shorten the caller context deadline. |
 
 An input reference has a required `type` of `file` or `inline`. A `file`
 reference requires `uri`; an `inline` reference requires `body`.
@@ -65,11 +68,13 @@ The optional `model` object accepts `endpoint`, `model`, `temperature`,
 `max_tokens`, `top_p`, `timeout_seconds`, `service_tier`,
 `reasoning_effort`, `api_key_env`, and `extra_params`. Numeric ranges and
 framework credential semantics are defined by the
-[Promptkit format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md).
+[Promptkit format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md).
 Explicit zero values for the numeric fields are overrides; zero
-`timeout_seconds` disables the per-generation deadline only, retaining the
+`timeout_seconds` disables the per-attempt generation deadline only, retaining the
 request context and configured transport cap. The timeout layers are defined in
-the [Promptkit outbound integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/integrations/openai-compatible-chat.md#timeout-and-cancellation).
+the [Promptkit outbound integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/integrations/openai-compatible-chat.md#timeout-and-cancellation).
+Each built-in transient retry gets a fresh generation timeout; the caller's
+request context remains the outer cancellation boundary.
 
 Raw API-key values are not accepted. `api_key` and any other unknown model
 field cause `400 invalid_json`.
@@ -90,6 +95,17 @@ Request decoding rejects malformed JSON, unknown fields at every request level,
 and trailing JSON tokens with `400 invalid_json`. A blank `prompt_id` returns
 `400 invalid_request`. Omitted or empty `inputs` are passed to Promptkit, which
 reports any definition-required or template-referenced inputs.
+`transient_retries` must be a JSON integer; a wrong type or noninteger number
+returns `400 invalid_json`, while a negative integer returns
+`400 invalid_request` before execution. A request override does not change the
+shared engine setting used by other requests. Promptkit owns eligible statuses,
+retry waits, and the separate transient budget for each repair generation.
+`run_timeout_seconds` must be a JSON int64. A wrong type or int64 overflow
+returns `400 invalid_json`; a negative value or one whose conversion to a Go
+duration would overflow returns `400 invalid_request`. The deadline starts
+immediately before engine execution and excludes body decoding and response
+writing. An earlier caller deadline or cancellation still wins. Synchronous
+work is not guaranteed to stop immediately.
 
 ### Success Response
 
@@ -103,6 +119,7 @@ validation contract. The response contains:
 - `metadata`: run, prompt, rendered-prompt, profile, optional backend identity, model, input-hash, usage,
   timing, validation, and repair-attempt metadata; and
 - optional `raw_model_output` when requested.
+- optional `execution` when requested and a terminal observation exists.
 
 `metadata.model_params` has `endpoint`, `model`, `temperature`,
 `max_tokens`, `top_p`, and `timeout_seconds`, plus optional
@@ -124,6 +141,38 @@ A validation failure has `validation.status: "failed"`, `is_valid: false`,
 and any available diagnostic errors, while still returning the artifact and
 metadata.
 
+### Execution Report
+
+The opt-in `execution` object projects one Promptkit terminal observation. It
+is independent of `metadata.usage`, which retains its existing result-based
+meaning. A terminal event can describe a failed invocation even when no provider
+call was dispatched; if there is no terminal event, `execution` is omitted.
+
+| Field | Meaning |
+| --- | --- |
+| `run_id`, `prompt_id`, `prompt_version`, `profile_id`, `backend_id`, `model` | Terminal identifiers; absent values are omitted. |
+| `outcome` | `success`, `invalid_output`, `error`, or `other` for an unrecognized upstream value. |
+| `failure` | The known terminal failure category, `none`, or `other` for an unrecognized value. |
+| `status_code` | Provider status, omitted when unavailable. |
+| `retries_exhausted` | Boolean independent of the failure category. |
+| `attempts`, `transient_retries`, `repair_calls` | Unsigned terminal aggregate counts. |
+| `duration_ms`, `capacity_wait_ms`, `generation_duration_ms`, `retry_delay_ms` | Terminal observed durations in integer milliseconds; their sum need not equal wall time. |
+| `usage` | Object containing `prompt_tokens`, `completion_tokens`, `total_tokens`, `cached_tokens`, and `cache_write_tokens`. |
+
+Each `usage` count contains `value` (integer), `known` (boolean), and
+`complete` (boolean). Reported zero differs from an unknown count; an
+incomplete total does not establish complete billing. All counts, durations,
+`retries_exhausted`, and `usage` are present even when zero or unknown. The
+recognized failure values are `none`, `canceled`, `deadline`, `transport`,
+`provider_status`, `malformed_response`, `capacity`, `invalid_config`,
+`invalid_request`, `prompt_not_found`, `profile_not_found`, `prompt_load`,
+`profile_load`, `artifact_load`, `prompt_render`, `validation`, `generation`,
+and `other`.
+
+The report excludes content, credentials, URLs, headers, and raw errors, but
+its identifiers can still be sensitive. Apply appropriate access and retention
+controls to responses that include it.
+
 ## Error Responses
 
 Errors have this shape:
@@ -133,6 +182,12 @@ Errors have this shape:
 ```
 
 Messages are concise and do not expose wrapped internal causes.
+For an opted-in failed invocation, the error envelope can also include the
+optional top-level `execution` object described above. Decode and validation
+failures before engine invocation do not have a report. Report-bearing error
+responses are subject to the complete encoded response-size limit; if they
+exceed it, the response is the compact `413 response_too_large` error without
+`execution`. The existing compact error fallback itself is not re-limited.
 
 | Status | Code | Meaning |
 | --- | --- | --- |
@@ -149,6 +204,7 @@ Messages are concise and do not expose wrapped internal causes.
 | `404` | `prompt_not_found` | Prompt ID or version does not exist. |
 | `404` | `profile_not_found` | Profile ID does not exist. |
 | `405` | `method_not_allowed` | The route does not accept the method. |
+| `408` | `request_canceled` | The caller context was canceled during a failed run. |
 | `413` | `request_too_large` | Encoded request exceeds its limit. |
 | `413` | `artifact_too_large` | File input exceeds its limit. |
 | `413` | `response_too_large` | Encoded response exceeds its limit. |
@@ -156,6 +212,12 @@ Messages are concise and do not expose wrapped internal causes.
 | `500` | `internal_error` | Unclassified server failure. |
 | `502` | `llm_failed` | Outbound model request failed. |
 | `503` | `capacity_exceeded` | The selected model backend has no admission capacity. No retry timing is supplied. |
+| `504` | `deadline_exceeded` | The execution context's deadline expired during a failed run. |
+
+These context outcomes require the execution context to have ended. A provider
+attempt timeout while it remains live retains `502 llm_failed`. A successful
+run remains successful even if cancellation races with response formatting;
+delivery after a client disconnect cannot be guaranteed.
 
 ## Retry And Idempotency
 

docs/cli.md

@@ -25,10 +25,22 @@ go run ./cmd/scriptorium render \
 - `scriptorium serve`: start the HTTP server.
 - `scriptorium inspect prompt`: inspect one prompt definition without model execution.
 - `scriptorium inspect profile`: inspect one effective profile without model execution.
+- `scriptorium config validate`: validate application configuration without building an engine.
+- `scriptorium config show`: show the resolved application configuration.
+- `scriptorium config sources`: show the ordered configuration sources.
+- `scriptorium config explain`: explain one effective or historical configuration path.
 
-All commands accept `--config <path>` and reject positional arguments. An
-effective `prompt_dir` is required for every command. Supply it through the
-configuration contract or the command's `--prompt-dir` flag.
+All commands accept `--config <path>` and `--config-overlay <name>` and reject
+positional arguments. `--config-overlay` selects one application configuration
+overlay; it is separate from `--profile`, which selects a Promptkit execution
+profile. See [configuration composition](config.md#explicit-imports-and-selected-overlays)
+for name, root, and no-base behavior. `run`,
+`render`, `serve`, and `inspect prompt` require an effective `prompt_dir`;
+`inspect profile` does not. Supply the directory through the configuration
+contract or a supported `--prompt-dir` flag. Explicit application flags,
+including empty values, override file settings. Configuration load diagnostics
+go to stderr and exit with status 1; successful load warnings are printed once
+before command work.
 
 ## `scriptorium run`
 
@@ -47,6 +59,7 @@ Optional flags:
 | Flag | Meaning |
 | --- | --- |
 | `--config <path>` | Application configuration file. |
+| `--config-overlay <name>` | Selected application configuration overlay. |
 | `--prompt-dir <dir>` | Prompt-definition directory override. |
 | `--profile-dir <dir>` | Custom profile-directory override. |
 | `--schema-dir <dir>` | Schema base-directory override. |
@@ -56,6 +69,7 @@ Optional flags:
 | `--input name=path` | Optional input file mapping; repeat or use comma-separated mappings. |
 | `--var name=value` | Template-variable mapping; repeat or use comma-separated mappings. |
 | `--out <path>` | Write generated content to this file instead of stdout. |
+| `--execution-report <path>` | Write an opt-in terminal execution report as newline-terminated JSON to a separate file. |
 | `--llm-base-url <url>` | Runtime endpoint override. |
 | `--model <name>` | Runtime model override. |
 | `--api-key-env <name>` | Runtime API-key environment-variable name override. |
@@ -64,17 +78,32 @@ Optional flags:
 | `--top-p <float>` | Runtime top-p override. |
 | `--reasoning-effort <value>` | Runtime reasoning-effort override. |
 | `--timeout <duration>` | Runtime timeout override using Go duration syntax. |
+| `--run-timeout <duration>` | Optional whole-run engine deadline using Go duration syntax; zero or omission adds no deadline, negative values are invalid. |
+| `--transient-retries <integer>` | Nonnegative additional transient attempt count for this request; `0` disables additional attempts. |
 
 Deprecated aliases: `--prompt-id` for `--prompt`, and `--profile-id` for
 `--profile`.
 
 Omitted numeric runtime flags preserve the selected effective value; explicit
-zero values override it. `--timeout 0s` disables the per-generation deadline
+zero values override it. `--timeout 0s` disables the per-attempt generation deadline
 only; the caller context and configured transport cap remain active. CLI
 durations are converted to whole seconds by truncation toward zero, so any
 duration whose absolute value is below one second becomes an explicit
 zero-second override. The timeout layers are defined in the
-[Promptkit outbound integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/integrations/openai-compatible-chat.md#timeout-and-cancellation).
+[Promptkit outbound integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/integrations/openai-compatible-chat.md#timeout-and-cancellation).
+With the built-in client, each transient retry gets a fresh generation timeout;
+this flag does not limit the complete run.
+`--run-timeout` keeps subsecond precision and starts immediately before engine
+execution. It bounds preparation, capacity waits, retry waits, generation, and
+repairs where they honor context. Parsing, engine construction, and output
+writes are outside that interval. A failed run whose execution context reaches
+this deadline exits 1 with a safe deadline diagnostic; a completed invalid
+output still exits 2. It does not promise immediate interruption of synchronous
+work.
+Omitting `--transient-retries` inherits the application setting or Promptkit's
+current default of two additional attempts. The flag overrides only this
+request; it does not alter the configured engine. Structural repair generations
+have separate transient retry budgets.
 
 There is no raw API-key flag. Use `--api-key-env`.
 
@@ -95,6 +124,32 @@ profile value, a nonblank value replaces that value, and
 `--reasoning-effort=` explicitly clears inherited reasoning. Promptkit treats
 nonblank values as provider-specific opaque strings.
 
+### Execution report file
+
+`--execution-report` requires a nonempty path. It is available only for `run`;
+`-` is an ordinary filename. Parent directories are not created. An existing
+file is replaced through ordinary file writing, without rollback or an atomic
+replacement guarantee. Generated content still goes to stdout or `--out`.
+Equal output/report paths, including existing paths that refer to the same
+file, are rejected before execution.
+
+The file is written after engine execution when Promptkit supplies a terminal
+event, including for a failed run or invalid generated output. A failure before
+invocation leaves the report path untouched. If no terminal event is supplied,
+Scriptorium leaves the path untouched and emits a diagnostic; an existing file
+may therefore contain an older report. Artifact output is attempted first on a
+successful run, and report writing is attempted even if artifact writing fails.
+Any output or report write failure exits 1; already-written output is not
+undone. Without a write failure, the actual run keeps its usual exit status.
+
+The [HTTP API reference](api.md#execution-report) defines the shared report
+fields and usage-presence semantics. The file contains that JSON object followed
+by a newline.
+
+The report excludes content, credentials, URLs, headers, and raw errors, but
+its identifiers can still be sensitive. The caller owns file retention and
+access.
+
 ## `scriptorium render`
 
 ```text
@@ -102,18 +157,19 @@ scriptorium render [flags]
 ```
 
 `--prompt <id>` is required. The following optional flags are supported:
-`--config`, `--prompt-dir`, `--profile-dir`, `--prompt-version`, `--profile`,
+`--config`, `--config-overlay`, `--prompt-dir`, `--profile-dir`, `--schema-dir`, `--prompt-version`, `--profile`,
 `--input`, `--var`, `--out`, `--llm-base-url`,
 `--model`, `--api-key-env`, `--temperature`, `--max-tokens`, `--top-p`,
-`--reasoning-effort`, `--session-id`, `--timeout`, and `--format text|json`. Their meanings match the corresponding
+`--reasoning-effort`, `--session-id`, `--timeout`, `--transient-retries`, and `--format text|json`. Their meanings match the corresponding
 `run` flags; `--format` selects prepared-run output and otherwise uses
 `defaults.render_format`.
 
 The same deprecated aliases and numeric/timeout behavior as `run` apply.
 The same session and reasoning inheritance, replacement, and clearing behavior
 also applies.
-`render` does not accept `--schema-dir`; configure `schema_dir` through the
-configuration file. It resolves profiles and schemas as part of preparation but
+`render` validates the retry override but does not generate model output or
+include a retry field in prepared output.
+`render` resolves profiles and schemas as part of preparation but
 does not call an LLM.
 
 ## `scriptorium serve`
@@ -127,6 +183,7 @@ Optional flags:
 | Flag | Meaning |
 | --- | --- |
 | `--config <path>` | Application configuration file. |
+| `--config-overlay <name>` | Selected application configuration overlay. |
 | `--addr <listen-address>` | HTTP listen-address override. |
 | `--prompt-dir <dir>` | Prompt-definition directory override. |
 | `--profile-dir <dir>` | Custom profile-directory override. |
@@ -143,7 +200,7 @@ schemas, and error codes are defined in the [HTTP API reference](api.md).
 
 ```text
 scriptorium inspect prompt --prompt ID [--prompt-version VERSION]
-    [--config PATH] [--prompt-dir DIR] [--format text|json] [--out PATH]
+    [--config PATH] [--config-overlay NAME] [--prompt-dir DIR] [--format text|json] [--out PATH]
 ```
 
 `--prompt` is required. Inspection uses normal configuration discovery and a
@@ -156,7 +213,7 @@ template, reserve backend capacity, or call a model.
 
 ```text
 scriptorium inspect profile --profile ID
-    [--config PATH] [--profile-dir DIR] [--format text|json] [--out PATH]
+    [--config PATH] [--config-overlay NAME] [--profile-dir DIR] [--format text|json] [--out PATH]
 ```
 
 `--profile` is required; built-in profiles need no prompt directory. Inspection
@@ -164,6 +221,49 @@ resolves profile inheritance and backend defaults, but never reads a credential
 value, loads a prompt, reserves capacity, or calls a model. Unset provider
 controls are shown as zero values when Promptkit leaves them unspecified.
 
+## Application configuration inspection
+
+```text
+scriptorium config validate [flags]
+scriptorium config show [flags]
+scriptorium config sources [flags]
+scriptorium config explain --path JSON_POINTER [flags]
+```
+
+Each accepts `--config`, `--config-overlay`, `--prompt-dir`, `--profile-dir`,
+`--schema-dir`, `--addr`, `--artifact-root`, `--max-request-bytes`,
+`--max-artifact-bytes`, and `--max-response-bytes`. These are application
+configuration selectors and overrides; request, credential, backend-editing,
+output-format, and output-file flags are not accepted. Positional arguments
+are rejected.
+
+`validate` writes no success output; status 0 means the application configuration
+loaded and passed its own validation. It does not establish that a prompt
+directory exists, credentials are present, or Promptkit can construct an
+engine. `show` writes ConfigKit's resolved, redacted YAML snapshot. `sources`
+writes one newline-terminated JSON array of source records in load order. Each
+record has `id`, `label`, and `layer`, plus `path` when available; explicit CLI
+overrides appear as synthetic sources. Defaults-only input without overrides
+produces `[]` for `sources`.
+
+`explain` requires an explicitly supplied `--path` in JSON Pointer syntax.
+The empty string selects the root; `/server/addr` selects a field; `~1` and
+`~0` escape `/` and `~` inside mapping keys. Numeric tokens are interpreted
+according to the containing mapping or sequence. Dotted paths, file paths,
+and URL fragments are not accepted. ConfigKit writes a human-readable account
+of the effective value and its chronological default, base, overlay, CLI,
+and normalization history, including removed historical values. This is not a
+JSON schema or a byte-for-byte stable prose format. Unknown and malformed paths
+exit 1 without successful explanation output.
+
+Diagnostics and warnings go to stderr, never into successful stdout. Parse,
+load, output, and diagnostic-write failures exit 1. Ordinary output writes may
+leave partial data on failure. The inspection output can contain paths,
+identifiers, `api_key_env` reference names, and ordinary `extra_params` values,
+including displaced historical values in explanations;
+protect it accordingly. Raw API-key values are not read or emitted by these
+commands.
+
 ## Input And Variable Syntax
 
 `--input name=path` maps an input name to a local file; `--var name=value`

docs/config.md

@@ -3,7 +3,7 @@
 This is the canonical reference for Scriptorium application settings. Prompt,
 profile, schema, execution-setting, built-in profile, and framework credential
 semantics are defined by the
-[Promptkit v0.9.0 format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md).
+[Promptkit v0.11.0 format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md).
 For command syntax, see the [CLI reference](cli.md); for HTTP request shapes and
 outcomes, see the [HTTP API reference](api.md).
 
@@ -20,17 +20,66 @@ When `--config` is omitted, Scriptorium searches
 If neither exists, it uses built-in defaults. An explicit `--config` path must
 exist and decode successfully.
 
-The maintained [minimal configuration](../examples/config.yml) and
-[complete configuration](../examples/config.full.yml) are copyable examples.
+The maintained [minimal configuration](../examples/config.yml),
+[complete configuration](../examples/config.full.yml), and
+[composed configuration](../examples/config.composed.yml) are copyable examples.
 
 ## Application Configuration File
 
-Configuration is strict YAML: unknown fields are rejected. Empty string values
-do not override a prior value. Raw API-key fields are not accepted.
+Configuration is decoded by ConfigKit v0.4.0 as strict YAML. Unknown fields,
+empty files, multiple YAML documents, and YAML aliases/anchors are rejected.
+An explicit empty string is a value: it clears a file-provided optional path
+or, for required `prompt_dir`, fails the command's prerequisite. A scalar
+`null` is not an empty string and is rejected for scalar fields; nullable
+fields such as `transient_retries` retain their documented meaning. Raw
+API-key fields are not accepted.
+
+ConfigKit applies defaults, the selected file, then only explicitly supplied
+application CLI flags before validating the result. Paths in the application
+document remain relative to the process working directory, not to the file's
+directory. Load diagnostics are printed safely on stderr; invalid application
+configuration exits with status 1.
+
+### Explicit imports and selected overlays
+
+A file-backed configuration may declare `configkit.imports` with relative
+`path` or `glob` entries. Each entry is explicit; Scriptorium does not scan
+for fragments on its own. ConfigKit composes the selected file and imports by
+conflict, so two fragments cannot define the same field merely to establish
+precedence. A required missing import fails; an entry may use
+`optional: true` to suppress target nonexistence. Imports are resolved from
+the declaring file's canonical directory and remain within the selected
+file's canonical directory by default. Scriptorium does not widen that root.
+
+`--config-overlay NAME` selects one file at `overlays/NAME.yml` below that
+root. The name must use letters, digits, `_`, or `-`; it is not a path. Empty
+and exact lowercase `default` select no overlay. There is no fallback,
+case folding, inherited stack, or automatic selection; a missing selected
+overlay fails. ConfigKit merges mapping fields and replaces other values,
+including sequences and explicit nulls, before applying explicit CLI flags.
+An active overlay needs a selected base file. No-source defaults-only loads
+cannot use imports or an active overlay.
+
+The [composed example](../examples/config.composed.yml) imports two fragments
+and has a selectable [local overlay](../examples/overlays/local.yml). From the
+repository root:
+
+```bash
+go run ./cmd/scriptorium render \
+  --config ./examples/config.composed.yml --config-overlay local \
+  --prompt generic.markdown_summary \
+  --input transcript=./examples/fixtures/transcript.md \
+  --input glossary=./examples/fixtures/glossary.yml
+```
+
+Import and overlay file locations are constrained by ConfigKit;
+*application path values* such as `prompt_dir` remain relative
+to the process working directory even when authored in a fragment or overlay.
+These source constraints are separate from HTTP artifact containment.
 
 | Field | Default | Meaning |
 | --- | --- | --- |
-| `prompt_dir` | unset | Promptkit prompt-definition source directory. `run`, `render`, and `serve` require an effective value. |
+| `prompt_dir` | unset | Promptkit prompt-definition source directory. `run`, `render`, `serve`, and `inspect prompt` require an effective value. |
 | `profile_dir` | unset | Optional custom Promptkit profile source directory overlaid on Promptkit built-ins. |
 | `schema_dir` | `.` | Promptkit schema source directory for relative schema paths. |
 | `server.addr` | `:8080` | Address used by `serve`. |
@@ -40,11 +89,28 @@ do not override a prior value. Raw API-key fields are not accepted.
 | `server.max_response_bytes` | `16777216` | Maximum encoded HTTP response bytes; `0` disables the limit. |
 | `defaults.render_format` | `text` | Default prepared-run output format: `text` or `json`. |
 | `backends` | unset | Optional mapping of custom Promptkit backend IDs to engine-scoped connection and capacity settings. |
+| `transient_retries` | Promptkit default | Optional nonnegative count of additional built-in transient generation attempts. Omitted or `null` delegates to Promptkit; `0` disables additional attempts. |
 
 The size fields must be zero or greater. The [HTTP API](api.md) defines how
 each limit is enforced and reported. `server.artifact_root` configures an HTTP
 deployment boundary; see [operations](operations.md) for deployment handling.
 
+The `config validate`, `config show`, `config sources`, and `config explain`
+commands inspect this
+application configuration without creating a Promptkit engine. `show` is a
+resolved inspection snapshot, not necessarily a reloadable file. Application
+validation does not check framework sources, backend connectivity, or credential
+availability. `api_key_env` is an environment-variable *name* and remains
+visible in inspection; its environment value is never read by ConfigKit.
+Ordinary backend `extra_params` values also remain visible, so keep secrets out
+of those fields and protect inspection output as potentially sensitive.
+`config explain --path /server/addr` uses JSON Pointer syntax to show the
+selected field's current value and history from defaults, files, overlays,
+explicit CLI overrides, and normalization. `--path ''` selects the root.
+History includes replaced and removed values; `show` presents the final
+resolved snapshot. Neither renderer looks up credential bytes, and neither
+claims to identify arbitrary secrets in unclassified parameter values.
+
 ## Custom Backends
 
 Use `backends` when a profile selects an application-defined backend ID:
@@ -86,13 +152,20 @@ parse those framework file formats.
   contracts.
 
 See the
-[tagged Promptkit format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md)
+[tagged Promptkit format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md)
 for all of those definitions. The files under
 [`examples/prompts`](../examples/prompts/),
 [`examples/profiles`](../examples/profiles/), and
 [`examples/schemas`](../examples/schemas/) are maintained Scriptorium
 application inputs using that tagged format.
 
+Promptkit supplies the opt-in profile IDs `promptkit-openrouter-standard` and
+`promptkit-openrouter-auto`. They can be selected as a prompt default or request
+profile, and custom profile sources retain precedence. Their effective model
+settings may change with a catalog upgrade; use a complete application-owned
+profile when fixed settings are required. See the tagged format reference for
+the maintained-profile contract.
+
 ## Credentials And Outbound Behavior
 
 Scriptorium maps `--api-key-env` and HTTP `model.api_key_env` into Promptkit
@@ -113,8 +186,16 @@ A positive Promptkit `repair_attempts` budget can add provider calls, latency,
 token use, and cost; see the tagged format reference for its permitted values
 and validation-mode requirements.
 
+Promptkit's built-in client makes up to two additional attempts for eligible
+transient HTTP statuses by default. The application `transient_retries` setting
+changes that budget for its constructed engine; an explicit `run` request
+override takes precedence. This can increase latency, request count, and cost.
+Each initial or repair generation has its own transient budget; an injected
+client retains its own retry behavior. The tagged integration contract owns
+eligibility and timing.
+
 Promptkit's
-[OpenAI-compatible integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/integrations/openai-compatible-chat.md)
+[OpenAI-compatible integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/integrations/openai-compatible-chat.md)
 defines outbound authentication, provider request mapping, transport limits,
 and timeout layering.
 
@@ -123,4 +204,4 @@ and timeout layering.
 - [CLI reference](cli.md)
 - [HTTP API reference](api.md)
 - [Operations guide](operations.md)
-- [Promptkit framework formats](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md)
+- [Promptkit framework formats](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md)

docs/consumers/api.md

@@ -14,7 +14,7 @@ and the HTTP service according to the boundary your application needs.
 
 Go applications that need an in-process prompt framework should import
 Promptkit directly. The tagged
-[Promptkit Go consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md)
+[Promptkit Go consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md)
 owns that interface; Scriptorium does not provide a Go library package.
 Consumers arriving from the former Scriptorium Go API should follow the
 [migration guide](migrating-to-promptkit.md).
@@ -35,4 +35,4 @@ Scriptorium does not persist run state. A retry can produce different output
 and can incur another provider request. CLI exits belong to the
 [CLI reference](../cli.md), HTTP status behavior belongs to the
 [HTTP API reference](../api.md), and framework semantics belong to
-[Promptkit v0.9.0](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md).
+[Promptkit v0.11.0](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md).

docs/consumers/migrating-to-promptkit.md

@@ -4,7 +4,7 @@
 
 Scriptorium `v0.11.1` at
 `gitea.maximumdirect.net/eric/scriptorium` is the final release that provides
-the former in-process Go framework. Promptkit `v0.9.0` at
+the former in-process Go framework. Promptkit `v0.11.0` at
 `gitea.maximumdirect.net/eric/promptkit` is the destination for that framework
 API. Scriptorium `v0.12.0` and later provide the CLI and HTTP application only.
 
@@ -19,7 +19,7 @@ Start from a clean consumer checkout and review the pending diff before
 committing it. Add the published Promptkit module:
 
 ```sh
-go get gitea.maximumdirect.net/eric/promptkit@v0.9.0
+go get gitea.maximumdirect.net/eric/promptkit@v0.11.0
 ```
 
 For an ordinary consumer that imports the former root package under its
@@ -51,14 +51,14 @@ go mod tidy
 go test ./...
 ```
 
-Confirm that `go.mod` selects Promptkit `v0.9.0` and that no Go file imports
+Confirm that `go.mod` selects Promptkit `v0.11.0` and that no Go file imports
 the former Scriptorium package:
 
 ```sh
 test "$(
   go list -m -f '{{.Path}}@{{.Version}}' \
     gitea.maximumdirect.net/eric/promptkit
-)" = 'gitea.maximumdirect.net/eric/promptkit@v0.9.0'
+)" = 'gitea.maximumdirect.net/eric/promptkit@v0.11.0'
 if git grep -n \
   'gitea.maximumdirect.net/eric/scriptorium' \
   -- '*.go'
@@ -73,20 +73,20 @@ fi
 Promptkit preserves the established engine, request, result, profile,
 source-option, model-client, artifact, validation-value, and public-error
 shapes where practical. Exact declarations and current behavior belong to the
-tagged [Promptkit consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md)
+tagged [Promptkit consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md)
 and Go source.
 
 Promptkit also includes migration-relevant public contracts that were not in
 Scriptorium `v0.11.1`:
 
-- [`WithArtifactReader`](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/engine.go#L182)
+- [`WithArtifactReader`](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/engine.go#L169-L174)
   and the
-  [`ArtifactReader` declaration](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/types.go#L304)
+  [`ArtifactReader` declaration](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/types.go#L315)
   provide the artifact-reading extension described by the tagged
-  [extension-interface guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md#extension-interfaces).
-- [`ErrProfileRequired` and `ErrAPIKeyEnvMissing`](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/engine.go#L46-L61)
+  [extension-interface guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md#extension-interfaces).
+- [`ErrProfileRequired` and `ErrAPIKeyEnvMissing`](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/engine.go#L33-L48)
   provide the specific identities described by the tagged
-  [error guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md#errors).
+  [error guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md#handle-errors).
 
 Use those tagged owners for exact signatures, wrapping guarantees, and
 extension behavior.

docs/development.md

@@ -2,17 +2,19 @@
 
 This is the contributor entry point for Scriptorium. Scriptorium is an
 application that consumes the public
-[Promptkit v0.9.0 package](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md);
-framework implementation work belongs in Promptkit.
+[Promptkit v0.11.0 package](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md);
+framework implementation work belongs in Promptkit. Application configuration
+uses ConfigKit v0.4.0; keep Scriptorium's schema and source selection local.
 
 ## Initial Orientation
 
-Before starting work:
+Before recommending, planning, or implementing work:
 
-1. inspect the working tree and preserve unrelated changes;
-2. read the [architecture policy](policy/architecture.md);
-3. follow the task-specific contracts and internal documents below; and
-4. inspect the relevant implementation and tests before changing them.
+1. read the [agent policy](policy/agents.md);
+2. inspect the working tree and preserve unrelated changes;
+3. read the [architecture policy](policy/architecture.md);
+4. follow the task-specific contracts and internal documents below; and
+5. inspect the relevant implementation and tests before changing them.
 
 Also read the [documentation policy](policy/documentation.md) before changing
 documentation and the [testing policy](policy/testing.md) before changing
@@ -25,14 +27,15 @@ tests.
 | Repository orientation or component responsibility | [Internal component overview](internal/overview.md) and [architecture policy](policy/architecture.md) |
 | CLI commands, flags, output, or exit behavior | [CLI contract](cli.md) and [adapter internals](internal/adapters.md) |
 | HTTP routes, DTOs, limits, status mapping, or artifact policy | [HTTP API contract](api.md), [adapter internals](internal/adapters.md), and [source internals](internal/sources.md) |
-| Application configuration or precedence | [Configuration contract](config.md), [adapter internals](internal/adapters.md), and [source internals](internal/sources.md) |
+| Application configuration or precedence | [Configuration contract](config.md), [adapter internals](internal/adapters.md), [source internals](internal/sources.md), and `internal/config` definition/load tests |
 | Prepared-run presentation | [CLI contract](cli.md), [adapter internals](internal/adapters.md), and `internal/format` |
-| Prompt, profile, schema, generation, or validation semantics | [Promptkit framework formats](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md) and the [Promptkit consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md) |
-| OpenAI-compatible outbound behavior or timeout layering | [Promptkit integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/integrations/openai-compatible-chat.md) |
+| Prompt, profile, schema, generation, or validation semantics | [Promptkit framework formats](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md) and the [Promptkit consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md) |
+| OpenAI-compatible outbound behavior or timeout layering | [Promptkit integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/integrations/openai-compatible-chat.md) |
 | Subprocess behavior | [Subprocess integration](integrations/subprocess.md) and [CLI contract](cli.md) |
 | Runtime operation or recovery | [Operations](operations.md) |
-| Release packaging or publication | The [release procedure](release.md), [hosted release workflow](../.woodpecker/release.yml), and [architecture policy](policy/architecture.md) |
-| Examples or copyable assets | The owning Scriptorium contract, the relevant [Promptkit format contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md), and the related files under `examples/` |
+| Dependencies, copied assets, or attribution | The [OSS policy](policy/oss.md), [component attributions](oss/attribution.md), and [architecture policy](policy/architecture.md) |
+| Release documentation, packaging, or publication | The [release policy](policy/release.md), [OSS policy](policy/oss.md), [release procedure](release.md), [hosted release workflow](../.woodpecker/release.yml), and [architecture policy](policy/architecture.md) |
+| Examples or copyable assets | The owning Scriptorium contract, the relevant [Promptkit format contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md), and the related files under `examples/` |
 | Architecture decisions or future work | The [documentation policy](policy/documentation.md), relevant accepted ADRs, and relevant roadmap documents |
 
 Cross-project changes land and release in Promptkit before Scriptorium adopts

docs/integrations/subprocess.md

@@ -17,6 +17,15 @@ Pass required API-key environment variables through the child environment. Do
 not place raw API keys in arguments. Keep the environment limited to the values
 needed for the selected profile.
 
+For application-only preflight, invoke `scriptorium config validate`; it exits
+0 with empty stdout when the application configuration loads, but does not
+prove that Promptkit sources or credentials are usable. `config show` emits
+resolved YAML and `config sources` emits newline-terminated JSON provenance.
+All three use the same source selection and explicit overrides as execution
+commands and require separate stderr capture for diagnostics. Their output can
+include source paths, identifiers, credential reference names, and ordinary
+backend parameters; treat it as potentially sensitive.
+
 ## Streams And Output Ownership
 
 Capture stdout and stderr separately. Stdout contains the requested artifact or
@@ -29,6 +38,13 @@ When using `--out`, the caller owns the output path, its permissions, and
 cleanup. Treat rendered prompts, generated artifacts, stdout, and stderr as
 potentially sensitive.
 
+For `run`, an optional [execution report file](../cli.md) contains terminal
+metadata separate from generated stdout or `--out`. The caller owns that
+path's permissions and retention. Output and report files are not a
+transaction: one may be written when the other fails, and a preexisting report
+may remain when the invocation has no terminal event. Check the exit status
+and diagnostics rather than assuming a report path contains this run's data.
+
 ## Cancellation And Recovery
 
 A CLI invocation performs one synchronous request and creates no durable run
@@ -37,5 +53,11 @@ process according to its own process-management policy. A later invocation is a
 new request and can make another model call; there is no resume or checkpoint
 protocol.
 
+For `run`, callers may set the [whole-run deadline](../cli.md) to bound the
+engine invocation. It is separate from the per-attempt generation timeout and
+does not include child startup or output writing. A deadline failure exits 1;
+supervisors still own process termination when they need to stop the child
+outside that engine interval.
+
 For deployment, filesystem permissions, and sensitive-artifact handling, see
 the [operations guide](../operations.md).

docs/internal/adapters.md

@@ -9,7 +9,7 @@ own IO and presentation mechanics, not framework decisions.
 External contracts are canonical in the [CLI reference](../cli.md) and
 [HTTP API reference](../api.md). Promptkit's public engine contract is
 described by its tagged
-[Go consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md).
+[Go consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md).
 
 ## Components And Collaborators
 
@@ -18,11 +18,17 @@ described by its tagged
 - `internal/adapter/cli` resolves settings through `internal/config`,
   constructs `promptkit.Engine`, maps CLI values to `promptkit.RunRequest`,
   and owns output files, summaries, and exit codes.
+- `internal/config` delegates application YAML decoding, default resolution,
+  overrides, and validation ordering to ConfigKit. The CLI supplies only
+  explicitly visited application flags as overrides and renders safe structured
+  load diagnostics on stderr.
 - `internal/adapter/http` strictly decodes request DTOs, maps them to Promptkit
   public values, calls its adapter-owned `Runner` interface, and maps results
   and errors to HTTP DTOs.
 - `internal/format` renders `promptkit.PreparedRun` values as deterministic text
   or JSON.
+- `internal/report` maps Promptkit's terminal execution event to a small
+  application-owned value without invoking the engine or writing output.
 
 ## Wiring Flows
 
@@ -33,6 +39,10 @@ described by its tagged
 and profile selection, optional file inputs, variables, and presence-aware
 execution overrides. Omitted framework settings remain zero values so Promptkit
 resolves its own defaults and definition-required inputs.
+The application retry setting enters shared engine construction, while a
+run/render flag enters only the mapped request. The server uses the same engine
+setting and retains one backend admission state. Render prepares without
+generation.
 
 `inspect prompt` maps the selected ID and optional version to
 `promptkit.Engine.InspectPrompt`, then formats Scriptorium-owned inspection
@@ -43,11 +53,16 @@ preparation or generation.
 formats a safe application-owned effective profile view. It intentionally does
 not require prompt, schema, or artifact sources.
 
+The `config` family loads the same application ConfigKit result once and uses
+its validation, effective-output, source, and explanation views directly. It
+does not construct a Promptkit engine. The CLI maps source records into a small
+JSON DTO for `sources`; `show` and `explain` use ConfigKit's own writers.
+
 `serve` constructs Scriptorium's restricted HTTP artifact reader, injects it
 with `promptkit.WithArtifactReader`, passes the engine through the HTTP
 adapter's consumer-owned `Runner` interface, and starts the server.
 
-All three CLI paths assemble the engine from the same resolved prompt, profile,
+The command paths assemble the engine from the same resolved prompt, profile,
 and schema directories plus configured custom backends. Each backend is mapped
 to Promptkit's public `Backend` value and registered during engine construction,
 so one constructed server engine retains one immutable backend registry and its
@@ -59,11 +74,30 @@ The handler enforces transport limits and strict JSON decoding before mapping
 DTOs into `promptkit.RunRequest`, `promptkit.ArtifactRef`, and
 `promptkit.ExecutionTargetOverride`. On success it reads Promptkit artifact,
 validation, model, usage, and metadata values directly.
+The optional HTTP retry count maps to `RunRequest.TransientRetries` without
+mutating the shared engine. Negative values are rejected before invocation;
+Promptkit controls retries and backend admission for concurrent requests.
 
 Failure mapping uses `errors.Is` against Promptkit's public sentinels and the
 HTTP reader's Scriptorium-owned containment and size errors. Wrapped reader
 errors preserve their identity through Promptkit's artifact-load boundary.
 
+The report collector is per invocation. It ignores nonterminal events and
+copies terminal aggregates once, including usage presence and completeness.
+It retains no history or shared state; adapters decide whether to attach it and
+how to present its snapshot.
+CLI `run` attaches one only when a report path is selected, writes generated
+output first on success, and then writes the terminal snapshot when present.
+Its report path remains untouched for pre-invocation failures or absent
+terminal observations; report and artifact writes are independent.
+HTTP `POST /v1/runs` attaches a separate collector to each invocation only when
+`include_execution_report` is true. Its optional top-level `execution` field
+appears in success or error envelopes only when Promptkit emits a terminal
+event. Pre-invocation adapter errors and absent terminal events leave it out.
+The complete serialized success or report-bearing error response is subject to
+the configured response-size limit; overflow returns the compact 413 error
+without a report. Field meanings are owned by the [API contract](../api.md#execution-report).
+
 ## Package-Local Guarantees
 
 - Adapters contain no copied framework types or orchestration.

docs/internal/overview.md

@@ -7,12 +7,13 @@ public behavior belongs in the linked contracts.
 | Component | Implemented responsibility | References |
 | --- | --- | --- |
 | `cmd/scriptorium` | Process entrypoint that delegates arguments and streams to the CLI adapter. | [CLI contract](../cli.md), [adapter internals](adapters.md) |
-| `internal/adapter/cli` | Parses commands, resolves application settings, constructs Promptkit engines, maps requests, and owns process output and exit behavior. | [CLI contract](../cli.md), [adapter internals](adapters.md) |
+| `internal/adapter/cli` | Parses execution and application-configuration inspection commands, resolves settings, constructs Promptkit engines only for execution/definition inspection, maps requests, and owns process output and exit behavior. | [CLI contract](../cli.md), [adapter internals](adapters.md) |
 | `internal/adapter/http` | Owns routes, DTOs, strict decoding, limits, Promptkit request/result mapping, public error mapping, and restricted HTTP artifact reading. | [HTTP API](../api.md), [adapter internals](adapters.md), [source internals](sources.md) |
-| `internal/config` | Discovers and strictly decodes application configuration and applies built-in and CLI precedence. | [configuration contract](../config.md), [adapter internals](adapters.md) |
+| `internal/config` | Selects application sources, supplies the ConfigKit definition and explicit overrides, and converts resolved settings for runtime use. | [configuration contract](../config.md), [adapter internals](adapters.md) |
 | `internal/defaults` | Holds Scriptorium-owned application and HTTP defaults. | [configuration contract](../config.md) |
 | `internal/format` | Formats Promptkit prepared-run and prompt/profile inspection values for CLI text or JSON output. | [CLI contract](../cli.md), [adapter internals](adapters.md) |
+| `internal/report` | Projects one terminal Promptkit observation into Scriptorium-owned execution metadata. | [adapter internals](adapters.md) |
 
 Framework implementation packages are provided by
-[Promptkit v0.9.0](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md)
+[Promptkit v0.11.0](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md)
 and are not part of this repository.

docs/internal/sources.md

@@ -5,7 +5,7 @@
 This document covers Scriptorium-owned source locations and the restricted HTTP
 artifact reader. Prompt, profile, schema, and ordinary artifact semantics are
 owned by the tagged
-[Promptkit format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md).
+[Promptkit format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md).
 
 ## Application Source Locations
 
@@ -24,6 +24,36 @@ The [configuration reference](../config.md) owns Scriptorium's source-location
 fields and precedence. Maintained files under `examples/` are application
 inputs that use Promptkit's tagged formats.
 
+## Application Configuration Boundary
+
+`internal/config` contains a ConfigKit-native `Document` definition. Its
+fresh defaults come from Scriptorium's existing default constants; its resolver
+records only actual trimming, path cleaning, and render-format normalization.
+The validator reports application retry, size-limit, and format errors without
+mutating the document or exposing rejected values. Application path values keep
+working-directory-relative meaning. ConfigKit owns strict YAML decoding and
+presence semantics, including dynamic integer values in backend parameters.
+
+The `Load` boundary selects an explicit file, the first existing implicit
+configuration file, or ConfigKit's defaults-only source. Only absence advances
+implicit discovery. One ConfigKit load applies explicit application overrides
+before validation and returns a result containing the resolved document and
+its frozen inspection records. `Settings` converts that document into the
+existing runtime settings with backend IDs sorted; it does not read files or
+apply another merge. File-backed loads enable explicit ConfigKit imports and
+select one named overlay when requested. ConfigKit uses the selected file's
+canonical directory as the source root; the application does not widen it.
+Application path values retain process-working-directory-relative meaning.
+
+Every configuration-loading command uses this boundary. The CLI maps only
+explicitly visited application flags to ConfigKit overrides and prints
+structured load diagnostics before engine construction. The old application
+YAML decoder and merge path have been removed.
+Application `config` inspection reuses the result's frozen effective,
+provenance, and history views without constructing an engine or reloading
+sources. It does not resolve API-key environment values; ordinary backend
+parameters remain visible by the selected inspection policy.
+
 ## Restricted HTTP Artifact Reader
 
 `internal/adapter/http` implements `promptkit.ArtifactReader` for HTTP
@@ -55,7 +85,8 @@ codes.
 
 Inspect:
 
-- `internal/config/config_test.go`
+- `internal/config/document_test.go`
+- `internal/config/load_test.go`
 - `internal/adapter/cli/run_test.go`
 - `internal/adapter/http/artifact_reader_test.go`
 - `internal/adapter/http/handler_test.go`

docs/operations.md

@@ -9,7 +9,7 @@ fields, or HTTP wire behavior.
 - [CLI reference](cli.md): commands, output destinations, and exit codes.
 - [Configuration reference](config.md): application settings, source
   locations, defaults, and credential mapping.
-- [Promptkit framework formats](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md):
+- [Promptkit framework formats](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md):
   prompt, profile, schema, execution-setting, and framework credential
   contracts.
 - [HTTP API reference](api.md): route, request/response schema, status codes,
@@ -26,6 +26,16 @@ configuration, or environment and submitting a new request.
 Generated artifacts, rendered prompts, model output, and run metadata are
 caller-owned data. Retention, encryption, backup, and deletion are deployment
 responsibilities.
+Optional [CLI execution reports](cli.md) also belong to the caller. They can
+contain sensitive identifiers even though they omit prompt and response bodies.
+Keep report paths separate from generated output and apply the same access and
+retention controls. Output and report writes are not transactional; an older
+report may remain when no terminal observation was produced.
+HTTP callers may also request a [terminal execution report](api.md#execution-report).
+It shares the same identifier sensitivity and should receive response access
+and retention controls. An opted-in report increases the encoded response size;
+an oversized success or report-bearing error becomes the compact 413 response
+without a report.
 
 ## Deploy The Filesystem And Process
 
@@ -34,6 +44,21 @@ schema sources that follow the tagged framework formats. For an HTTP deployment
 that accepts file artifacts, use a dedicated, narrow artifact directory rather
 than a general-purpose or sensitive filesystem tree.
 
+If application configuration uses explicit imports or a selected overlay,
+deploy those files beneath the selected configuration file's canonical
+directory. ConfigKit keeps imported and overlay files within that root;
+Scriptorium does not widen it. Application path values inside those files still
+resolve from the process working directory, so set that directory deliberately
+in the service manager. This configuration-source boundary is distinct from
+the HTTP artifact root and does not constrain prompt or schema directories.
+
+When upgrading existing definitions, review selected profile IDs.
+`deepseek-3-2`, `deepseek-4-flash`, and `deepseek-4-pro` were removed from the
+selected catalog. Choose an available catalog profile, one of Promptkit's
+maintained names, or a complete application-owned profile. New DeepSeek names
+are not aliases for the removed ones; see the
+[v0.10 migration guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.10.0/docs/releases/v0.10.0.md).
+
 Run Scriptorium under an identity that can:
 
 - read only the prompt, profile, schema, and allowed input-artifact paths it
@@ -62,14 +87,26 @@ and apply the same retention and access rules as for model input and output.
 
 Before changing production inputs, profiles, or schemas:
 
-1. confirm the deployed configuration selects the intended sources and model
+1. use `config validate`, `config show`, or `config sources` to check the
+   application configuration and its selected sources without constructing an
+   engine; this does not verify prompt files, provider access, or credentials;
+2. confirm the deployed configuration selects the intended sources and model
    credentials;
-2. use [`render`](cli.md) with the same request inputs and variables to confirm
+3. use [`render`](cli.md) with the same request inputs and variables to confirm
    preparation without a model call;
-3. use [`run`](cli.md) for generation; and
-4. retain or discard validation-failed output according to the caller's
+4. use [`run`](cli.md) for generation; and
+5. retain or discard validation-failed output according to the caller's
    policy.
 
+Inspection output can include configuration paths, names, and ordinary backend
+parameters. `config explain --path` uses JSON Pointer paths and can show
+displaced historical values as well as the final value. Protect both the
+selected paths and the output, especially when backend parameters contain
+sensitive identifiers; ConfigKit does not infer arbitrary secrets in those
+parameters. Apply
+access and retention controls; the application stores only API-key environment
+reference names, not the resolved credential bytes.
+
 The [maintained render script](../examples/render-markdown-summary.sh) is a
 copyable preflight example. The CLI reference owns its complete invocation and
 exit semantics.
@@ -102,6 +139,30 @@ runs. When admission is exhausted, HTTP returns `503 capacity_exceeded` without
 retry timing; callers choose any retry policy that is safe for another model
 call.
 
+Promptkit's built-in client now allows two additional attempts for eligible
+transient provider statuses by default. Account for added calls, latency, and
+cost when planning capacity. Each repair generation has its own transient
+budget, separate from the structural repair count. Configure the engine budget
+with [`transient_retries`](config.md); `run` and `render` accept a request
+override described in the [CLI contract](cli.md), and HTTP callers can select
+one through the [API contract](api.md). Omission inherits the engine
+setting and then Promptkit's default, while explicit zero disables additional
+attempts. Existing generation timeouts apply
+per attempt, while the caller context covers the complete operation. Provider
+`Retry-After` timing can exceed Promptkit's ordinary backoff cap; consult the
+[outbound integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/integrations/openai-compatible-chat.md)
+for details.
+
+CLI `run` can set a [whole-run deadline](cli.md) when total execution time must
+be bounded. The existing generation timeout applies afresh to each attempt and
+does not cap retry waits or the complete invocation. The whole-run interval
+covers the engine call, including capacity waiting and repairs as supported by
+context cancellation; it excludes parsing, engine setup, and output writes.
+HTTP callers can use [`run_timeout_seconds`](api.md) to shorten their request
+context for the same engine interval. The API distinguishes an expired
+execution deadline from caller cancellation and from a provider-attempt
+timeout. A canceled capacity waiter releases its Promptkit admission state.
+
 Before increasing a limit:
 
 1. measure representative input, generated-output, and optional raw-output

docs/oss/attribution.md

@@ -0,0 +1,127 @@
+# Open-Source Attribution
+
+Scriptorium's own terms are in the repository [LICENSE](../../LICENSE).
+The [OSS policy](../policy/oss.md) governs these entries. Dependency versions
+are recorded in [go.mod](../../go.mod) and [go.sum](../../go.sum); source and
+notice links below identify the exact commits for the selected versions.
+
+## Promptkit
+
+Copyright (C) 2026 Eric Rakestraw
+
+SPDX-License-Identifier: LGPL-3.0-or-later
+
+License and notices: [LGPL v3](licenses/LGPL-3.0.txt) |
+[GPL v3 companion](licenses/GPL-3.0.txt) |
+[upstream LICENSE](https://gitea.maximumdirect.net/eric/promptkit/src/commit/db05b56166ca2d8650dfaf660e1cf0eed3932e1d/LICENSE)
+
+Source: [Promptkit v0.11.0](https://gitea.maximumdirect.net/eric/promptkit/src/commit/db05b56166ca2d8650dfaf660e1cf0eed3932e1d)
+
+## ConfigKit
+
+Upstream copyright statement: no project-specific statement is present in the
+selected module; the FSF copyright in its license text belongs to that license
+document and is not substituted for a project notice.
+
+SPDX-License-Identifier: LGPL-3.0-only
+
+License and notices: [LGPL v3](licenses/LGPL-3.0.txt) |
+[GPL v3 companion](licenses/GPL-3.0.txt) |
+[upstream LICENSE](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/e6ced0d3d1194b58e0b3dc8970089471ba175258/LICENSE)
+
+Source: [ConfigKit v0.4.0](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/e6ced0d3d1194b58e0b3dc8970089471ba175258)
+
+## Promptkit OpenRouter Catalog
+
+Upstream copyright statement: no project-specific statement is present in the
+selected module. The linked upstream README records the profile assets'
+Promptkit origin and the copyright holder's deliberate relicensing.
+
+SPDX-License-Identifier: LGPL-3.0-only
+
+License and notices: [LGPL v3](licenses/LGPL-3.0.txt) |
+[GPL v3 companion](licenses/GPL-3.0.txt) |
+[upstream LICENSE](https://gitea.maximumdirect.net/eric/promptkit-backend-openrouter/src/commit/f7518043ebdd70b8959506201639903dff0b85ab/LICENSE) |
+[asset licensing notice](https://gitea.maximumdirect.net/eric/promptkit-backend-openrouter/src/commit/f7518043ebdd70b8959506201639903dff0b85ab/README.md)
+
+Source: [OpenRouter catalog v2.20260922.1](https://gitea.maximumdirect.net/eric/promptkit-backend-openrouter/src/commit/f7518043ebdd70b8959506201639903dff0b85ab)
+
+## Promptkit Rakestrawhome Catalog
+
+Upstream copyright statement: no project-specific statement is present in the
+selected module. The linked upstream README records the profile asset's
+Promptkit origin and the copyright holder's deliberate relicensing.
+
+SPDX-License-Identifier: LGPL-3.0-only
+
+License and notices: [LGPL v3](licenses/LGPL-3.0.txt) |
+[GPL v3 companion](licenses/GPL-3.0.txt) |
+[upstream LICENSE](https://gitea.maximumdirect.net/eric/promptkit-backend-rakestrawhome/src/commit/d664696142bf7c2d93db2a6b2e07f3deb55f2226/LICENSE) |
+[asset licensing notice](https://gitea.maximumdirect.net/eric/promptkit-backend-rakestrawhome/src/commit/d664696142bf7c2d93db2a6b2e07f3deb55f2226/README.md)
+
+Source: [Rakestrawhome catalog v2.20260922.0](https://gitea.maximumdirect.net/eric/promptkit-backend-rakestrawhome/src/commit/d664696142bf7c2d93db2a6b2e07f3deb55f2226)
+
+## jsonschema/v6
+
+Upstream copyright statement: no project-specific statement is present in the
+selected module; no substitute statement is inferred here.
+
+SPDX-License-Identifier: Apache-2.0
+
+License and notices: [Apache 2.0](licenses/Apache-2.0.txt) |
+[upstream LICENSE](https://github.com/santhosh-tekuri/jsonschema/blob/29cbed948d24a04700eb94436416b18a07953b71/LICENSE)
+
+Source: [jsonschema/v6 v6.0.2](https://github.com/santhosh-tekuri/jsonschema/tree/29cbed948d24a04700eb94436416b18a07953b71)
+
+## golang.org/x/text
+
+Copyright (c) 2009 The Go Authors. All rights reserved.
+
+SPDX-License-Identifier: BSD-3-Clause
+
+License and notices: [BSD 3-Clause](licenses/BSD-3-Clause.txt) |
+[upstream LICENSE](https://go.googlesource.com/text/+/6c97a165dd661335ff7bce6104a008558123c353/LICENSE)
+
+Source: [golang.org/x/text v0.14.0](https://go.googlesource.com/text/+/6c97a165dd661335ff7bce6104a008558123c353)
+
+## gopkg.in/yaml.v3
+
+Copyright (c) 2006-2010 Kirill Simonov
+
+Copyright (c) 2006-2011 Kirill Simonov
+
+Copyright (c) 2011-2019 Canonical Ltd
+
+Copyright 2011-2016 Canonical Ltd.
+
+SPDX-License-Identifier: MIT AND Apache-2.0
+
+The upstream LICENSE assigns MIT to files ported from libyaml and Apache 2.0
+to the remaining files; these are file-specific terms, not alternatives.
+
+License and notices: [MIT](licenses/MIT.txt) |
+[Apache 2.0](licenses/Apache-2.0.txt) |
+[upstream LICENSE](https://github.com/go-yaml/yaml/blob/f6f7691b1fdeb513f56608cd2c32c51f8194bf51/LICENSE) |
+[upstream NOTICE](https://github.com/go-yaml/yaml/blob/f6f7691b1fdeb513f56608cd2c32c51f8194bf51/NOTICE)
+
+Source: [gopkg.in/yaml.v3 v3.0.1](https://github.com/go-yaml/yaml/tree/f6f7691b1fdeb513f56608cd2c32c51f8194bf51)
+
+## go.yaml.in/yaml/v4
+
+Copyright 2006-2010 Kirill Simonov
+
+Copyright 2011-2019 Canonical Ltd
+
+Copyright 2025 The go-yaml Project Contributors
+
+SPDX-License-Identifier: MIT AND Apache-2.0
+
+The upstream NOTICE assigns MIT to files ported from libyaml and Apache 2.0
+to the remaining files; these are file-specific terms, not alternatives.
+
+License and notices: [MIT](licenses/MIT.txt) |
+[Apache 2.0](licenses/Apache-2.0.txt) |
+[upstream LICENSE](https://github.com/yaml/go-yaml/blob/1c17b9cee72bf81ae73a8be50fbd7dd3a9985125/LICENSE) |
+[upstream NOTICE](https://github.com/yaml/go-yaml/blob/1c17b9cee72bf81ae73a8be50fbd7dd3a9985125/NOTICE)
+
+Source: [go.yaml.in/yaml/v4 v4.0.0-rc.6](https://github.com/yaml/go-yaml/tree/1c17b9cee72bf81ae73a8be50fbd7dd3a9985125)

docs/oss/licenses/Apache-2.0.txt

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

docs/oss/licenses/BSD-3-Clause.txt

@@ -0,0 +1,11 @@
+Copyright (c) <year> <owner>. 
+
+Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

docs/oss/licenses/GPL-3.0.txt

@@ -0,0 +1,674 @@
+                    GNU GENERAL PUBLIC LICENSE
+                       Version 3, 29 June 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+                            Preamble
+
+  The GNU General Public License is a free, copyleft license for
+software and other kinds of works.
+
+  The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works.  By contrast,
+the GNU General Public License is intended to guarantee your freedom to
+share and change all versions of a program--to make sure it remains free
+software for all its users.  We, the Free Software Foundation, use the
+GNU General Public License for most of our software; it applies also to
+any other work released this way by its authors.  You can apply it to
+your programs, too.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+  To protect your rights, we need to prevent others from denying you
+these rights or asking you to surrender the rights.  Therefore, you have
+certain responsibilities if you distribute copies of the software, or if
+you modify it: responsibilities to respect the freedom of others.
+
+  For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must pass on to the recipients the same
+freedoms that you received.  You must make sure that they, too, receive
+or can get the source code.  And you must show them these terms so they
+know their rights.
+
+  Developers that use the GNU GPL protect your rights with two steps:
+(1) assert copyright on the software, and (2) offer you this License
+giving you legal permission to copy, distribute and/or modify it.
+
+  For the developers' and authors' protection, the GPL clearly explains
+that there is no warranty for this free software.  For both users' and
+authors' sake, the GPL requires that modified versions be marked as
+changed, so that their problems will not be attributed erroneously to
+authors of previous versions.
+
+  Some devices are designed to deny users access to install or run
+modified versions of the software inside them, although the manufacturer
+can do so.  This is fundamentally incompatible with the aim of
+protecting users' freedom to change the software.  The systematic
+pattern of such abuse occurs in the area of products for individuals to
+use, which is precisely where it is most unacceptable.  Therefore, we
+have designed this version of the GPL to prohibit the practice for those
+products.  If such problems arise substantially in other domains, we
+stand ready to extend this provision to those domains in future versions
+of the GPL, as needed to protect the freedom of users.
+
+  Finally, every program is threatened constantly by software patents.
+States should not allow patents to restrict development and use of
+software on general-purpose computers, but in those that do, we wish to
+avoid the special danger that patents applied to a free program could
+make it effectively proprietary.  To prevent this, the GPL assures that
+patents cannot be used to render the program non-free.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+                       TERMS AND CONDITIONS
+
+  0. Definitions.
+
+  "This License" refers to version 3 of the GNU General Public License.
+
+  "Copyright" also means copyright-like laws that apply to other kinds of
+works, such as semiconductor masks.
+
+  "The Program" refers to any copyrightable work licensed under this
+License.  Each licensee is addressed as "you".  "Licensees" and
+"recipients" may be individuals or organizations.
+
+  To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of an
+exact copy.  The resulting work is called a "modified version" of the
+earlier work or a work "based on" the earlier work.
+
+  A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+  To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy.  Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+  To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies.  Mere interaction with a user through
+a computer network, with no transfer of a copy, is not conveying.
+
+  An interactive user interface displays "Appropriate Legal Notices"
+to the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License.  If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+  1. Source Code.
+
+  The "source code" for a work means the preferred form of the work
+for making modifications to it.  "Object code" means any non-source
+form of a work.
+
+  A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+  The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form.  A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+  The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities.  However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work.  For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+  The Corresponding Source need not include anything that users
+can regenerate automatically from other parts of the Corresponding
+Source.
+
+  The Corresponding Source for a work in source code form is that
+same work.
+
+  2. Basic Permissions.
+
+  All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met.  This License explicitly affirms your unlimited
+permission to run the unmodified Program.  The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work.  This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+  You may make, run and propagate covered works that you do not
+convey, without conditions so long as your license otherwise remains
+in force.  You may convey covered works to others for the sole purpose
+of having them make modifications exclusively for you, or provide you
+with facilities for running those works, provided that you comply with
+the terms of this License in conveying all material for which you do
+not control copyright.  Those thus making or running the covered works
+for you must do so exclusively on your behalf, under your direction
+and control, on terms that prohibit them from making any copies of
+your copyrighted material outside their relationship with you.
+
+  Conveying under any other circumstances is permitted solely under
+the conditions stated below.  Sublicensing is not allowed; section 10
+makes it unnecessary.
+
+  3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+  No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+  When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such circumvention
+is effected by exercising rights under this License with respect to
+the covered work, and you disclaim any intention to limit operation or
+modification of the work as a means of enforcing, against the work's
+users, your or third parties' legal rights to forbid circumvention of
+technological measures.
+
+  4. Conveying Verbatim Copies.
+
+  You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+  You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+  5. Conveying Modified Source Versions.
+
+  You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these conditions:
+
+    a) The work must carry prominent notices stating that you modified
+    it, and giving a relevant date.
+
+    b) The work must carry prominent notices stating that it is
+    released under this License and any conditions added under section
+    7.  This requirement modifies the requirement in section 4 to
+    "keep intact all notices".
+
+    c) You must license the entire work, as a whole, under this
+    License to anyone who comes into possession of a copy.  This
+    License will therefore apply, along with any applicable section 7
+    additional terms, to the whole of the work, and all its parts,
+    regardless of how they are packaged.  This License gives no
+    permission to license the work in any other way, but it does not
+    invalidate such permission if you have separately received it.
+
+    d) If the work has interactive user interfaces, each must display
+    Appropriate Legal Notices; however, if the Program has interactive
+    interfaces that do not display Appropriate Legal Notices, your
+    work need not make them do so.
+
+  A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit.  Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+  6. Conveying Non-Source Forms.
+
+  You may convey a covered work in object code form under the terms
+of sections 4 and 5, provided that you also convey the
+machine-readable Corresponding Source under the terms of this License,
+in one of these ways:
+
+    a) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by the
+    Corresponding Source fixed on a durable physical medium
+    customarily used for software interchange.
+
+    b) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by a
+    written offer, valid for at least three years and valid for as
+    long as you offer spare parts or customer support for that product
+    model, to give anyone who possesses the object code either (1) a
+    copy of the Corresponding Source for all the software in the
+    product that is covered by this License, on a durable physical
+    medium customarily used for software interchange, for a price no
+    more than your reasonable cost of physically performing this
+    conveying of source, or (2) access to copy the
+    Corresponding Source from a network server at no charge.
+
+    c) Convey individual copies of the object code with a copy of the
+    written offer to provide the Corresponding Source.  This
+    alternative is allowed only occasionally and noncommercially, and
+    only if you received the object code with such an offer, in accord
+    with subsection 6b.
+
+    d) Convey the object code by offering access from a designated
+    place (gratis or for a charge), and offer equivalent access to the
+    Corresponding Source in the same way through the same place at no
+    further charge.  You need not require recipients to copy the
+    Corresponding Source along with the object code.  If the place to
+    copy the object code is a network server, the Corresponding Source
+    may be on a different server (operated by you or a third party)
+    that supports equivalent copying facilities, provided you maintain
+    clear directions next to the object code saying where to find the
+    Corresponding Source.  Regardless of what server hosts the
+    Corresponding Source, you remain obligated to ensure that it is
+    available for as long as needed to satisfy these requirements.
+
+    e) Convey the object code using peer-to-peer transmission, provided
+    you inform other peers where the object code and Corresponding
+    Source of the work are being offered to the general public at no
+    charge under subsection 6d.
+
+  A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+  A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal, family,
+or household purposes, or (2) anything designed or sold for incorporation
+into a dwelling.  In determining whether a product is a consumer product,
+doubtful cases shall be resolved in favor of coverage.  For a particular
+product received by a particular user, "normally used" refers to a
+typical or common use of that class of product, regardless of the status
+of the particular user or of the way in which the particular user
+actually uses, or expects or is expected to use, the product.  A product
+is a consumer product regardless of whether the product has substantial
+commercial, industrial or non-consumer uses, unless such uses represent
+the only significant mode of use of the product.
+
+  "Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to install
+and execute modified versions of a covered work in that User Product from
+a modified version of its Corresponding Source.  The information must
+suffice to ensure that the continued functioning of the modified object
+code is in no case prevented or interfered with solely because
+modification has been made.
+
+  If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information.  But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+  The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or updates
+for a work that has been modified or installed by the recipient, or for
+the User Product in which it has been modified or installed.  Access to a
+network may be denied when the modification itself materially and
+adversely affects the operation of the network or violates the rules and
+protocols for communication across the network.
+
+  Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+  7. Additional Terms.
+
+  "Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law.  If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+  When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it.  (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.)  You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+  Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders of
+that material) supplement the terms of this License with terms:
+
+    a) Disclaiming warranty or limiting liability differently from the
+    terms of sections 15 and 16 of this License; or
+
+    b) Requiring preservation of specified reasonable legal notices or
+    author attributions in that material or in the Appropriate Legal
+    Notices displayed by works containing it; or
+
+    c) Prohibiting misrepresentation of the origin of that material, or
+    requiring that modified versions of such material be marked in
+    reasonable ways as different from the original version; or
+
+    d) Limiting the use for publicity purposes of names of licensors or
+    authors of the material; or
+
+    e) Declining to grant rights under trademark law for use of some
+    trade names, trademarks, or service marks; or
+
+    f) Requiring indemnification of licensors and authors of that
+    material by anyone who conveys the material (or modified versions of
+    it) with contractual assumptions of liability to the recipient, for
+    any liability that these contractual assumptions directly impose on
+    those licensors and authors.
+
+  All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10.  If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term.  If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+  If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+  Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions;
+the above requirements apply either way.
+
+  8. Termination.
+
+  You may not propagate or modify a covered work except as expressly
+provided under this License.  Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+  However, if you cease all violation of this License, then your
+license from a particular copyright holder is reinstated (a)
+provisionally, unless and until the copyright holder explicitly and
+finally terminates your license, and (b) permanently, if the copyright
+holder fails to notify you of the violation by some reasonable means
+prior to 60 days after the cessation.
+
+  Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+  Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License.  If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+  9. Acceptance Not Required for Having Copies.
+
+  You are not required to accept this License in order to receive or
+run a copy of the Program.  Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance.  However,
+nothing other than this License grants you permission to propagate or
+modify any covered work.  These actions infringe copyright if you do
+not accept this License.  Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+  10. Automatic Licensing of Downstream Recipients.
+
+  Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License.  You are not responsible
+for enforcing compliance by third parties with this License.
+
+  An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations.  If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+  You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License.  For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+  11. Patents.
+
+  A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based.  The
+work thus licensed is called the contributor's "contributor version".
+
+  A contributor's "essential patent claims" are all patent claims
+owned or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version.  For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+  Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+  In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement).  To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+  If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients.  "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+  If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+  A patent license is "discriminatory" if it does not include within
+the scope of its coverage, prohibits the exercise of, or is
+conditioned on the non-exercise of one or more of the rights that are
+specifically granted under this License.  You may not convey a covered
+work if you are a party to an arrangement with a third party that is
+in the business of distributing software, under which you make payment
+to the third party based on the extent of your activity of conveying
+the work, and under which the third party grants, to any of the
+parties who would receive the covered work from you, a discriminatory
+patent license (a) in connection with copies of the covered work
+conveyed by you (or copies made from those copies), or (b) primarily
+for and in connection with specific products or compilations that
+contain the covered work, unless you entered into that arrangement,
+or that patent license was granted, prior to 28 March 2007.
+
+  Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+  12. No Surrender of Others' Freedom.
+
+  If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you may
+not convey it at all.  For example, if you agree to terms that obligate you
+to collect a royalty for further conveying from those to whom you convey
+the Program, the only way you could satisfy both those terms and this
+License would be to refrain entirely from conveying the Program.
+
+  13. Use with the GNU Affero General Public License.
+
+  Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU Affero General Public License into a single
+combined work, and to convey the resulting work.  The terms of this
+License will continue to apply to the part which is the covered work,
+but the special requirements of the GNU Affero General Public License,
+section 13, concerning interaction through a network will apply to the
+combination as such.
+
+  14. Revised Versions of this License.
+
+  The Free Software Foundation may publish revised and/or new versions of
+the GNU General Public License from time to time.  Such new versions will
+be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+  Each version is given a distinguishing version number.  If the
+Program specifies that a certain numbered version of the GNU General
+Public License "or any later version" applies to it, you have the
+option of following the terms and conditions either of that numbered
+version or of any later version published by the Free Software
+Foundation.  If the Program does not specify a version number of the
+GNU General Public License, you may choose any version ever published
+by the Free Software Foundation.
+
+  If the Program specifies that a proxy can decide which future
+versions of the GNU General Public License can be used, that proxy's
+public statement of acceptance of a version permanently authorizes you
+to choose that version for the Program.
+
+  Later license versions may give you additional or different
+permissions.  However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+  15. Disclaimer of Warranty.
+
+  THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
+OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
+IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
+ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+  16. Limitation of Liability.
+
+  IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
+THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
+GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
+USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
+DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
+PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
+EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGES.
+
+  17. Interpretation of Sections 15 and 16.
+
+  If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+                     END OF TERMS AND CONDITIONS
+
+            How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+state the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+Also add information on how to contact you by electronic and paper mail.
+
+  If the program does terminal interaction, make it output a short
+notice like this when it starts in an interactive mode:
+
+    <program>  Copyright (C) <year>  <name of author>
+    This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
+    This is free software, and you are welcome to redistribute it
+    under certain conditions; type `show c' for details.
+
+The hypothetical commands `show w' and `show c' should show the appropriate
+parts of the General Public License.  Of course, your program's commands
+might be different; for a GUI interface, you would use an "about box".
+
+  You should also get your employer (if you work as a programmer) or school,
+if any, to sign a "copyright disclaimer" for the program, if necessary.
+For more information on this, and how to apply and follow the GNU GPL, see
+<https://www.gnu.org/licenses/>.
+
+  The GNU General Public License does not permit incorporating your program
+into proprietary programs.  If your program is a subroutine library, you
+may consider it more useful to permit linking proprietary applications with
+the library.  If this is what you want to do, use the GNU Lesser General
+Public License instead of this License.  But first, please read
+<https://www.gnu.org/licenses/why-not-lgpl.html>.

docs/oss/licenses/LGPL-3.0.txt

@@ -0,0 +1,165 @@
+                   GNU LESSER GENERAL PUBLIC LICENSE
+                       Version 3, 29 June 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <https://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+
+  This version of the GNU Lesser General Public License incorporates
+the terms and conditions of version 3 of the GNU General Public
+License, supplemented by the additional permissions listed below.
+
+  0. Additional Definitions.
+
+  As used herein, "this License" refers to version 3 of the GNU Lesser
+General Public License, and the "GNU GPL" refers to version 3 of the GNU
+General Public License.
+
+  "The Library" refers to a covered work governed by this License,
+other than an Application or a Combined Work as defined below.
+
+  An "Application" is any work that makes use of an interface provided
+by the Library, but which is not otherwise based on the Library.
+Defining a subclass of a class defined by the Library is deemed a mode
+of using an interface provided by the Library.
+
+  A "Combined Work" is a work produced by combining or linking an
+Application with the Library.  The particular version of the Library
+with which the Combined Work was made is also called the "Linked
+Version".
+
+  The "Minimal Corresponding Source" for a Combined Work means the
+Corresponding Source for the Combined Work, excluding any source code
+for portions of the Combined Work that, considered in isolation, are
+based on the Application, and not on the Linked Version.
+
+  The "Corresponding Application Code" for a Combined Work means the
+object code and/or source code for the Application, including any data
+and utility programs needed for reproducing the Combined Work from the
+Application, but excluding the System Libraries of the Combined Work.
+
+  1. Exception to Section 3 of the GNU GPL.
+
+  You may convey a covered work under sections 3 and 4 of this License
+without being bound by section 3 of the GNU GPL.
+
+  2. Conveying Modified Versions.
+
+  If you modify a copy of the Library, and, in your modifications, a
+facility refers to a function or data to be supplied by an Application
+that uses the facility (other than as an argument passed when the
+facility is invoked), then you may convey a copy of the modified
+version:
+
+   a) under this License, provided that you make a good faith effort to
+   ensure that, in the event an Application does not supply the
+   function or data, the facility still operates, and performs
+   whatever part of its purpose remains meaningful, or
+
+   b) under the GNU GPL, with none of the additional permissions of
+   this License applicable to that copy.
+
+  3. Object Code Incorporating Material from Library Header Files.
+
+  The object code form of an Application may incorporate material from
+a header file that is part of the Library.  You may convey such object
+code under terms of your choice, provided that, if the incorporated
+material is not limited to numerical parameters, data structure
+layouts and accessors, or small macros, inline functions and templates
+(ten or fewer lines in length), you do both of the following:
+
+   a) Give prominent notice with each copy of the object code that the
+   Library is used in it and that the Library and its use are
+   covered by this License.
+
+   b) Accompany the object code with a copy of the GNU GPL and this license
+   document.
+
+  4. Combined Works.
+
+  You may convey a Combined Work under terms of your choice that,
+taken together, effectively do not restrict modification of the
+portions of the Library contained in the Combined Work and reverse
+engineering for debugging such modifications, if you also do each of
+the following:
+
+   a) Give prominent notice with each copy of the Combined Work that
+   the Library is used in it and that the Library and its use are
+   covered by this License.
+
+   b) Accompany the Combined Work with a copy of the GNU GPL and this license
+   document.
+
+   c) For a Combined Work that displays copyright notices during
+   execution, include the copyright notice for the Library among
+   these notices, as well as a reference directing the user to the
+   copies of the GNU GPL and this license document.
+
+   d) Do one of the following:
+
+       0) Convey the Minimal Corresponding Source under the terms of this
+       License, and the Corresponding Application Code in a form
+       suitable for, and under terms that permit, the user to
+       recombine or relink the Application with a modified version of
+       the Linked Version to produce a modified Combined Work, in the
+       manner specified by section 6 of the GNU GPL for conveying
+       Corresponding Source.
+
+       1) Use a suitable shared library mechanism for linking with the
+       Library.  A suitable mechanism is one that (a) uses at run time
+       a copy of the Library already present on the user's computer
+       system, and (b) will operate properly with a modified version
+       of the Library that is interface-compatible with the Linked
+       Version.
+
+   e) Provide Installation Information, but only if you would otherwise
+   be required to provide such information under section 6 of the
+   GNU GPL, and only to the extent that such information is
+   necessary to install and execute a modified version of the
+   Combined Work produced by recombining or relinking the
+   Application with a modified version of the Linked Version. (If
+   you use option 4d0, the Installation Information must accompany
+   the Minimal Corresponding Source and Corresponding Application
+   Code. If you use option 4d1, you must provide the Installation
+   Information in the manner specified by section 6 of the GNU GPL
+   for conveying Corresponding Source.)
+
+  5. Combined Libraries.
+
+  You may place library facilities that are a work based on the
+Library side by side in a single library together with other library
+facilities that are not Applications and are not covered by this
+License, and convey such a combined library under terms of your
+choice, if you do both of the following:
+
+   a) Accompany the combined library with a copy of the same work based
+   on the Library, uncombined with any other library facilities,
+   conveyed under the terms of this License.
+
+   b) Give prominent notice with the combined library that part of it
+   is a work based on the Library, and explaining where to find the
+   accompanying uncombined form of the same work.
+
+  6. Revised Versions of the GNU Lesser General Public License.
+
+  The Free Software Foundation may publish revised and/or new versions
+of the GNU Lesser General Public License from time to time. Such new
+versions will be similar in spirit to the present version, but may
+differ in detail to address new problems or concerns.
+
+  Each version is given a distinguishing version number. If the
+Library as you received it specifies that a certain numbered version
+of the GNU Lesser General Public License "or any later version"
+applies to it, you have the option of following the terms and
+conditions either of that published version or of any later version
+published by the Free Software Foundation. If the Library as you
+received it does not specify a version number of the GNU Lesser
+General Public License, you may choose any version of the GNU Lesser
+General Public License ever published by the Free Software Foundation.
+
+  If the Library as you received it specifies that a proxy can decide
+whether future versions of the GNU Lesser General Public License shall
+apply, that proxy's public statement of acceptance of any version is
+permanent authorization for you to choose that version for the
+Library.

docs/oss/licenses/MIT.txt

@@ -0,0 +1,18 @@
+MIT License
+
+Copyright (c) <year> <copyright holders>
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software and
+associated documentation files (the "Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the
+following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or substantial
+portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT
+LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO
+EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+USE OR OTHER DEALINGS IN THE SOFTWARE.

docs/policy/agents.md

@@ -0,0 +1,77 @@
+# Agent Policy
+
+**Base Policy Version:** 1.1
+
+**Last Updated:** 2026-09-26
+
+## Purpose
+
+Agents working in this repository must read and adhere to the following house rules.
+
+## Implementation Discipline
+
+**KEY PRINCIPLES:** Write the simplest idiomatic code that satisfies the user's
+request. Do not implicitly strengthen requirements, restrictions, or guarantees
+beyond what the user actually requested.
+
+**Keep implementation machinery proportionate.** Use straightforward,
+idiomatic code and introduce only the machinery needed to support the requested
+behavior and flexibility. Optimize for conceptual simplicity and
+maintainability, not for the fewest lines, fewest options, or narrowest product
+surface.
+
+**Do not invent restrictions.** Do not reject, prohibit, constrain, cap,
+harden, or otherwise narrow behavior unless required by the user, an explicit
+repository contract, or an unavoidable correctness constraint. Preserve
+operator and caller choice where policy has not been established. If an
+additional restriction appears advisable, present it as a decision rather than
+silently implementing it.
+
+**Do not invent guarantees.** Do not strengthen public contracts, guarantees,
+or invariants beyond those expressly requested by the user.
+
+**Use the standard library.** For example: in almost every case, the best
+approach to copying files is to use the standard library. Writing custom code to
+implement power-loss protection at the application level, or to guard against
+bad actors with privileged and concurrent access to the host's filesystem,
+is wasteful, unnecessary, and not idiomatic. Stick to the simple, standard, and
+idiomatic approach unless another path is expressly requested by the user.
+
+**An implementation detail is not repository policy.** Tests and documentation
+must not elevate incidental behavior into a durable guarantee or restriction
+unless that behavior was expressly selected as part of the contract.
+
+## Git Commit Policy
+
+By default, create a commit after completing each coherent unit of work, using a
+concise, plain-English message describing what changed. This policy is standing
+authorization to create those local commits; a separate user request is not
+required. An explicit user instruction to leave work uncommitted takes precedence.
+
+A unit of work is a focused, reviewable outcome, including its relevant tests and
+documentation. Do not commit after every intermediate edit or create empty commits.
+Complete the appropriate validation before committing. If work is incomplete or
+validation is blocked, report that state rather than presenting it as completed.
+
+Before committing:
+
+1. Run `git status --short` and review the relevant diff, including any staged
+   changes. Identify changes that predate the current unit of work.
+2. Exclude secrets, local environment files, generated junk, and unrelated changes.
+3. Stage only the files or hunks belonging to the completed unit. Use `git add .`
+   only when every included change belongs to that unit. Do not disturb unrelated
+   staged or unstaged work.
+4. Review `git diff --staged` and run appropriate checks for the change. Do not
+   repeat checks already passed for unchanged content merely to create a commit.
+5. Review `git log --oneline -n 10` and use
+   `git commit -m "<concise plain-English commit message>"`.
+6. Report the commit identifier and any remaining uncommitted work.
+
+If the unit cannot be safely separated from pre-existing work, leave it uncommitted
+and explain the overlap rather than silently including that work. If a commit
+fails, preserve the changes and report the failure; do not bypass hooks.
+
+Default commit authorization does not authorize pushing, publishing, amending,
+resetting, reverting, or rewriting history. Those actions require a separate user
+instruction. Do not use `git commit -a` or `git commit --no-verify` to bypass the
+staging and validation discipline above.

docs/policy/architecture.md

@@ -1,7 +1,21 @@
-# Architecture
+# Architecture Policy
 
-This document defines Scriptorium's current application architecture and
-durable development boundaries.
+**Base Policy Version:** 1.1
+
+**Last Updated:** 2026-09-26
+
+## Purpose And Ownership
+
+This policy defines Scriptorium's purpose, responsibility and dependency
+boundaries, established invariants, and non-goals. Concrete component inventory
+and implementation mechanics belong in [internal documentation](../internal/overview.md);
+public behavior belongs in its canonical contracts.
+
+Adopted constraints govern future development without implying that every
+capability they could govern exists or is planned. Keep proposed structure and
+implementation status in roadmaps, and decision rationale in ADRs. Acceptance
+of a decision is not proof of implementation. Do not turn incidental choices
+into architectural invariants.
 
 ## System Shape
 
@@ -11,16 +25,32 @@ The `inspect` command has prompt and profile modes. Scriptorium does not expose
 a reusable root Go package.
 
 The application consumes
-[Promptkit v0.9.0](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/consumers/pkg-promptkit.md)
+[Promptkit v0.11.0](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md)
 through its supported root package. Promptkit owns prompt execution,
 preparation, source formats, built-in profiles, model-client behavior, and
 validation. Scriptorium owns application configuration, executable adapters,
 prepared-run and definition-inspection presentation, process behavior, and HTTP
 deployment policy.
 
+Scriptorium also consumes ConfigKit v0.4.0 for application configuration
+decoding, resolution, and validation flow. Scriptorium owns its application
+schema, defaults, normalization, validation rules, source selection, and CLI
+flag mapping; ConfigKit does not own Promptkit framework sources or HTTP
+artifact containment.
+
 The concrete package inventory is maintained in the
 [internal overview](../internal/overview.md).
 
+As established by [ADR 0004](../adr/0004-definition-boundary.md), Scriptorium
+aims to support the complete set of features in valid Promptkit prompt and
+profile definitions for its selected upstream version. Promptkit remains the
+canonical parser and validator. Library-only features are considered
+individually against the simple predefined-prompt execution model.
+
+Maintain adapter boundaries with proportionate, idiomatic code. This approach
+does not require an interface for every type, additional wrappers, or a fixed
+directory tree for integrations that do not exist.
+
 ## Dependency Direction
 
 ```text
@@ -30,7 +60,7 @@ cmd/scriptorium
 CLI and HTTP adapters, configuration, defaults, and formatting
         |
         v
-gitea.maximumdirect.net/eric/promptkit
+Promptkit (execution) and ConfigKit (application configuration)
 ```
 
 - Retained application packages may import Promptkit's root package.
@@ -46,21 +76,25 @@ gitea.maximumdirect.net/eric/promptkit
 The repository architecture guard enforces these import and removal
 invariants.
 
+Cross-project changes land and release in Promptkit before Scriptorium adopts
+the tagged dependency. Do not commit Go workspaces, local replacements,
+vendored Promptkit implementation, or unpublished dependency assumptions.
+
 ## Retained Boundaries
 
 - `internal/adapter/cli` owns commands, flags, configuration precedence,
   process streams, output files, summaries, and exit codes.
 - `internal/adapter/http` owns routes, strict JSON DTOs, size limits, response
   mapping, status mapping, and the restricted artifact reader.
-- `internal/config` owns discovery and strict decoding of Scriptorium
-  application configuration.
+- `internal/config` owns discovery and Scriptorium's application definition;
+  ConfigKit owns strict YAML decoding and the load/override pipeline.
 - `internal/defaults` owns Scriptorium application and HTTP defaults only.
 - `internal/format` owns deterministic prepared-run and definition-inspection
   text and JSON presentation.
 - Promptkit owns framework orchestration and contracts. Its
-  [format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/formats.md)
+  [format reference](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md)
   and
-  [outbound integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.9.0/docs/integrations/openai-compatible-chat.md)
+  [outbound integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/integrations/openai-compatible-chat.md)
   are canonical.
 
 ## HTTP Artifact Security Boundary
@@ -71,6 +105,13 @@ That reader denies file references without an artifact root, enforces the
 configured byte limit, and applies Scriptorium's lexical root-containment rule.
 The operating system still follows symlinks after the lexical check.
 
+This boundary handles untrusted request references under the documented
+deployment model; it is not a filesystem sandbox. Defending against a
+privileged actor concurrently modifying the host filesystem is outside the
+adopted scope, not a claim that such activity cannot occur. Do not infer
+additional defenses or guarantees from hypothetical threats; raise unsettled
+scope decisions under the [agent policy](agents.md).
+
 The [HTTP API](../api.md) owns observable request outcomes, and
 [operations](../operations.md) owns deployment permissions and root selection.
 
@@ -80,6 +121,14 @@ Scriptorium has no durable run-state store, checkpoint, cache, or resume
 mechanism. Recovery is a new request after correcting inputs, configuration, or
 environment.
 
+Each request is self-contained from the caller's perspective, without
+conversation or workflow state retained from completed requests. This does
+not mean the process is internally stateless. Shared engines, backend capacity
+and queue accounting, in-flight coordination, cancellation, and server
+lifecycle resources are permitted and must be shared where necessary to
+enforce configured backend limits across concurrent requests. Such state must
+be concurrency-safe and does not create a durable user-workflow contract.
+
 Adapters map Promptkit public error identities into CLI exits or HTTP statuses
 without classifying by message text. Raw API keys are not accepted in
 Scriptorium configuration, CLI arguments, or HTTP payloads, and resolved
@@ -102,5 +151,7 @@ secrets must not be emitted.
   into Scriptorium.
 - Do not move CLI, inbound HTTP, process, or deployment policy into Promptkit.
 - Do not add durable workflow, archive, or resume behavior.
+- Do not add application-managed prompt modification, follow-up message flows,
+  conversation management, or multi-step workflow orchestration.
 
 Work that is not implemented belongs in `docs/roadmap/`.

docs/policy/documentation.md

@@ -1,5 +1,9 @@
 # Documentation Policy
 
+**Base Policy Version:** 1.4
+
+**Last Updated:** 2026-09-26
+
 ## Purpose
 
 This policy assigns each documentation topic to one canonical owner. Its goal is
@@ -30,10 +34,14 @@ semantics.
 Outside `docs/roadmap/`, documentation describes implemented behavior only.
 Partial features may be described only to their implemented boundary.
 
-ADRs are the narrow exception: an ADR may record an accepted architectural
-decision before implementation, but acceptance must not be presented as proof
-that the behavior exists. The roadmap owns implementation status and sequencing
-until the decision is implemented. Current architecture, user, operator,
+Architectural constraints may govern future changes without implying that the
+components or capabilities they would govern already exist. This does not
+permit describing a proposed system structure as implemented.
+
+ADRs may record an accepted architectural decision before implementation, but
+acceptance must not be presented as proof that the behavior exists. The 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.
 
 ### Audience And Detail
@@ -63,16 +71,22 @@ secret values.
 
 | Topic | Canonical owner | Owned content | Content owned elsewhere |
 | --- | --- | --- | --- |
-| Product orientation and minimal end-to-end quickstart | `README.md` | What this project is, why it is useful, one shortest successful invocation, and links onward. | Complete command reference, configuration reference, operational procedures, implementation detail. |
+| Product orientation and minimal end-to-end quickstart | `README.md` | What this project is, why it is useful, one shortest successful invocation, and links onward, including license and attribution navigation. | Complete command reference, configuration reference, operational procedures, implementation detail. |
 | Contributor entry point | `docs/development.md` | Task-oriented reading guide, minimal contributor orientation, baseline validation commands, and links to canonical docs. | Package inventory, architecture rules, subsystem behavior, and detailed change recipes, which belong in the relevant internal component document. |
-| Current application architecture | `docs/policy/architecture.md` | System shape, normative ownership, dependency direction, architectural boundaries, invariants, safety properties, and non-goals. | Concrete package inventory, implementation mechanics, contributor procedures, decision history, future work. |
+| Application architecture policy | `docs/policy/architecture.md` | System shape, normative ownership, dependency direction, architectural boundaries, invariants, safety properties, and non-goals. | Concrete package inventory, implementation mechanics, contributor procedures, decision history, future work. |
+| Agent policy | `docs/policy/agents.md` | Agent implementation discipline, proportionality, idiomatic simplicity, local commit rules, and rules against inventing requirements, restrictions, guarantees, or repository policy. | Application contracts, architecture, testing policy, and contributor procedures. |
 | Documentation organization | `docs/policy/documentation.md` | Documentation ownership, audience boundaries, maintenance rules, and ADR/document lifecycle. | Application architecture or product behavior. |
 | Testing policy | `docs/policy/testing.md` | Test philosophy, risk-based sufficiency, test boundaries, doubles, coverage guidance, regression-test policy, and criteria for adding, rewriting, or deleting tests. | Subsystem behavior, application contracts, subsystem-specific test inventories, and implementation plans. |
 | CLI contract | `docs/cli.md` | Commands, arguments, flags, invocation semantics, and exit codes. | End-to-end operating procedures, configuration field definitions, runtime filesystem layout, module implementation details. |
 | Configuration contract | `docs/config.md` | Application discovery and precedence, source locations, server fields, render default, HTTP limits, and credential mapping. | Promptkit framework formats and defaults, complete example files, CLI syntax, runtime lifecycle, and implementation detail. |
 | Operations | `docs/operations.md` | Runtime workflows, physical filesystem and state layout, output, cache, and debug handling, resume, cleanup, permissions, recovery, and operational limits. | CLI flag syntax, configuration field definitions, logical output schemas, implementation mechanics. |
-| Release procedure | `docs/release.md` | Candidate validation, version and tag operations, hosted-workflow observation, and published-artifact verification. | Runtime operations, version-specific announcements, and complete application-interface contracts. |
+| Release policy | `docs/policy/release.md` | Versioning and compatibility, release integrity and immutability, documentation requirements, proportionate gates, publication controls, source completion, and optional verification boundaries. | Repository-specific commands, installation guidance, historical summaries, and CI implementation details. |
+| Release procedure | `docs/release.md` | Maintainer prerequisites, candidate preparation, publication command and guards, asynchronous delivery, optional diagnostics, and failure recovery. | General release requirements, runtime operations, historical announcements, and current application contracts. |
 | Version-specific release notes | `docs/releases/` | Immutable release summaries, compatibility notices, and migration announcements for one published version. | Complete CLI, HTTP, configuration, operations, or dependency contracts. |
+| Open source licensing and attribution policy | `docs/policy/oss.md` | House attribution format, license-text and notice-link conventions, discoverability, maintenance, and handling of concrete exceptions. | Component entries, license texts, release commands, and dependency security review. |
+| Project license | `LICENSE` | The project's own license terms. | Upstream component attribution and dependency license texts. |
+| Upstream component attribution | `docs/oss/attribution.md` | Component names, copyright statements, license identification, and license, notice, and source links. | General attribution policy and complete standard license texts. |
+| Standard dependency license texts | `docs/oss/licenses/` | Complete, verbatim standard license texts shared by attribution entries where applicable. | Project-specific upstream notices, attribution prose, and policy interpretation. |
 | Public HTTP contract | `docs/api.md` | Routes, authentication, media types, request and response schemas, status codes, pagination, caching, idempotency, rate limits, and HTTP retry semantics. | Client walkthroughs, upstream or downstream integration internals, implementation detail. |
 | Consumer guidance | `docs/consumers/` | Choosing between Scriptorium's executable interfaces and understanding consumer responsibilities. | HTTP wire semantics, CLI syntax, Promptkit's Go package, and internal implementation detail. |
 | External and durable integration contracts | `docs/integrations/` | Scriptorium-owned process and executable integration contracts. | Promptkit framework formats and outbound provider protocols, physical runtime placement, internal transformations, CLI syntax, and configuration defaults. |
@@ -103,6 +117,22 @@ runtime state and how to operate or recover the application. When a workflow
 crosses these topics, choose the document that owns the task and link to the
 other contracts.
 
+### Releases
+
+The release policy owns release requirements; the release procedure supplies
+the commands that satisfy them. Release notes are historical summaries, not
+current-state contract owners. The checked-in note at an immutable tag records
+that release; current documentation changes with the behavior it describes.
+Neither a procedure nor a note may weaken policy or defer required updates to
+canonical contracts.
+
+### Open Source Attribution
+
+The [OSS policy](oss.md) defines the House Format and README and release-link
+conventions. Attribution entries own component-specific information. README
+and release documents provide navigation to those canonical materials; the
+release procedure owns URL construction and publication mechanics.
+
 ### Contracts And Implementation
 
 Integration and API documents define externally observable shapes and
@@ -150,6 +180,11 @@ ADR; rejected product ideas belong in the roadmap.
 
 ## Maintenance
 
+Every policy document records its underlying template version and local
+last-updated date immediately below the title. Update the date when revising
+local policy; change the base version only when adopting a newer template.
+Do not retain template adoption instructions in the local policy.
+
 When behavior changes, update its canonical owner in the same change. If
 ownership moves, remove the old definition and replace it with a link where
 navigation remains useful.

docs/policy/oss.md

@@ -0,0 +1,139 @@
+# Open Source Licensing And Attribution Policy
+
+**Base Policy Version:** 1.1
+
+**Last Updated:** 2026-09-26
+
+## Purpose And Scope
+
+This policy establishes a repeatable House Format for open source attribution
+and license notices. It covers upstream components incorporated into or
+distributed with the project, including relevant transitive dependencies and
+copied code or assets. Treatment follows the component's terms and actual use,
+rather than its classification as a direct dependency.
+
+The baseline is a publicly available source repository with releases published
+alongside it. The format is independent of language, forge, and build system.
+It does not require attribution for every development tool, establish license
+compatibility, or resolve obligations beyond attribution and notice handling.
+
+## Canonical Locations
+
+The [documentation policy](documentation.md) assigns ownership:
+
+- `LICENSE` contains the project's own license. Dependency attribution does
+  not relicense the project.
+- `docs/oss/attribution.md` contains the canonical component entries.
+- `docs/oss/licenses/` contains complete, verbatim standard license texts,
+  including companion texts where applicable.
+
+Use recognizable license filenames, such as `Apache-2.0.txt` or `LGPL-3.0.txt`.
+A shared standard text may serve multiple entries. Keep explanations outside
+license texts and preserve distinct versions and exceptions.
+
+## House Format
+
+Each component entry has five fields:
+
+1. **Project name**, normally as the entry heading.
+2. **Upstream copyright statement or statements**, preserving their wording.
+3. **SPDX license identifier or expression**, when applicable.
+4. **License and notice links**, covering the applicable standard texts and
+   upstream project-specific material.
+5. **Source repository link**, identifying the source corresponding to the
+   imported version.
+
+Illustrative Markdown, with placeholders for the actual upstream information:
+
+```markdown
+## Example Library
+
+Copyright <upstream copyright statement>
+
+SPDX-License-Identifier: Apache-2.0
+
+License and notices: [Apache 2.0](licenses/Apache-2.0.txt) | [NOTICE](<upstream NOTICE URL at the imported commit>)
+
+Source: [Example Library](<upstream repository URL at the imported commit>)
+```
+
+These are five fields, not a line limit. Include multiple statements and links
+where needed. Omit the SPDX field when no applicable identifier exists; identify
+nonstandard terms accurately through the linked text. Preserve licensing
+alternatives and exceptions, and record a selected option when relevant. Do not
+invent an identifier or silently resolve an unsettled licensing choice.
+
+## License Texts And Upstream Links
+
+Link standard license texts to their local copies. Link project-specific
+notices to the upstream copies instead of storing them locally. Every upstream
+repository or notice link must identify the exact commit corresponding to the
+imported version, including the selected source revision for copied material.
+Do not use moving branches or version tags as substitutes for that commit.
+
+Classify material by its contents, not its filename. An upstream `LICENSE`,
+`COPYING`, or similar file may contain project-specific notices or exceptions
+in addition to standard wording. Link both the local standard text and the
+commit-specific upstream file when needed. Make companion license texts directly
+reachable from the entry rather than leaving readers to infer their location.
+
+For project-specific license and attribution material, this policy accepts
+direct links to the applicable upstream texts as satisfying reproduction,
+inclusion, or copy requirements. This is the adopted house interpretation.
+Agents must apply it without adding local-copy requirements or seeking repeated
+approval solely because a license uses those terms.
+
+This interpretation concerns notice reproduction. Concrete obligations such as
+marking modified files or supplying corresponding source receive their own
+applicable treatment. Preserve required notices already present in incorporated
+source files; the central attribution entry does not authorize stripping them.
+
+## Discoverability
+
+End the project's README with two friendly statements linking to its actual
+license and the attribution document. For Scriptorium:
+
+```markdown
+Scriptorium is licensed under the [BSD 3-Clause License](LICENSE).
+
+[Open Source attribution notices](docs/oss/attribution.md) are available.
+```
+
+Each new checked-in release document and its hosted download page must end
+with links whose visible text is `LICENSE | ATTRIBUTION`. Use absolute URLs to
+those files at the release's immutable tag or corresponding immutable source
+revision so the links work in both locations. This release navigation is distinct
+from upstream component links, which must use the imported commit.
+
+Reuse the checked-in release document as the hosted release body where supported.
+The release procedure owns concrete URLs and publication mechanics for delivering
+the links through the selected channel. Do not maintain duplicate notice lists.
+These conventions do not by themselves require separate downloadable license
+assets, source bundles, or a different binary package format.
+
+## Maintenance And Exceptions
+
+Update notices in the change that adds, removes, upgrades, copies, or changes
+the distribution of a relevant component. Check the selected source's license
+and notices rather than assuming they are unchanged. Update affected entries,
+commit links, and texts; retain historical versions through Git.
+
+Check affected statements, faithful standard-text copies, and link destinations
+as part of ordinary change review. For a future release tag, check intended file
+paths against the candidate source; the tag need not resolve remotely before
+publication. Do not add publication-time network checks, CI polling, recurring
+legal sign-offs, mandatory scanners, or a full software bill of materials merely
+to implement this policy.
+
+Use the House Format routinely. Raise an exception only for a concrete obligation
+or unresolved rights question it does not address, including any arising from a
+particular distribution channel. Record the issue and selected treatment in the
+adopted policy or affected attribution entry as appropriate. Resolve the issue
+before the affected use or distribution without reopening settled choices.
+Common ownership of projects is neither required nor a blanket exception.
+
+Apply the formatting convention prospectively. Do not rewrite published tags,
+notes, binaries, or checksums to retrofit it. Address actual omissions deliberately
+under the applicable terms and the [release correction policy](release.md#immutability-and-failure-recovery).
+Ordinary attribution maintenance and release links are preparation work, not new
+release approval gates.

docs/policy/release.md

@@ -0,0 +1,213 @@
+# Release Policy
+
+**Base Policy Version:** 1.2
+
+**Last Updated:** 2026-09-26
+
+## Purpose
+
+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.
+
+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.
+
+## Release Model And Completion
+
+A source release identifies one exact committed source revision through an
+immutable version tag and a version-matched, checked-in release note. The
+primary release branch is `main`; the authoritative publication remote is
+`origin`.
+
+A successful push of the intended tag, with its source commit available on the
+authoritative primary branch, completes the release. Publish `main` before
+invoking the publication command; the command requires
+that the remote branch exactly identifies the validated candidate and pushes
+only the selected annotated tag.
+
+Release commands must finish at successful publication. They must not wait for
+or poll CI, require successful CI status, inspect downloadable artifacts, or
+check consumer installation or Go module availability as completion gates.
+A failed or ambiguous push requires resolving the authoritative remote state
+before declaring success or retrying; successful pushes do not require a
+separate remote read-back gate.
+
+Scriptorium publishes Linux binaries and checksums asynchronously from 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.
+
+Artifact delivery is a separate status. A successful tag push is not evidence
+that artifacts are available. Delivery failures do not reverse source-release
+success, and optional investigation must not delay release completion.
+
+## Versioning And Compatibility
+
+Stable releases use Semantic Versioning in the form `vMAJOR.MINOR.PATCH`, with
+no leading zeroes. Prereleases are not part of the adopted release channel.
+
+The selected version must be unused locally and on the authoritative remote.
+Do not require queries to every configured remote or publication service.
+A version known to have already been published must not be reused, even if the
+normal guards do not inspect that channel. Adding a distribution channel
+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 patch release may make backward-compatible fixes.
+
+Before `v1.0.0`, a minor release may change a documented public, operational,
+configuration, data, integration, or compatibility contract when the release
+note explains the impact and required consumer action. A patch release must
+not intentionally break those contracts within its minor line.
+
+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.
+
+## 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.
+
+Include a short summary and any relevant compatibility impact or upgrade
+instructions. Keep the note proportionate to the change. Fixed sections and
+explicit no-action boilerplate are not required. Existing historical notes
+need not be rewritten to adopt this format.
+
+Include the release footer defined by the [OSS policy](oss.md#discoverability)
+in each new release document and preserve it on the hosted download page. The
+release procedure defines the concrete URLs and publication mechanics. Prepare
+these links with the release note; they do not add a network check or approval
+gate.
+
+Keep exact current contracts in their canonical documents and update affected
+documentation with the behavior change. Link to those documents from the note
+where useful. A release note must not substitute for current documentation.
+Use annotated tags with their message derived from the checked-in note rather
+than requiring a separately authored account of the same release.
+
+Version selection, release-note content, and retirement of completed roadmaps
+are preparation responsibilities, not separate human approval gates. Preserve
+planned and deferred work and repair affected links when retiring roadmaps.
+Ordinary change review covers unintended files, credentials, and private data;
+do not repeat it as a release-specific sign-off. Ignored local files are not
+prohibited merely because they exist: they are not included in the tagged source.
+
+## Candidate Validation
+
+Provide one publication command that validates the final committed candidate,
+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.
+
+Run the complete deterministic, offline default test suite once for the final
+candidate, following the [testing policy](testing.md), and 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.
+
+The selected test invocation includes the race detector because the HTTP
+service shares engines and backend capacity state across concurrent requests.
+This single invocation covers both ordinary assertions and race detection.
+Build both declared Linux architectures with the hosted release flags to catch
+cross-compilation and packaging failures; native test hardware for each target
+is not required. Existing architecture tests enforce dependency boundaries.
+
+Retain narrow source guards against committed workspaces, vendor trees, and
+module replacements: these can otherwise make a passing local build depend on
+unpublished or substituted framework code. Disable workspace discovery and
+vendor mode during validation. These guards preserve the established
+cross-project boundary; they do not require exact dependency-version audits or
+module-metadata rewrites at release time.
+
+Additional gates must address a concrete risk that the existing checks do not
+adequately cover. Do not automatically require linting, race testing, module
+hygiene audits, documentation audits, credential scans, example validation, or
+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.
+
+Release tests must not require real credentials, mutable application services,
+paid APIs, provider probes, or inference calls. The procedure describes
+dependency preparation separately from offline validation.
+
+Avoid duplicate validation across release phases. Successful checks for the
+same final candidate need not be repeated solely to satisfy another phase;
+any reuse must clearly apply to that exact candidate. Persistent validation
+caches are not required. If the candidate changes, validate the new candidate.
+Cheap publication-state checks may be repeated immediately before pushing
+because local or remote state may have changed during validation.
+
+Automate mechanical guards. Human review gates are strongly disfavored: add one
+only through an explicit project-specific policy decision explaining the
+concrete risk and why ordinary change review or automation is insufficient.
+Do not add routine confirmations for checks the command can perform itself.
+
+## Publication Guards
+
+Automate the checks needed to publish the intended source under a fresh version:
+
+- the version has the adopted format and is unused locally and on the
+  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;
+- the remote primary branch already identifies that exact candidate; and
+- the tag identifies the validated candidate and only the intended refs are
+  published.
+
+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
+refs rather than all local tags. Do not require CI success, publication-service
+availability checks, or a separate manual confirmation for each guard.
+
+## Immutability And Failure Recovery
+
+Published version tags and published contents are immutable. Do not move,
+overwrite, recreate, or delete a published tag, or replace a published artifact,
+checksum, signature, provenance statement, or release note under that version.
+The checked-in note at the immutable tag is that release's historical record.
+
+If validation fails before publication, correct and commit the candidate and
+rerun the publication command. If a push fails or its outcome is ambiguous,
+inspect the authoritative remote before retrying. Remove an unpublished local
+tag only after confirming it was never published or used by a release channel.
+An unavailable remote leaves publication status unknown; do not overwrite a tag
+to resolve uncertainty.
+
+Transient network, authentication, CI, or consumer-availability failures do not
+by themselves require a new version. Asynchronous delivery may be retried from
+the same immutable tag. Missing artifacts may be published later, provided
+existing published contents are not replaced and the additions remain consistent
+with the tagged source and any already-published checksum manifest. If that
+cannot be achieved, publish a new version.
+
+A defect requiring changes to published source or contents requires a corrected
+commit, a new semantically appropriate version, a matching release note, and
+the release procedure again. Leave the original release intact. A channel's
+yank, deprecation, or advisory mechanism may warn consumers away from a defective
+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:
+
+1. prerequisites, primary branch, and authoritative remote;
+2. version and release-note preparation;
+3. one publication command with its selected local checks and automated guards;
+4. declared deliverables, required build targets, and automation responsibilities;
+5. the successful-push completion point and separately reported delivery status;
+   and
+6. failure handling, ambiguous-push recovery, and asynchronous delivery retries.
+
+Use copyable commands where practical. Keep optional diagnostics and consumer
+verification clearly optional. The procedure must not turn preparation work,
+ordinary change review, or post-publication investigation into additional
+release gates.

docs/policy/testing.md

@@ -1,5 +1,9 @@
 # Testing Policy
 
+**Base Policy Version:** 1.1
+
+**Last Updated:** 2026-09-26
+
 ## Purpose
 
 Our tests exist to make **incorrect changes expensive and correct changes cheap**.
@@ -8,29 +12,16 @@ We do not optimize for test count, line coverage, exhaustive isolation, or the f
 
 ## Every test has a cost
 
-Testing is not an unqualified good. Every test imposes both an immediate cost and a continuing lifetime cost.
-
-A test must be:
-
-- written and reviewed;
-- understood by future maintainers and coding agents;
-- executed in local and CI workflows;
-- diagnosed when it fails;
-- updated when legitimate behavior changes;
-- maintained as fixtures, APIs, and dependencies evolve; and
-- removed or rewritten when it becomes redundant, brittle, misleading, or obsolete.
-
-Tests also create cognitive and architectural friction. They can constrain refactoring, duplicate policy, slow feedback loops, add noise to failures, and cause harmless implementation changes to require unrelated edits across the suite.
+Every test has an immediate and continuing cost: writing, review, execution,
+diagnosis, fixture maintenance, and updates when legitimate behavior changes.
+Tests can also constrain refactoring, duplicate policy, and add failure noise.
 
 A test is warranted only when the confidence it provides justifies these costs.
-
-Apply this cost-benefit analysis at two levels:
+Evaluate that tradeoff at two levels:
 
 1. **Per test:** What realistic defect does this test detect, how consequential would that defect be, and is that protection worth the test's lifetime cost?
 2. **Across the suite:** Does this collection provide materially more confidence than a smaller, simpler suite would?
 
-The preferred test suite is a **lean suite that provides sufficient confidence in the risks that matter, without redundant or low-value tests**. We seek sufficient confidence with the least unnecessary testing friction, not the fewest possible tests.
-
 Some friction is intentional. Tests should make dangerous changes—such as breaking compatibility, corrupting data, violating security boundaries, or reintroducing subtle bugs—require deliberate review. They should not make ordinary internal changes needlessly expensive.
 
 The cost of a test is not a reason to omit testing by default. Do not cite maintenance cost abstractly. When omitting a plausible test, be able to state why the protected failure is low-risk, already covered, obvious, reversible, or cheaper to detect elsewhere. For consequential, subtle, or difficult-to-observe behavior, the presumption should favor testing.
@@ -55,7 +46,7 @@ Control clocks, randomness, environment variables, and other process-global or m
 
 ## What deserves tests
 
-Prioritize tests for:
+To the extent the following features are present in the application, prioritize tests for:
 
 1. Public and package-level contracts.
 2. Domain rules and important invariants.
@@ -67,7 +58,18 @@ Prioritize tests for:
 
 A package-level contract is behavior relied upon by another package or major collaborator, not every observable detail of a package implementation.
 
-For behavior involving **data integrity, destructive operations, compatibility, security, concurrency, idempotency, or recovery**, presume that durable tests are required unless the behavior is already credibly protected at another layer.
+Choose tests for requested behavior and established contracts. Tests must not
+introduce new restrictions, guarantees, or threat scenarios merely because they
+could be asserted. Follow the [agent policy](agents.md) when deciding whether a
+behavior belongs to the contract; current implementation behavior alone is not
+sufficient authority.
+
+For established behavior involving **data integrity, destructive operations,
+compatibility, security, concurrency, idempotency, or recovery**, favor durable
+tests for realistic failures whose consequences warrant protection. Identify
+the failure and the existing contract it would violate. These categories call
+for attention, not an automatic test inventory or permission to strengthen the
+application's guarantees. Account for protection already provided at other layers.
 
 Do not add tests merely because a function, branch, or line exists. Do not add a test when the same meaningful risk is already adequately protected elsewhere.
 
@@ -75,9 +77,16 @@ Do not add tests merely because a function, branch, or line exists. Do not add a
 
 Test through the narrowest stable boundary that expresses the behavior clearly.
 
+For application configuration, ConfigKit owns exhaustive YAML grammar,
+source traversal, import/overlay composition, provenance, and inspection
+renderer tests. Scriptorium tests its selected definition, source discovery,
+CLI override presence, command prerequisites, runtime conversion, credential
+boundary, and representative execution/inspection workflows with the tagged
+library. Do not copy the upstream parser test matrix into this repository.
+
 This is often the package API, but it may instead be:
 
-- a smaller pure function when dense application logic is most clearly isolated there;
+- a smaller pure function when dense domain logic is most clearly isolated there;
 - a package-level operation when several internal collaborators jointly produce the behavior; or
 - a larger integration boundary when correctness emerges from interaction with a real dependency.
 
@@ -85,11 +94,7 @@ Do not force all behavior through oversized end-to-end tests. Do not test every
 
 ## Test behavior, not implementation
 
-A test should protect a decision, contract, or invariant—not memorialize the current implementation.
-
-Before adding or retaining a test, ask:
-
-> What realistic defect would this test catch?
+A test should protect a decision, contract, or invariant. A test must not memorialize the current implementation.
 
 A test is suspect when its main purpose is to detect that someone:
 
@@ -102,8 +107,6 @@ A test is suspect when its main purpose is to detect that someone:
 
 Refactoring should normally require no test edits unless the refactored structure is itself part of the contract.
 
-A test can be factually correct and still have negative value. Accurately describing current behavior is not enough; the protected behavior must be important enough to justify the future friction.
-
 ## Expected effects of different changes
 
 Use the following expectations when evaluating test failures and test maintenance:
@@ -121,10 +124,11 @@ A test failing is not the same as a test needing to be edited. Many tests may co
 
 Configurable thresholds and defaults must not be duplicated throughout the test suite.
 
-For example, do not encode an internal concurrency limit indirectly:
+For this example, assume the selected contract rejects a request to start more
+processes than a configured limit. Do not encode its production default indirectly:
 
 ```go
-// Production policy:
+// Production default:
 const maxConcurrency = 4
 
 // Brittle test:
@@ -142,18 +146,23 @@ require.NoError(t, runner.Start(limit))
 require.ErrorIs(t, runner.Start(limit+1), ErrTooMuchConcurrency)
 ```
 
-The test should prove:
+For this selected contract, the test should prove:
 
 - the configured limit is accepted; and
 - one beyond the configured limit is rejected.
 
+Rejection is not a general requirement for concurrency control. If the selected
+contract queues excess work or limits active workers while accepting larger
+batches, test that behavior instead. Do not add rejection or a configurable
+limit solely to reproduce this example.
+
 The production default should be tested exactly only when its literal value is itself a public, operational, safety, protocol, or compatibility requirement.
 
 Apply the same rule to limits, timeouts, capacities, retry counts, and ranges: test relationships and behavior, not duplicated literals.
 
-For concurrency limits, test both kinds of behavior when relevant:
+For established concurrency limits, distinguish these behaviors where the contract requires them:
 
-1. **Configuration enforcement:** invalid or excessive requested values are handled correctly.
+1. **Configuration handling:** values are accepted or rejected according to the selected contract.
 2. **Runtime enforcement:** observed peak concurrency never exceeds the configured limit.
 
 Use a test-controlled limit and measure the behavior relative to that limit. Do not merely assert today's default value.
@@ -162,21 +171,21 @@ Use a test-controlled limit and measure the behavior relative to that limit. Do
 
 Each behavior should have a clear test owner.
 
-- Configuration tests own application YAML, discovery, precedence, and
-  application defaults.
+- Configuration tests own application YAML, strict decoding, discovery,
+  precedence, and application defaults.
 - CLI tests own argument mapping, streams, summaries, exit behavior, and
   representative command workflows.
 - HTTP tests own DTOs, strict decoding, limits, status mapping, and restricted
   artifact policy.
-- Formatter tests own prepared-run text and JSON presentation.
+- Formatter tests own prepared-run and definition-inspection text and JSON
+  presentation.
 - Architecture tests own dependency direction and removal invariants.
 - Promptkit owns framework parsing, orchestration, validation, profiles, and
-  model-client behavior.
+  model-client behavior. Scriptorium's definition compatibility fixtures protect
+  its adapter plumbing without duplicating the upstream format contract.
 
 Higher-level tests should not repeat every lower-level case. A single intentional policy change should not require unrelated edits across many test files.
 
-Tests that are individually reasonable may still be collectively redundant. Evaluate the marginal value of each additional test in light of the protection already provided by the rest of the suite.
-
 ## Use test doubles deliberately
 
 Choose the least elaborate test double that provides the required control or observation.
@@ -200,15 +209,16 @@ Do not use mocks merely to isolate every object or reproduce the implementation'
 
 ## Go-specific guidance
 
-Use:
+Select techniques according to the behavior and risk being tested. The following
+are options, not a checklist that every package or application must satisfy:
 
 - table-driven tests for meaningful behavioral categories and boundaries;
 - `t.TempDir()` for real filesystem behavior;
 - `httptest.Server` for realistic HTTP interactions;
-- fuzz tests for parsers, normalization, path handling, and broad input spaces;
+- fuzz tests when exploring broad input spaces, such as parsers or path handling, adds useful protection beyond representative cases;
 - golden files only when the complete output is intentionally stable;
 - integration tests where correctness depends on component interaction; and
-- a small number of representative end-to-end tests.
+- a small number of representative end-to-end tests when assembled workflows expose meaningful risks not adequately covered at narrower boundaries.
 
 Avoid exact error-string assertions unless the wording is itself contractual. Prefer `errors.Is`, `errors.As`, typed errors, or structured error fields.
 
@@ -228,8 +238,6 @@ Security-sensitive HTTP containment and external mappings may warrant denser
 coverage than straightforward process wiring. Uneven coverage is acceptable
 when it reflects risk.
 
-Increasing coverage is valuable only when the newly covered behavior protects a meaningful risk at an acceptable cost.
-
 ## Regression tests
 
 A bug fix should normally include a regression test that fails before the fix and passes afterward.
@@ -258,44 +266,26 @@ Strong candidates include tests that:
 
 Several brittle tests may encode one genuine requirement. Replace them with one durable behavior-level or invariant test rather than preserving all of them.
 
-Deleting a low-value test can improve the quality of the suite by reducing noise, maintenance burden, and friction around legitimate change.
-
 ## Reviewing a proposed test
 
 Use the following questions when the value, boundary, or durability of a proposed test is not self-evident. Significant test additions should be reviewable against them, but written answers are not required for every routine test.
 
-1. What realistic defect would it catch?
-2. How likely is that defect?
-3. How consequential would it be?
-4. Is the behavior already protected elsewhere?
-5. At which layer should this behavior be owned?
-6. Does the test assert a durable contract or an incidental implementation detail?
-7. Could the implementation be refactored without changing the behavior and without editing this test?
-8. What should cause this test to fail?
-9. What legitimate changes should not cause this test to fail?
-10. What ongoing maintenance, execution, and diagnostic cost will the test impose?
-11. Is there a smaller or more direct test that protects the same risk?
-
-Do not add the test when its expected lifetime cost exceeds its expected protective value.
-
-When deciding not to test plausible behavior, record or be able to explain why the risk is low, already protected, obvious, reversible, or cheaper to detect elsewhere.
+1. Which requested behavior or established contract does it protect, and what realistic defect would violate it?
+2. How likely and consequential is that defect, and is it already protected elsewhere?
+3. What is the narrowest stable boundary that can test it clearly?
+4. Would a correct internal refactor or a change to an incidental default require test edits?
+5. Does its confidence justify its maintenance, execution, and diagnostic cost, or would a smaller test suffice?
 
 ## Definition of sufficient
 
 A test suite is sufficient when:
 
-- important contracts and invariants are protected;
-- meaningful boundaries and failure modes are exercised;
-- realistic and consequential regressions are credibly protected against silent recurrence;
-- behavior involving data integrity, destructive operations, compatibility, security, concurrency, idempotency, and recovery is credibly protected;
-- important external boundaries have realistic integration coverage;
-- representative complete workflows are tested;
+- important requested behavior and established contracts are protected against realistic, consequential failures;
+- boundaries and component interactions have coverage where they introduce meaningful risks not protected elsewhere;
 - failures provide useful signal rather than redundant noise;
 - legitimate internal changes usually do not require test edits; and
 - additional tests would mostly repeat existing protection or preserve inconsequential implementation details.
 
-Sufficiency is a risk judgment, not a coverage percentage or test count. Reassess it as the application, its users, and the consequences of failure evolve.
-
-The governing rule is:
-
-> Test heavily where failure is consequential, subtle, or difficult to detect after the fact. Test lightly where failure is obvious, reversible, and inexpensive—and retain no test whose lifetime cost exceeds the confidence it provides.
+Sufficiency is a risk judgment, not a coverage percentage, test count, or required
+mix of testing techniques. Reassess it as the application, its users, and the
+consequences of failure evolve.

docs/release.md

@@ -1,377 +1,143 @@
 # Release Procedure
 
-## Release Model And Status
+The [release policy](policy/release.md) owns versioning, compatibility,
+publication completion, and immutability. This runbook supplies Scriptorium's
+commands; release notes under `docs/releases/` are historical summaries. The
+[OSS policy](policy/oss.md#discoverability) owns license and attribution
+discoverability requirements.
 
-Scriptorium publishes annotated semantic tags and tag-triggered Linux binary
-releases. The hosted
-[release workflow](../.woodpecker/release.yml) builds `amd64` and `arm64`
-executables, publishes their SHA-256 checksums, and uses the matching file
-under `docs/releases/` as the hosted release body.
+## Prepare The Candidate
 
-`v0.12.0` is the first published application-only release. For each later
-release, select a new `vMAJOR.MINOR.PATCH` version according to the intended
-compatibility change. A selected version remains an unreleased candidate until
-its annotated tag is published, the hosted workflow succeeds, and every
-published artifact is verified.
-
-Run this procedure from the Scriptorium repository root. A release must not
-depend on a Go workspace, module replacement, vendor tree, sibling checkout,
-unpublished dependency, or unpushed source commit.
-
-## Establish The Candidate
-
-Start a POSIX shell, choose a semantic version that has not been published, and
-export it as `RELEASE_VERSION`. For example, if `v0.12.1` is the intended next
-version and remains unpublished, select:
+Use a POSIX shell, Git with a configured tagger identity and access to
+`origin`, and a Go toolchain compatible with `go.mod`. The native host must
+support Go's race detector and have its required C compiler. If Git is
+configured to sign tags, its signing tool and key must be available. Run from the
+repository root. Provision the toolchain and module cache before offline
+validation; when dependencies are not cached:
 
 ```sh
-export RELEASE_VERSION=v0.12.1
+GOWORK=off go mod download
 ```
 
-Use the version appropriate to the actual compatibility change rather than
-assuming that the example is the next release. Then run the following guard in
-that same shell:
+Choose an unused stable version under the release policy. Commit its note as
+`docs/releases/<version>.md`, beginning with a title such as
+`# Scriptorium v0.14.0` (an example, not an instruction to select that version).
+Include a concise summary and relevant compatibility impact or upgrade action.
+End the note with the following footer, replacing `<version>` in both absolute
+URLs with the selected version tag (for example, `v0.14.0`):
+
+```markdown
+[LICENSE](https://gitea.maximumdirect.net/eric/scriptorium/src/tag/<version>/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/scriptorium/src/tag/<version>/docs/oss/attribution.md)
+```
+
+Check that both URLs name files in the candidate tree at that tag. The tag may
+not resolve online until publication. The checked-in note is copied unchanged
+into the hosted release body by the existing workflow, so do not maintain a
+second footer there.
+Update affected canonical documentation with the change, and retire completed
+roadmap material when appropriate while preserving deferred work and links.
+No extra release-specific review signoff is required.
+
+Finish and commit the intended candidate on `main`, then publish that branch:
 
 ```sh
-set -eu
-
-: "${RELEASE_VERSION:?export an unpublished vMAJOR.MINOR.PATCH version}"
-if ! printf '%s\n' "$RELEASE_VERSION" |
-    grep -Eq '^v(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)$'
-then
-    printf '%s\n' "invalid release version: $RELEASE_VERSION" >&2
-    exit 1
-fi
-
-RELEASE_COMMIT=$(git rev-parse --verify 'HEAD^{commit}')
-export RELEASE_COMMIT
-
-check_release_candidate() {
-    test "$(git branch --show-current)" = main
-    test -z "$(git status --porcelain)"
-
-    gowork_value=$(go env GOWORK)
-    case "$gowork_value" in
-        ''|off) ;;
-        *)
-            printf '%s\n' "active Go workspace: $gowork_value" >&2
-            return 1
-            ;;
-    esac
-
-    test -z "$(git ls-files go.work go.work.sum)"
-    test ! -e vendor
-    if grep -Eq '^[[:space:]]*replace([[:space:]]|\()' go.mod
-    then
-        printf '%s\n' 'go.mod contains a replacement' >&2
-        return 1
-    fi
-
-    git fetch origin main --tags
-    test "$RELEASE_COMMIT" = \
-        "$(git rev-parse --verify 'refs/remotes/origin/main^{commit}')"
-
-    if git show-ref --verify --quiet "refs/tags/$RELEASE_VERSION"
-    then
-        printf '%s\n' "local tag already exists: $RELEASE_VERSION" >&2
-        return 1
-    fi
-    if test -n "$(
-        git ls-remote --tags origin \
-            "refs/tags/$RELEASE_VERSION" \
-            "refs/tags/$RELEASE_VERSION^{}"
-    )"
-    then
-        printf '%s\n' "remote tag already exists: $RELEASE_VERSION" >&2
-        return 1
-    fi
-}
-
-check_release_candidate
+git -c push.followTags=false push origin refs/heads/main:refs/heads/main
 ```
 
-Do not continue unless this guard succeeds. It deliberately requires the
-candidate to be the exact clean commit already published at `origin/main`.
+The publication command inspects the configured push endpoint of `origin`,
+not a potentially stale remote-tracking ref. If a separate push URL is
+configured, it is authoritative for both checks and publication. Configure one
+publication URL; multiple push endpoints cannot be validated as one remote.
 
-## Verify Modules And Repository Boundaries
+## Publish
 
-Confirm the module path and declared Go version:
+Invoke the single [publication command](../scripts/release.sh), substituting
+the selected version:
 
 ```sh
-test "$(
-    GOWORK=off go list -m -f '{{.Path}} {{.GoVersion}}'
-)" = 'gitea.maximumdirect.net/eric/scriptorium 1.25.5'
+sh scripts/release.sh v0.14.0
 ```
 
-Require Promptkit `v0.9.0` as both the direct module-graph edge and the selected
-module version:
+The command validates a clean worktree and index on `main`, an unused local and
+remote version, exact agreement with authoritative remote `main`, and a
+checked-in note whose first-level title contains Scriptorium and the exact
+version. Ignored local files do not make the candidate dirty.
+
+The existing source-boundary guards reject committed workspaces, vendor trees,
+and module replacements. No migration-specific exact Promptkit version audit
+is required.
+
+It then runs the complete suite once with the race detector and builds Linux
+amd64 and arm64 binaries using `CGO_ENABLED=0`, `-trimpath`, and
+`-ldflags '-s -w'`. The native test run uses the host platform and enables cgo.
+Validation disables workspaces, uses read-only module mode, and disables module
+and toolchain downloads. Missing dependencies must be provisioned separately;
+fix the environment and rerun the command. No persistent validation cache is
+maintained by the release tool.
+
+The command rechecks candidate and publication state, creates an annotated tag
+from the committed note, and pushes only that tag's exact object to its explicit
+ref. It does not push branches, follow other tags, force updates, or wait for
+CI. Temporary notes and validation builds are removed on exit. The successful
+summary identifies the source commit and tag object and reports binary
+delivery separately as asynchronous and unverified.
+
+## Asynchronous Binary Delivery
+
+The tag-triggered [hosted workflow](../.woodpecker/release.yml) builds and
+publishes:
+
+- `scriptorium-<version>-linux-amd64`;
+- `scriptorium-<version>-linux-arm64`; and
+- `SHA256SUMS`.
+
+The matching checked-in note supplies the hosted release body. The workflow
+owns image selection, packaging, checksum generation, and upload configuration.
+Its upload step skips existing files and does not overwrite the release.
+Source-publication success does not attest to workflow success or asset
+availability.
+
+Workflow inspection, downloads, checksum verification, and a compatible-host
+smoke test are optional delivery diagnostics. They must not delay or redefine
+source-release completion. No polling or consumer-installation check is part
+of the publication command.
+
+## Failure Recovery
+
+Before tag creation, correct and commit a failed candidate, publish `main`,
+and invoke the publication command again. A changed candidate must pass its
+own validation.
+
+After an unsuccessful push, the command queries the authoritative endpoint.
+If it confirms the exact tag object and candidate, it reports source success.
+Otherwise it exits unsuccessfully and preserves the local tag. It never retries
+a push or deletes a tag automatically. If the endpoint is unavailable, the
+publication status is unknown, not evidence that nothing was published.
+
+For manual diagnosis, select the affected version and compare the local tag
+object and peeled commit with the authoritative remote:
 
 ```sh
-direct_promptkit=$(
-    GOWORK=off go mod graph |
-        awk '
-            $1 == "gitea.maximumdirect.net/eric/scriptorium" &&
-            $2 ~ /^gitea\.maximumdirect\.net\/eric\/promptkit@/ {
-                print $2
-            }
-        '
-)
-test "$direct_promptkit" = \
-    'gitea.maximumdirect.net/eric/promptkit@v0.9.0'
-test "$(
-    GOWORK=off go list -m -f '{{.Path}}@{{.Version}}' \
-        gitea.maximumdirect.net/eric/promptkit
-)" = 'gitea.maximumdirect.net/eric/promptkit@v0.9.0'
-GOWORK=off go list -m all
+release_version=v0.14.0
+release_remote=$(git remote get-url --push origin)
+git rev-parse "refs/tags/$release_version" "refs/tags/$release_version^{commit}"
+git ls-remote "$release_remote" refs/heads/main \
+    "refs/tags/$release_version" "refs/tags/$release_version^{}"
 ```
 
-Require tidy module metadata and recheck the repository exclusions:
+Do not retry while the outcome remains unknown. If the tag exists remotely,
+do not move, recreate, or delete it. If the tag was never published or used by
+any release channel, an explicitly authorized local cleanup may delete only
+that unpublished tag; then rerun the command to validate and publish the
+candidate. The command intentionally rejects an existing local tag rather than
+assuming that a previous process validated it.
 
-```sh
-GOWORK=off go mod tidy -diff
-test -z "$(git ls-files go.work go.work.sum)"
-test ! -e vendor
-if grep -Eq '^[[:space:]]*replace([[:space:]]|\()' go.mod
-then
-    printf '%s\n' 'go.mod contains a replacement' >&2
-    exit 1
-fi
-test -z "$(git status --porcelain)"
-```
-
-## Validate The Application
-
-Run the complete application validation:
-
-```sh
-GOWORK=off go test ./...
-GOWORK=off go test -race ./...
-GOWORK=off go vet ./...
-validation_build_dir=$(mktemp -d)
-GOWORK=off go build \
-    -o "$validation_build_dir/scriptorium" \
-    ./cmd/scriptorium
-```
-
-The ordinary test run includes the architecture guard that rejects a root Go
-package, former framework package families, and imports of Promptkit internal
-packages. Inspect the repository for generated binaries, credentials,
-temporary output, sibling paths, and other files that do not belong in the
-tracked release source.
-
-Check every tracked Go file. This command must produce no output:
-
-```sh
-unformatted=$(
-    git ls-files '*.go' |
-        while IFS= read -r go_file
-        do
-            gofmt -l "$go_file"
-        done
-)
-test -z "$unformatted"
-```
-
-Run the maintained render and Promptkit v0.9 feature scripts, then smoke-test
-both maintained configuration examples without a model call:
-
-```sh
-GOWORK=off ./examples/render-markdown-summary.sh
-GOWORK=off ./examples/render-v0.9-features.sh
-for config_file in examples/config.yml examples/config.full.yml
-do
-    GOWORK=off go run ./cmd/scriptorium render \
-        --config "$config_file" \
-        --prompt generic.markdown_summary \
-        --input transcript=./examples/fixtures/transcript.md \
-        --input glossary=./examples/fixtures/glossary.yml \
-        --format json >/dev/null
-done
-```
-
-Exercise usage output and offline rendering with the temporary native
-executable:
-
-```sh
-usage_output="$validation_build_dir/usage.txt"
-if "$validation_build_dir/scriptorium" >"$usage_output" 2>&1
-then
-    printf '%s\n' 'expected an invocation without a command to fail' >&2
-    exit 1
-fi
-grep -F 'usage: scriptorium' "$usage_output"
-"$validation_build_dir/scriptorium" render \
-    --config ./examples/config.yml \
-    --prompt generic.markdown_summary \
-    --input transcript=./examples/fixtures/transcript.md \
-    --input glossary=./examples/fixtures/glossary.yml \
-    --format text >/dev/null
-```
-
-Follow every maintained local, Promptkit-tagged, and other external Markdown
-link. Confirm that all repository-relative link targets exist. Finish the
-application checks with:
-
-```sh
-git diff --check
-test -z "$(git status --porcelain)"
-```
-
-## Require Release Notes And Reproduce Packaging
-
-The immutable release note must exist before tagging:
-
-```sh
-release_notes="docs/releases/$RELEASE_VERSION.md"
-test -f "$release_notes"
-test -s "$release_notes"
-```
-
-Validate every local and currently published link in the note. For links
-pinned to the candidate Scriptorium tag, confirm that the corresponding
-repository-relative path exists even though its tag URL is not live yet.
-
-Reproduce the hosted build flags, targets, and filenames in a temporary
-directory:
-
-```sh
-release_dist=$(mktemp -d)
-release_package='gitea.maximumdirect.net/eric/scriptorium/cmd/scriptorium'
-
-build_release_binary() {
-    target_os="$1"
-    target_arch="$2"
-    output="$release_dist/scriptorium-$RELEASE_VERSION-$target_os-$target_arch"
-
-    CGO_ENABLED=0 GOOS="$target_os" GOARCH="$target_arch" GOWORK=off \
-        go build -trimpath -ldflags '-s -w' \
-        -o "$output" "$release_package"
-}
-
-build_release_binary linux amd64
-build_release_binary linux arm64
-
-test -s "$release_dist/scriptorium-$RELEASE_VERSION-linux-amd64"
-test -s "$release_dist/scriptorium-$RELEASE_VERSION-linux-arm64"
-file "$release_dist/scriptorium-$RELEASE_VERSION-linux-amd64"
-file "$release_dist/scriptorium-$RELEASE_VERSION-linux-arm64"
-```
-
-Require `file` to identify Linux executables for `x86-64` and `ARM aarch64`,
-respectively. Inspect the
-[hosted workflow](../.woodpecker/release.yml) and confirm that it uses the
-same build flags and names, copies the selected release note to
-`dist/RELEASE_NOTES.md`, publishes only `dist/scriptorium-*`, and keeps
-checksum generation enabled.
-
-## Create And Publish The Tag
-
-Run the candidate guard again immediately before creating the tag:
-
-```sh
-check_release_candidate
-test -f "$release_notes"
-test -s "$release_notes"
-```
-
-Create an annotated tag explicitly bound to the validated commit, using the
-version-specific release note as its message:
-
-```sh
-git tag --annotate "$RELEASE_VERSION" \
-    --file "$release_notes" \
-    "$RELEASE_COMMIT"
-```
-
-Inspect the tag and require it to resolve to the validated source:
-
-```sh
-test "$(git cat-file -t "refs/tags/$RELEASE_VERSION")" = tag
-git show --no-patch --decorate "refs/tags/$RELEASE_VERSION"
-test "$(
-    git rev-parse --verify "refs/tags/$RELEASE_VERSION^{commit}"
-)" = "$RELEASE_COMMIT"
-```
-
-If inspection finds an error, delete only the unpublished local tag, correct
-the candidate, and repeat the complete validation. Never move or recreate a
-published tag.
-
-Push only the selected tag ref:
-
-```sh
-git push origin \
-    "refs/tags/$RELEASE_VERSION:refs/tags/$RELEASE_VERSION"
-```
-
-## Observe And Verify Publication
-
-Open the hosted workflow run for the selected tag. Require
-`build-release-assets` to succeed before `publish-release`, then require the
-publication step and hosted release to succeed. A queued, running, failed, or
-partially published workflow is not a verified release.
-
-Compare the local and remote annotated-tag objects and their source commits:
-
-```sh
-remote_tag=$(
-    git ls-remote --tags origin "refs/tags/$RELEASE_VERSION" |
-        awk 'NR == 1 { print $1 }'
-)
-remote_commit=$(
-    git ls-remote --tags origin "refs/tags/$RELEASE_VERSION^{}" |
-        awk 'NR == 1 { print $1 }'
-)
-test -n "$remote_tag"
-test "$remote_tag" = \
-    "$(git rev-parse --verify "refs/tags/$RELEASE_VERSION")"
-test "$remote_commit" = "$RELEASE_COMMIT"
-```
-
-Download the hosted binaries and checksum file into a temporary directory:
-
-```sh
-release_base="https://gitea.maximumdirect.net/eric/scriptorium/releases/download/$RELEASE_VERSION"
-download_dir=$(mktemp -d)
-(
-    cd "$download_dir"
-    for asset in \
-        "scriptorium-$RELEASE_VERSION-linux-amd64" \
-        "scriptorium-$RELEASE_VERSION-linux-arm64" \
-        SHA256SUMS
-    do
-        curl --fail --location --remote-name "$release_base/$asset"
-    done
-
-    test -s "scriptorium-$RELEASE_VERSION-linux-amd64"
-    test -s "scriptorium-$RELEASE_VERSION-linux-arm64"
-    test -s SHA256SUMS
-    sha256sum --check SHA256SUMS
-    test "$(wc -l < SHA256SUMS | tr -d ' ')" = 2
-    file "scriptorium-$RELEASE_VERSION-linux-amd64"
-    file "scriptorium-$RELEASE_VERSION-linux-arm64"
-)
-```
-
-Require the same Linux architectures observed in the local packaging check and
-confirm that the hosted release contains no unexpected asset. On a compatible
-Linux host, make the matching downloaded binary executable and repeat the
-usage-output and offline-render smoke checks against it.
-
-Only after the tag, workflow, release body, binaries, architectures, and
-checksums all pass verification is the candidate a verified published release.
-
-## Handle Failures
-
-Before tag publication, correct the release commit or note and restart the
-complete procedure. After tag publication, never delete, move, overwrite, or
-recreate the tag. A transient hosted failure may be retried only against the
-same immutable tag and commit and only when doing so cannot overwrite or
-silently retain partial assets. A source, packaging, note, or artifact defect
-requires a new corrective semantic version from a new validated commit.
-
-Record the selected version, validated commit, tag object, workflow result,
-artifact names, checksum result, and smoke-check outcome in the release
-checkpoint. Keep temporary builds and downloaded assets outside the repository,
-and require a clean `main` synchronized with `origin/main` when verification
-is complete.
+For transient delivery failures, inspect what has already been uploaded before
+retrying the same immutable tag. The upload step's skip behavior alone does not
+prove checksum consistency. Add missing assets only if they match the tagged
+source and any already-published manifest, without replacing existing contents.
+Compare rebuilt bytes with published checksums when recovering a partial
+upload; do not assume toolchain changes reproduce identical binaries. If safe
+completion is impossible, or published source, notes, packaging, or artifact
+contents need correction, leave the original release intact and publish a new
+version from a corrected commit.

docs/releases/v0.14.0.md

@@ -0,0 +1,66 @@
+# Scriptorium v0.14.0
+
+This minor release adopts Promptkit v0.11.0 and ConfigKit v0.4.0, adds
+application-configuration inspection, and exposes request retry controls,
+whole-run deadlines, and optional terminal execution reports.
+
+## Configuration migration
+
+Scriptorium now uses ConfigKit for application configuration. This intentionally
+changes accepted input and precedence:
+
+- An existing empty file is invalid; use `{}` for an empty mapping. Explicit
+  empty strings and zeroes replace lower values, while `null` is accepted only
+  for nullable fields. `--config ''` is invalid. An explicitly empty server
+  address reaches the standard HTTP server rather than restoring `:8080`.
+- YAML decoding is stricter: unknown fields, additional documents,
+  anchors/aliases, unsupported scalar forms, and other inputs rejected by
+  ConfigKit now fail. The `configkit` key is reserved, and ConfigKit's source
+  and resource limits apply. Backend parameter integers use its `int64` shape.
+- File-backed configurations can explicitly import path or glob fragments and
+  select one named overlay with `--config-overlay`. These files stay under the
+  selected configuration file's canonical directory. Application path *values*
+  remain relative to the process working directory. Only explicitly supplied
+  application CLI flags override file or overlay settings.
+
+Use `config validate`, `config show`, `config sources`, and `config explain` to
+inspect application settings without creating an engine or contacting a model.
+Validation is not a provider probe or complete Promptkit definition check;
+effective output is not guaranteed reloadable. Inspection keeps ordinary
+`extra_params` values, including historical values, and `api_key_env` reference
+names visible. Keep secrets out of parameters and protect inspection output;
+actual API-key values remain with Promptkit and are never resolved by ConfigKit.
+Request retry overrides and Promptkit credential requirements are separate
+from application configuration.
+
+## Promptkit and execution changes
+
+Promptkit v0.11.0 uses its `/v2` catalogs. The `deepseek-3-2`,
+`deepseek-4-flash`, and `deepseek-4-pro` profile IDs are removed without aliases;
+select an available catalog profile, a maintained Promptkit profile, or a
+complete application-owned profile.
+
+Promptkit's built-in transient retry default is two additional attempts. This
+can increase provider calls, latency, and cost; an application or request
+retry override of zero disables additional attempts. Repair generations have
+their own retry budgets. Generation timeouts apply per attempt. CLI
+`--run-timeout` and HTTP `run_timeout_seconds` add optional whole-run deadlines;
+an expired HTTP execution returns `504 deadline_exceeded`, caller cancellation
+returns `408 request_canceled`, and a live-context provider-attempt failure
+remains a 502 response.
+
+CLI `--execution-report` and HTTP `include_execution_report` expose optional
+terminal reports. Reports are absent when no terminal observation exists, can
+contain sensitive identifiers, and do not replace result usage. CLI report
+writes are separate and nontransactional; HTTP report bytes count toward the
+response-size limit.
+
+For complete behavior, see the
+[CLI](https://gitea.maximumdirect.net/eric/scriptorium/src/tag/v0.14.0/docs/cli.md),
+[configuration](https://gitea.maximumdirect.net/eric/scriptorium/src/tag/v0.14.0/docs/config.md),
+[HTTP API](https://gitea.maximumdirect.net/eric/scriptorium/src/tag/v0.14.0/docs/api.md),
+and [operations](https://gitea.maximumdirect.net/eric/scriptorium/src/tag/v0.14.0/docs/operations.md)
+references. The repository now includes
+[dependency attribution and license texts](https://gitea.maximumdirect.net/eric/scriptorium/src/tag/v0.14.0/docs/oss/attribution.md).
+
+[LICENSE](https://gitea.maximumdirect.net/eric/scriptorium/src/tag/v0.14.0/LICENSE) | [ATTRIBUTION](https://gitea.maximumdirect.net/eric/scriptorium/src/tag/v0.14.0/docs/oss/attribution.md)

docs/roadmap/configkit.md

@@ -0,0 +1,231 @@
+# ConfigKit Adoption
+
+## Status And Intent
+
+Implemented feature direction, drafted on 2026-09-26 and delivered through the
+[implementation plan](implementation.md). This roadmap records the target
+behavior and changes from the reviewed baseline; the implementation plan owns
+sequencing, concrete interfaces, and verification evidence.
+
+The user is adopting ConfigKit across an application suite to standardize
+configuration semantics. Scriptorium will make a clean cutover to ConfigKit's
+native conventions, including intentional breaking changes. Preserving the
+legacy loader's behavior is not a goal: there will be no parallel legacy loader,
+compatibility mode, or recreation of the old merging rules.
+
+Alignment applies to application configuration, not to Promptkit-owned prompt,
+profile, or schema definitions. Departures from ConfigKit conventions require an
+application-specific reason or a genuine ownership boundary, not backward
+compatibility alone. The [architecture policy](../policy/architecture.md) and
+[ADR 0004](../adr/0004-definition-boundary.md) remain the scope boundaries.
+
+## Reviewed Baseline (Historical)
+
+Scriptorium consumes Promptkit v0.11.0. Its `internal/config` package uses
+yaml.v3 directly and manually applies defaults, file settings, and CLI overrides.
+It has a parser-specific integer decoder, ignores empty string overrides, and
+treats an empty configuration file as defaults-only input. It has no application
+configuration imports, selected overlays, or provenance/inspection interface.
+The [configuration reference](../config.md), [CLI reference](../cli.md), and
+[source internals](../internal/sources.md) own the current behavior.
+
+The reviewed ConfigKit release is v0.4.0 at commit
+`e6ced0d3d1194b58e0b3dc8970089471ba175258`. Adopt that published release as the
+initial target, rather than a sibling checkout or unpublished replacement.
+Its canonical references are:
+
+- [Configuration contract](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/e6ced0d3d1194b58e0b3dc8970089471ba175258/docs/config.md).
+- [Consumer adoption guide](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/e6ced0d3d1194b58e0b3dc8970089471ba175258/docs/consumers/README.md).
+- [Release notes](https://gitea.maximumdirect.net/eric/ConfigKit/src/commit/e6ced0d3d1194b58e0b3dc8970089471ba175258/docs/releases/v0.4.0.md).
+
+ConfigKit requires Go 1.25, within Scriptorium's existing toolchain requirement.
+It selects `go.yaml.in/yaml/v4 v4.0.0-rc.6`. Removing Scriptorium's direct parser
+usage does not remove yaml.v3 from the dependency graph: Promptkit still uses it.
+The initial review passed ConfigKit's default test suite and Scriptorium's
+configuration and CLI tests; that is baseline evidence, not migration validation.
+
+## Target Ownership And Loading Model
+
+| Owner | Responsibility |
+| --- | --- |
+| ConfigKit | Application YAML policy, source acquisition, composition, overlays, typed decoding, explicit override application, provenance, and redacted inspection. |
+| Scriptorium | Source discovery, application schema and defaults, CLI parsing, recorded application normalization, semantic validation, command prerequisites, and runtime conversion. |
+| Promptkit | Framework definition formats, backend semantics, credential lookup, engine defaults, and prompt execution. |
+
+`internal/config` becomes a small application-specific ConfigKit integration,
+not a second general-purpose configuration framework. Execution and inspection
+share one configuration definition and loading pipeline. Defaults, selected base
+and imports, selected overlay, and explicit application CLI overrides participate
+in ConfigKit's lifecycle; resolution and validation follow its upstream rules.
+Normalization that changes inspected values is recorded through its resolver.
+Runtime conversion does not perform a hidden second merge or apply new defaults
+after the inspection snapshot has been created.
+
+Scriptorium retains its application field names and domain meanings unless a
+specific change below requires otherwise. ConfigKit's public root API replaces
+direct YAML decoding and parser-specific hooks. Application defaults remain
+Scriptorium-owned; omitted framework settings still delegate to Promptkit.
+Backend conversion retains deterministic registration and engine-scoped shared
+capacity state. This work does not change the HTTP request contract or move
+configuration loading into each HTTP request.
+
+## Native Configuration Semantics
+
+The target accepts ConfigKit's upstream YAML grammar, schema binding, limits,
+composition, and value-presence rules without a legacy preprocessing layer.
+Migration-visible changes include:
+
+- An existing empty file is invalid; `{}` expresses an empty application mapping.
+  Defaults-only operation with no discovered file uses `NoSource` instead.
+- Explicit empty strings, false values, and zeroes replace defaults. They are
+  not silently treated as omission. Application validation still decides whether
+  a resulting value is meaningful for its field or selected command.
+- Null follows the application's declared nullable types. It is not a universal
+  spelling for omission. Scalars and structs do not accept null merely because
+  the former parser tolerated it.
+- Anchors, aliases, merge keys, explicit tags, additional documents, unsupported
+  scalar spellings, and coercions rejected by ConfigKit are rejected here too.
+- ConfigKit's reserved `configkit` key and resource limits apply, including in
+  dynamic application data. These constraints do not apply to independently
+  loaded Promptkit definitions or HTTP request bodies.
+- Dynamic integers in backend `extra_params` use ConfigKit's `int64`
+  representation. Runtime mapping preserves their supported value semantics
+  rather than reserializing through YAML or demanding the old Go scalar type.
+
+Presence remains meaningful at the framework boundary: omitted/null
+`transient_retries` delegates to Promptkit, while explicit zero disables retries;
+omitted backend queue capacity differs from explicit zero. Application HTTP
+size defaults and zero-as-unlimited behavior remain. No new value ranges or
+backend restrictions are introduced merely to adopt the library.
+
+## Discovery, Imports, Overlays, And Paths
+
+Source discovery remains Scriptorium-owned. The target retains the documented
+explicit `--config` selection and ordered system-location search. An explicitly
+empty config path is invalid rather than an implicit request for a default file.
+If neither implicit candidate exists, loading uses defaults only; a selected
+file that fails loading is not silently replaced with defaults. No implicit
+working-directory configuration search, environment overlay, or directory scan
+is added.
+
+Application configuration supports ConfigKit's explicit path and glob imports.
+Imports compose disjoint definitions; repeated non-mapping definitions are
+conflicts, not last-writer-wins overrides. There is no implicit `conf.d` scan.
+
+An optional `--config-overlay` selects ConfigKit's one named overlay using its
+standard `overlays/<name>.yml` location, naming rules, and merge behavior. No
+overlay is selected by default. Selecting an active overlay without a base
+source follows ConfigKit's rejection of overlays with `NoSource`; Scriptorium
+does not invent a root. Application overlays do not replace `--profile`,
+`profile_dir`, or Promptkit's execution-profile selection.
+
+Use the selected file's canonical directory as ConfigKit's default import and
+overlay root. This root governs configuration source traversal only. Values of
+`prompt_dir`, `profile_dir`, `schema_dir`, and `server.artifact_root` retain
+working-directory-relative interpretation; absolute values remain absolute.
+This is an explicit application path policy, not an upstream requirement or a
+legacy parsing mode. ConfigKit's consumer examples demonstrate alternatives but
+do not mandate source-file-relative application paths. This adoption does not
+extend import containment to application data or alter HTTP artifact containment.
+
+## CLI Overrides And Configuration Inspection
+
+Only flags explicitly supplied by the user become ConfigKit application
+override records. Parser defaults do not overwrite file or overlay settings.
+Explicit empty values are not dropped. The existing invocation-specific prompt,
+profile, input, model, timeout, output, and reporting options remain invocation
+state rather than new application configuration fields. In particular, the
+run/render retry flag remains a request override; it does not mutate the engine's
+application retry setting or backend state.
+
+Expose application configuration validation, effective redacted YAML, source
+provenance, and explanation of an individual configuration path. These are
+included features, not merely library capabilities available for future work.
+They use ConfigKit diagnostics and inspection records from the shared loading
+pipeline, including defaults, imports, overlays, explicit CLI values, and
+recorded normalization. No separate inspection-only interpretation is maintained.
+
+Configuration inspection does not execute prompts, make provider calls, or
+resolve API-key values. Application validation is not a promise that every
+Promptkit backend or definition is valid or that a provider is reachable.
+Command-specific execution prerequisites remain with the relevant command;
+inspection can explain an incomplete application configuration without requiring
+an executable prompt selection. The interface is a CLI configuration command
+family, not a second CLI framework or an HTTP configuration endpoint; the
+implementation plan specifies its concrete commands and mappings.
+
+Safe diagnostics must not restore raw parser errors or configuration values
+that ConfigKit deliberately suppresses. Effective output and explanation use
+the upstream redacted renderers; they are inspection output, not guaranteed
+reloadable configuration. Under the maintainer's selected policy, backend
+`extra_params` are visible in effective output and explanation, including
+historical values. There is no blanket container masking, sensitive-key heuristic,
+or new selective-redaction configuration surface.
+
+## Credential Boundary And Deferred Features
+
+Existing `api_key_env` values remain environment-name strings passed to Promptkit.
+They do not become ConfigKit `Secret` fields. ConfigKit eagerly resolves effective
+typed references and requires nonempty values; that would change Promptkit's
+optional/required and selected-credential behavior. This departure from the
+secret-reference feature is an ownership decision, not compatibility plumbing.
+Raw API-key fields remain unsupported under the existing architecture policy.
+
+API-key values are the only secrets requiring ConfigKit redaction in this
+adoption. The application configuration holds environment-variable names, not
+those values, so inspection neither resolves nor contains the actual keys.
+Reference names remain visible. Do not tag `api_key_env` as secret merely to
+hide its name, add an eager key lookup, or mask unrelated application settings.
+The existing raw-key prohibition and Promptkit boundary provide the protection
+without introducing a redundant credential-bearing configuration field.
+
+Secrets-directory bootstrap and application schema versioning are deferred until
+there is a concrete application need. Their availability in ConfigKit does not
+make them mandatory. Hot reload, runtime resource replacement, remote imports,
+automatic migrations, and general environment-variable overlays are outside
+this adoption. These are scope limits for this feature, not new permanent
+prohibitions on future explicitly requested work.
+
+## Required Changes And Completion Criteria
+
+- Replace manual decoding/default/override machinery with the shared ConfigKit
+  integration and retain only application-specific discovery, rules, and mapping.
+- Support imports and selected overlays across commands that load application
+  configuration, and deliver the configuration inspection capabilities above.
+- Update application configuration types, presence handling, CLI explicit-set
+  mapping, and diagnostics to match the selected native semantics.
+- Update maintained examples and canonical configuration, CLI, operations,
+  architecture, and internal documentation when behavior lands. Upstream generic
+  contracts remain linked rather than copied into a competing specification.
+- Add proportionate integration coverage for the cutover: defaults-only input,
+  explicit values and nulls, imports/conflicts, overlays, CLI precedence,
+  inspection agreement and API-key non-disclosure, visible dynamic backend
+  parameters, unchanged credential ownership, and representative
+  execution/server startup.
+  ConfigKit owns exhaustive parser and composition tests; Scriptorium tests its
+  adoption and application boundaries.
+- Pin the published dependency and maintain ConfigKit and YAML attribution under
+  the [OSS policy](../policy/oss.md), using imported-commit source/notice links
+  and applicable standard and companion license texts.
+- Document migration in a release appropriate for the intentional breaking
+  configuration changes under the [release policy](../policy/release.md).
+  Release selection and publication remain separate preparation work; this
+  roadmap adds no release gate, network probe, or new distribution format.
+
+Completion means one ConfigKit-native application configuration path serves
+execution and inspection, the selected user-facing features work, and migrated
+examples and regression coverage demonstrate the documented semantics. It does
+not require adopting every ConfigKit feature or changing Promptkit definitions.
+
+## Resolved Inspection Policy
+
+On 2026-09-26 the maintainer selected no redaction for `extra_params` and limited
+the relevant secret to the API key. Inspection therefore preserves parameter
+values and environment-reference names, while actual API-key values remain
+outside the application configuration pipeline. Visibility is intentional;
+this is not a promise to detect secrets placed inside arbitrary parameter values.
+
+The delivered `config show` and `config explain` commands use ConfigKit's
+inspection writers without secret tags or a second redaction path. Both
+effective and historical parameter values remain visible; `api_key_env` names
+remain visible while Promptkit alone resolves selected credential values.

docs/roadmap/policy.md

@@ -0,0 +1,63 @@
+# OSS Policy Adoption
+
+## Status
+
+Completed on 2026-09-26. The upstream template work proposed here is now
+available and adopted locally. This document records the adoption; the
+[OSS policy](../policy/oss.md) is the canonical owner of the ongoing rules.
+
+The adopted baseline is the standard policy repository at commit
+`eea8bcd21a964b395e31971077d7fddb445f2c28`:
+
+- OSS policy base version 1.1;
+- documentation policy base version 1.4; and
+- release policy base version 1.2.
+
+Each affected policy records its base version and local update date. Template
+adoption instructions were removed. Unchanged agent, architecture, and testing
+policies retain their existing versions.
+
+## Adoption Outcome
+
+- [Documentation ownership](../policy/documentation.md) distinguishes the
+  project license, component entries, standard texts, and OSS policy.
+- The [contributor guide](../development.md) routes dependency, copied-asset,
+  attribution, and release-document work to the OSS policy.
+- [Component entries](../oss/attribution.md) cover the six selected direct and
+  transitive Go modules. Source and project-specific notice links identify the
+  imported commits. LGPL entries link directly to both LGPL v3 and GPL v3.
+- Project-specific notices are linked upstream under the adopted house
+  interpretation. YAML's former local LICENSE and NOTICE copies were removed;
+  their history remains in Git. Missing upstream copyright statements are
+  identified in the affected entries rather than fabricated.
+- The five applicable standard texts in `docs/oss/licenses/` match the upstream
+  reference collection byte-for-byte: GPL v3, LGPL v3, Apache 2.0, BSD 3-Clause,
+  and MIT. The reference collection documents their publisher/SPDX provenance.
+  Adoption removed a stray LGPL prefix, restored the complete Apache text,
+  separated the standard BSD text from the Go-specific notice, and added MIT.
+- The [README](../../README.md) ends with friendly project-license and
+  attribution links. The [release policy](../policy/release.md) delegates footer
+  conventions to the OSS policy; the [release procedure](../release.md) retains
+  Scriptorium's immutable-tag URLs and existing hosted-body reuse.
+
+## Preserved Scope
+
+This adoption changes documentation and attribution, not Scriptorium's license,
+dependencies, application architecture, or release machinery. Published notes,
+tags, and assets remain unchanged. Existing source-release completion and
+asynchronous binary-delivery behavior remain intact.
+
+There is no new publication-time network check, approval gate, scanner, source
+bundle, downloadable license asset, or modified-dependency rebuild exercise.
+Concrete exceptions continue to receive the treatment specified by the adopted
+OSS policy; the house format is not a universal legal-sufficiency guarantee.
+
+## Validation
+
+The standard texts were compared against the upstream reference collection.
+The BSD reference intentionally retains its publisher's trailing space in the
+copyright placeholder; it is not reformatted to satisfy a whitespace check.
+Dependency commit references were resolved from Go module origin metadata or
+the selected upstream tag, and affected local links and upstream attribution
+destinations were checked. No runtime code, commands, or executable examples
+changed.

docs/roadmap/remediation.md

@@ -0,0 +1,77 @@
+# Promptkit Upgrade Review Remediation
+
+## Scope And Status
+
+Review baseline: `e177d7c`, covering implementation commits after `5c3b0cf`.
+This document records a resolved test-contract issue; it does not authorize
+unrelated implementation changes. No runtime defect or missing planned feature
+was identified in this review. Release preparation remains separate work.
+
+## R1: Remove The Unrequested Context-Identity Invariant
+
+Priority: P3 (low). Status: Complete.
+
+Original location: `internal/adapter/http/handler_test.go:334–349`,
+`TestHandlerRunTimeoutValidationAndContextMapping`, introduced in `7de2b02`.
+
+The test's `child` field and `(ctx != parent) != tc.child` assertion required
+the exact parent context object for omitted/null/zero timeouts and a distinct
+object for a positive timeout. The requested contract concerns effective
+deadlines and cancellation propagation, not Go interface identity. In
+particular, a harmless derived context that preserves the parent's deadline,
+values, and cancellation would fail the omission cases despite satisfying the
+application contract.
+
+This elevated an incidental implementation choice into a test-enforced
+invariant, contrary to the [agent policy](../policy/agents.md) and
+[testing policy](../policy/testing.md). The current handler's context behavior
+is correct; this is a maintainability issue, not a runtime cancellation bug.
+
+### Required Change
+
+Remove the identity comparison and the identity-only table field. Retain
+behavioral assertions that an earlier parent deadline is not extended and
+parent cancellation reaches execution. For omitted/null/zero values, verify
+that no deadline is introduced when the parent has none. Retain coverage of
+positive deadlines and the distinction between a canceled execution context
+and a provider-attempt timeout while the execution context remains live.
+
+Use the existing runner seam and focused cases; do not introduce a context
+wrapper framework, timer abstraction, new public guarantee, or production
+refactor solely to change this test.
+
+### Completion Criteria
+
+- Tests protect effective deadline/cancellation behavior without requiring
+  context object identity.
+- A semantically equivalent derived context does not fail merely because it
+  is a different object.
+- Existing timeout validation and HTTP outcome coverage remains intact.
+
+### Resolution And Validation
+
+Removed the identity-only table field and context comparisons. The test now
+checks effective deadlines, including omitted/null/zero requests with no
+parent deadline, and explicitly checks that parent cancellation reaches the
+runner. Existing positive-deadline integration tests and live-context
+provider-timeout mapping remain intact. No production code changed.
+
+Focused HTTP timeout tests, `go test ./...`, `go test -race ./...`,
+`go vet ./...`, a native build outside the repository, and `git diff --check`
+all passed after remediation.
+
+## Original Review Verification
+
+- `go test -race -count=1 ./...` passed, including retry, shared-capacity,
+  deadline, report, definition-conformance, and architecture-boundary tests.
+- `go vet ./...` passed.
+- A native `cmd/scriptorium` build passed, with output outside the repository.
+- The selected dependency is Promptkit v0.11.0 with its intended v2 catalogs.
+- Application code, tests, release automation, and historical releases were
+  not changed during this review.
+
+The runtime ownership boundaries remain as intended: Promptkit owns framework
+formats, execution, retries, and backend admission; Scriptorium owns executable
+adapters, application settings, HTTP artifact policy, deadlines, and terminal
+report presentation. No additional runtime restrictions, durable workflow
+state, or extra release compliance gates were identified in the reviewed diff.

docs/roadmap/upgrade-kit.md

@@ -0,0 +1,273 @@
+# Promptkit v0.11 Adoption
+
+## Status And Intent
+
+Accepted feature direction, recorded on 2026-09-26. Implementation is complete
+to the pre-release boundary; version selection, release-candidate validation,
+and publication remain separate release preparation. The baseline descriptions
+below record the state when this roadmap was written, not current behavior.
+Upgrade directly from Promptkit v0.9.0 to v0.11.0, including the intervening
+v0.10.0 catalog transition. This document defines the desired end state and
+the changes from the current baseline. The [implementation plan](implementation.md)
+owns sequencing, concrete interface choices, verification steps, and unresolved
+implementation decisions.
+
+The target is complete definition compatibility, controllable transient
+retries, an optional whole-run deadline, and modest opt-in execution diagnostics.
+This is a focused adapter and configuration update, not another architecture
+migration. Optional diagnostics are included in the desired feature set.
+“Optional” means callers opt in, not that diagnostics may be silently omitted
+from the completed adoption or appear in default output.
+
+[ADR 0004](../adr/0004-definition-boundary.md) and the
+[architecture policy](../policy/architecture.md) govern scope. Promptkit remains
+the parser, validator, execution engine, and owner of framework defaults.
+Scriptorium remains a simple executable application for independent prompt
+executions, with shared operational state where required for backend limits.
+
+## Baseline And Upstream Changes
+
+Scriptorium currently selects Promptkit v0.9.0 and the v1.0.0 backend catalog
+modules. It already supports CLI execution, rendering, prompt/profile
+inspection, HTTP execution, configured backends, and presence-aware execution
+overrides. It does not expose transient retry settings, a whole-run deadline,
+or execution observers. CLI execution currently supplies a background context.
+
+The upstream review covered the local tagged release v0.11.0 at commit
+`db05b56166ca2d8650dfaf660e1cf0eed3932e1d`. Its canonical references are:
+
+- [v0.10.0 migration guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.10.0/docs/releases/v0.10.0.md);
+- [v0.11.0 release notes](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/releases/v0.11.0.md);
+- [definition formats](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/formats.md);
+- [execution and retry integration contract](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/integrations/openai-compatible-chat.md);
+- [consumer guide](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/docs/consumers/pkg-promptkit.md); and
+- [observation declarations](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/observations.go).
+
+A diagnostic run using a temporary module file compiled Scriptorium against
+v0.11.0. The existing test suite reported one failing profile-inspection case
+that selects the removed `deepseek-4-flash` profile; other packages passed.
+This establishes an initial migration issue, not proof of complete behavioral
+compatibility or coverage of the new capabilities. The diagnostic did not
+change the repository dependency pin.
+
+Neither intervening release adds declarative prompt/profile fields requiring
+new adapter plumbing. The definition-facing changes concern catalog contents
+and resolution precedence. Retry and observer APIs are execution controls;
+their adoption is a deliberate case-by-case choice under ADR 0004.
+
+## Dependency And Definition Compatibility
+
+### Target End State
+
+Scriptorium uses tagged Promptkit v0.11.0 through its supported root package,
+with the catalog versions selected by that release:
+
+- `promptkit-backend-openrouter/v2` v2.20260922.1;
+- `promptkit-backend-rakestrawhome/v2` v2.20260922.0.
+
+Valid definitions for the selected Promptkit version remain usable through
+Scriptorium's existing sources and adapters, subject to its established HTTP
+artifact and deployment boundaries. Rendering and inspection remain offline
+operations that do not generate model output.
+
+The maintained names `promptkit-openrouter-standard` and
+`promptkit-openrouter-auto` work wherever a profile ID is accepted, including
+prompt defaults, explicit selection, inheritance, and profile inspection.
+They are opt-in choices, not new Scriptorium defaults. Ordinary custom-source
+precedence remains intact.
+
+A stable maintained name is not a frozen model/settings selection. External
+catalog entries can change or disappear across upgrades; inheriting from one
+does not create a frozen copy. Complete application-owned profiles remain the
+appropriate choice when an application owns its model and settings.
+
+### Change From Baseline
+
+The module graph and active definition fixtures reflect the released v0.11
+catalogs rather than v0.9's catalogs. The selected catalog removes
+`deepseek-3-2`, `deepseek-4-flash`, and `deepseek-4-pro`; newer DeepSeek names
+are not compatibility aliases. Migration guidance offers an available catalog
+profile, a maintained Promptkit name, or a complete application-owned definition.
+Application-owned examples retain fixed settings where that is their purpose.
+
+Definition conformance and focused resolution coverage protect Scriptorium's
+adapter boundary without duplicating the upstream parser or catalog. Existing
+dependency boundaries also apply to the new `/v2` module paths.
+
+## Controllable Transient Retries
+
+### Target End State
+
+Expose `transient_retries` in application configuration, with a CLI execution
+override and an HTTP request override. Preserve value presence throughout:
+
+| Value | Meaning |
+| --- | --- |
+| Omitted | Inherit the engine setting, ultimately Promptkit's default. |
+| Zero | Disable additional transient attempts. |
+| Positive integer | Permit that many additional transient attempts. |
+| Negative integer | Invalid configuration or request. |
+
+Adopt Promptkit's default of two additional attempts rather than silently
+pinning the previous single-attempt behavior. Users must be able to select zero
+explicitly. Request overrides take precedence over application configuration;
+omission must not accidentally become an explicit zero.
+
+Promptkit alone owns eligible statuses, attempt execution, backoff,
+`Retry-After`, and capacity coordination. Scriptorium adds no outer model retry
+loop. Transient retries and structural repairs have distinct budgets: each
+generation, including each repair generation, receives its transient budget.
+Retries do not imply exactly-once generation or complete billing knowledge.
+
+### Change From Baseline
+
+Application settings reach `promptkit.Config.TransientRetries`, including the
+shared server engine, and request overrides reach
+`promptkit.RunRequest.TransientRetries`. This adds execution controls, not
+prompt/profile schema fields or provider model parameters. Strict decoding and
+presence-aware mapping preserve inheritance and explicit zero. User and
+operations guidance makes the changed default, latency/cost implications, and
+independent repair budget visible. Cancellation and shared backend limits
+remain effective during retrying executions.
+
+## Optional Whole-Run Deadline
+
+### Target End State
+
+Callers can opt into a total execution deadline independently of the existing
+generation timeout. No new default deadline is imposed. Existing timeout
+controls must not silently change meaning: in v0.11.0 they apply per generation
+attempt, whereas a whole-run deadline bounds the overall operation, including
+preparation, capacity waiting, retry delays, attempts, and repairs as supported
+by the upstream context contract.
+
+Expose this choice through CLI execution and HTTP execution. Derive any timeout
+context from the caller's existing context, so a caller cancellation or earlier
+deadline still wins. This is a cancellation boundary, not a guarantee of
+instantaneous interruption of arbitrary callback or filesystem code.
+
+### Change From Baseline
+
+CLI execution can carry a selected deadline instead of always using an
+unbounded background context. HTTP execution can shorten the incoming context
+without losing parent cancellation or changing server lifecycle behavior.
+Total-run duration handling is separate from existing attempt-timeout conversion.
+The interval covers the engine invocation, not adapter parsing, engine creation,
+or output writing. Provider retry timing can exceed local backoff limits;
+per-attempt timeouts do not cap that complete invocation.
+
+## Opt-In Execution Reports
+
+### Target End State
+
+Provide a modest terminal execution report for CLI and HTTP callers who opt
+in. Capture selected metadata through `RunRequest.Observer`, including attempt,
+retry, repair, timing, outcome, and presence-aware usage information. Reports
+should be available for completed successful and failed execution invocations
+when Promptkit supplies terminal observations. Do not invent execution data for
+failures that occur before invocation.
+
+Keep generated CLI content on stdout or the existing output path. Execution
+reports use a separate destination. Ordinary CLI output and HTTP response
+shapes remain unchanged when reporting is not requested.
+
+The report has a Scriptorium-owned schema rather than serializing upstream
+events directly: `ExecutionEvent` has no stable JSON contract. Preserve unknown
+usage versus reported zero, and incomplete versus complete observed totals.
+Existing successful-result usage semantics remain unchanged. A terminal total
+is an aggregate, not another attempt to add to the bill; no report is a billing
+ledger.
+
+### Change From Baseline
+
+Both adapters have opt-in reporting and explicit application-owned report DTOs.
+Per-request collection adds no shared mutable report state, durable storage,
+callback I/O, or event archive. Operational failures can carry reports without
+manufacturing partial `RunResult` values. Default output remains unchanged.
+
+Reports exclude prompt and response bodies, credentials, headers, endpoint URLs,
+provider diagnostic text, and raw errors. Selected identifiers remain potentially
+sensitive even when observations are content-free. Reports are concise terminal
+metadata, not a general telemetry framework, event stream, or durable history.
+
+## Errors And Compatibility
+
+Preserve existing CLI exit classifications and HTTP `502 / llm_failed`
+classification for generation failures. Keep safe public messages rather than
+exposing provider error bodies or parsing diagnostic strings.
+
+Promptkit adds `ErrTransport`, `ErrMalformedResponse`, `ErrRetriesExhausted`, and
+`RetryExhaustedError`. Use their public identities to support optional structured
+failure details in execution diagnostics, including actual attempts where
+available. Retry exhaustion and the final cause are distinct facts; do not lose
+the underlying failure category by flattening them into one label.
+
+Capacity and validation behavior remains intact. Whole-run deadline and caller
+cancellation outcomes are deliberate adapter classifications, distinct from a
+provider-attempt timeout while the caller context is still live. The implementation
+plan owns the exact presentation choices. Existing generation-error codes do
+not need a breaking replacement for this upgrade.
+
+## Licensing And Distribution Review
+
+Promptkit v0.11.0 changes from GPLv3 to LGPL-3.0-or-later. Scriptorium's current
+binary workflow produces statically linked Go executables, and upstream's
+[license summary](https://gitea.maximumdirect.net/eric/promptkit/src/tag/v0.11.0/README.md#license)
+explicitly calls out distribution considerations for that model. Catalog and
+other dependency licenses remain separate.
+
+Before distributing the upgraded binaries, resolve the applicable notices,
+license texts, source availability, and any required materials or permissions
+for recombination with a modified library. Record the chosen distribution
+approach and make any corresponding packaging/documentation changes. This
+roadmap is not a legal determination or a complete compliance checklist.
+
+Do not automatically relicense Scriptorium, assume that library source alone
+settles every distribution requirement, or alter historical release contents.
+Licensing review is a concrete adoption prerequisite, not a new blanket manual
+approval gate for every future release.
+
+## Documentation And Delivery
+
+Current documentation links to the selected v0.11.0 contracts once implemented.
+CLI, HTTP, configuration, and operations documents own their respective new
+behavior; internal docs explain adapter mapping; release notes summarize
+compatibility impact. Historical release notes and ADR decision content remain
+unchanged. Current examples and migration guidance reflect the selected version.
+
+The upgrade merits a minor Scriptorium release, not a compatibility patch,
+with prominent migration guidance for removed profiles, changed retry behavior,
+the explicit single-attempt option, and timeout layering. Any later decision to
+defer part of the target leaves that feature visibly incomplete until delivered.
+
+Use the existing [release policy](../policy/release.md) and
+[runbook](../release.md). This roadmap authorizes no publication. No new ADR is
+needed unless a later decision changes the established product or compatibility
+boundaries.
+
+## Completion Criteria
+
+- The tagged dependency and selected catalog modules are adopted without
+  unsupported imports or local dependency substitutions.
+- Existing definition conformance tests and focused maintained-profile tests
+  pass; removed profile IDs have explicit migration guidance, not silent aliases.
+- Retry settings preserve omission and zero across configuration, CLI, and HTTP;
+  Promptkit remains the sole retry/capacity implementation.
+- Optional total deadlines bound the caller operation without redefining
+  per-attempt timeouts or disabling parent cancellation.
+- Opt-in terminal reports provide safe, isolated, presence-aware execution
+  metadata without changing default output or creating durable workflow state.
+  Any deferred reporting work remains visibly incomplete.
+- Existing error classifications remain compatible, with deliberate handling
+  of the new public identities and context outcomes.
+- The licensing/distribution approach is resolved before upgraded binary
+  distribution, and any required materials are included in the chosen workflow.
+- Affected current contracts, examples, and migration guidance match
+  implemented behavior. Relevant tests remain offline and credential-free,
+  and normal development validation passes. Release-candidate validation and
+  the version-matched note remain release preparation.
+
+Concrete interface decisions and the selected distribution approach are recorded
+in the completed [implementation plan](implementation.md). They do not expand
+the product into conversations, follow-up messages, application-owned semantic
+retry loops, prepared-handle coordination, or multi-step workflows.

examples/config-fragments/backend.yml

@@ -0,0 +1,4 @@
+backends:
+  local-gpu:
+    endpoint: http://localhost:11434/v1
+    api_key_env: LOCAL_GPU_API_KEY

examples/config-fragments/sources.yml

@@ -0,0 +1,3 @@
+prompt_dir: ./examples/prompts
+profile_dir: ./examples/profiles
+schema_dir: ./examples/schemas

examples/config.composed.yml

@@ -0,0 +1,8 @@
+# Select from the repository root; application paths remain working-directory-relative.
+configkit:
+  imports:
+    - path: config-fragments/sources.yml
+    - path: config-fragments/backend.yml
+
+server:
+  addr: :8080

examples/overlays/local.yml

@@ -0,0 +1,4 @@
+defaults:
+  render_format: json
+server:
+  addr: 127.0.0.1:8080

go.mod

@@ -3,13 +3,15 @@ module gitea.maximumdirect.net/eric/scriptorium
 go 1.25.5
 
 require (
-	gitea.maximumdirect.net/eric/promptkit v0.9.0
-	gopkg.in/yaml.v3 v3.0.1
+	gitea.maximumdirect.net/eric/ConfigKit v0.4.0
+	gitea.maximumdirect.net/eric/promptkit v0.11.0
 )
 
 require (
-	gitea.maximumdirect.net/eric/promptkit-backend-openrouter v1.0.0 // indirect
-	gitea.maximumdirect.net/eric/promptkit-backend-rakestrawhome v1.0.0 // indirect
+	gitea.maximumdirect.net/eric/promptkit-backend-openrouter/v2 v2.20260922.1 // indirect
+	gitea.maximumdirect.net/eric/promptkit-backend-rakestrawhome/v2 v2.20260922.0 // indirect
 	github.com/santhosh-tekuri/jsonschema/v6 v6.0.2 // indirect
+	go.yaml.in/yaml/v4 v4.0.0-rc.6 // indirect
 	golang.org/x/text v0.14.0 // indirect
+	gopkg.in/yaml.v3 v3.0.1 // indirect
 )

go.sum

@@ -1,13 +1,17 @@
-gitea.maximumdirect.net/eric/promptkit v0.9.0 h1:IpvDRC8L6xRxQ9hpuyKOmMc5b6MeLTKYyx+h1YAjy08=
-gitea.maximumdirect.net/eric/promptkit v0.9.0/go.mod h1:oMJ/WUJImUtwJ5e+6MAGECPYAErAkOaKel0G+3T/b4E=
-gitea.maximumdirect.net/eric/promptkit-backend-openrouter v1.0.0 h1:lc062euk2qseO//D762i3JaFyulDNML3eQQX7DkYTho=
-gitea.maximumdirect.net/eric/promptkit-backend-openrouter v1.0.0/go.mod h1:AIa7kAu2mfrRQgcspe4L+DW51WqgnALQT60lqkEywJI=
-gitea.maximumdirect.net/eric/promptkit-backend-rakestrawhome v1.0.0 h1:j9YY7wsTVjzke2kHH4YAzpU0oUpM+x+nXwl1IeS+2eg=
-gitea.maximumdirect.net/eric/promptkit-backend-rakestrawhome v1.0.0/go.mod h1:4RNS+LILDg4JbS4Ts9Lwy1C92wauXJIbeQaalps4Koo=
+gitea.maximumdirect.net/eric/ConfigKit v0.4.0 h1:fo+vKff3zTGeWqJIA38NtT84b6OUP//9ZmBH3lFuKgk=
+gitea.maximumdirect.net/eric/ConfigKit v0.4.0/go.mod h1:tYWpsrbLyu05r8+1MapWp5fN6ZrX9I8hU2xEPedYmYk=
+gitea.maximumdirect.net/eric/promptkit v0.11.0 h1:x7JguOUiInPVXK0GEitjyoRFqqvcuSTjYUvqJ+whtik=
+gitea.maximumdirect.net/eric/promptkit v0.11.0/go.mod h1:0zEBVlJJC+ldDdXazAgybYBXNSe74vwFx5dyOuSos/w=
+gitea.maximumdirect.net/eric/promptkit-backend-openrouter/v2 v2.20260922.1 h1:S2/Q/JDE+m8Ca3pDhAjVbfYrdlEhs9blj0STgNopHo0=
+gitea.maximumdirect.net/eric/promptkit-backend-openrouter/v2 v2.20260922.1/go.mod h1:a3VBbh9oaGXD6ruxhtwHj+AWW6DY/SSyby1QtMmYSUo=
+gitea.maximumdirect.net/eric/promptkit-backend-rakestrawhome/v2 v2.20260922.0 h1:Y0EXc3eh12eKW2NrOviVqdk3oBo6UUjopt1YAa1B3Oc=
+gitea.maximumdirect.net/eric/promptkit-backend-rakestrawhome/v2 v2.20260922.0/go.mod h1:mtgD1E0SGm0e/I8Ffhz+AYX7GN7RKUWJXehNFIKHs4Y=
 github.com/dlclark/regexp2 v1.11.0 h1:G/nrcoOa7ZXlpoa/91N3X7mM3r8eIlMBBJZvsz/mxKI=
 github.com/dlclark/regexp2 v1.11.0/go.mod h1:DHkYz0B9wPfa6wondMfaivmHpzrQ3v9q8cnmRbL6yW8=
 github.com/santhosh-tekuri/jsonschema/v6 v6.0.2 h1:KRzFb2m7YtdldCEkzs6KqmJw4nqEVZGK7IN2kJkjTuQ=
 github.com/santhosh-tekuri/jsonschema/v6 v6.0.2/go.mod h1:JXeL+ps8p7/KNMjDQk3TCwPpBy0wYklyWTfbkIzdIFU=
+go.yaml.in/yaml/v4 v4.0.0-rc.6 h1:1h7H1ohdUh93/FyE4YaDa1Zh64K6VVbjF4K6WUxMtH4=
+go.yaml.in/yaml/v4 v4.0.0-rc.6/go.mod h1:aZqd9kCMsGL7AuUv/m/PvWLdg5sjJsZ4oHDEnfPPfY0=
 golang.org/x/text v0.14.0 h1:ScX5w1eTa3QqT8oi6+ziP7dTV1S2+ALU0bI+0zXKWiQ=
 golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=

internal/adapter/cli/application_config.go

@@ -0,0 +1,71 @@
+package cli
+
+import (
+	"errors"
+	"flag"
+	"fmt"
+	"io"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+	appconfig "gitea.maximumdirect.net/eric/scriptorium/internal/config"
+)
+
+// applicationOverrides keeps explicit application flags separate from invocation state.
+func applicationOverrides(fs *flag.FlagSet) []configkit.Override {
+	var overrides []configkit.Override
+	fs.Visit(func(selected *flag.Flag) {
+		var path configkit.Path
+		switch selected.Name {
+		case "prompt-dir":
+			path = configkit.Keys("prompt_dir")
+		case "profile-dir":
+			path = configkit.Keys("profile_dir")
+		case "schema-dir":
+			path = configkit.Keys("schema_dir")
+		case "addr":
+			path = configkit.Keys("server", "addr")
+		case "artifact-root":
+			path = configkit.Keys("server", "artifact_root")
+		case "max-request-bytes":
+			path = configkit.Keys("server", "max_request_bytes")
+		case "max-artifact-bytes":
+			path = configkit.Keys("server", "max_artifact_bytes")
+		case "max-response-bytes":
+			path = configkit.Keys("server", "max_response_bytes")
+		default:
+			return
+		}
+		overrides = append(overrides, configkit.Override{
+			Path: path, Value: selected.Value.(flag.Getter).Get(), Flag: "--" + selected.Name,
+		})
+	})
+	return overrides
+}
+
+func loadApplicationSettings(configPath, overlay string, explicit bool, overrides []configkit.Override) (appconfig.AppSettings, []configkit.Diagnostic, error) {
+	result, err := appconfig.Load(appconfig.LoadOptions{
+		Path: configPath, Overlay: overlay, Explicit: explicit, Overrides: overrides,
+	})
+	if err != nil {
+		return appconfig.AppSettings{}, nil, err
+	}
+	return appconfig.Settings(result.Config), result.Diagnostics(), nil
+}
+
+func writeApplicationError(output io.Writer, label string, err error) {
+	var loadError *configkit.LoadError
+	if errors.As(err, &loadError) {
+		if writeErr := configkit.WriteDiagnostics(output, loadError.Diagnostics()); writeErr != nil {
+			_, _ = fmt.Fprintf(output, "%s diagnostic write error: %v\n", label, writeErr)
+		}
+		return
+	}
+	_, _ = fmt.Fprintf(output, "%s error: %v\n", label, err)
+}
+
+func writeApplicationWarnings(output io.Writer, diagnostics []configkit.Diagnostic) error {
+	if len(diagnostics) == 0 {
+		return nil
+	}
+	return configkit.WriteDiagnostics(output, diagnostics)
+}

internal/adapter/cli/config.go

@@ -0,0 +1,104 @@
+package cli
+
+import (
+	"encoding/json"
+	"flag"
+	"fmt"
+	"io"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+	appconfig "gitea.maximumdirect.net/eric/scriptorium/internal/config"
+)
+
+type configurationSource struct {
+	ID    string          `json:"id"`
+	Label string          `json:"label"`
+	Layer configkit.Layer `json:"layer"`
+	Path  string          `json:"path,omitempty"`
+}
+
+func configCommand(args []string, stdout, stderr io.Writer) int {
+	if len(args) == 0 {
+		fmt.Fprintln(stderr, "config parse error: configuration mode is required")
+		return ExitRuntimeError
+	}
+	mode := args[0]
+	if mode != "validate" && mode != "show" && mode != "sources" && mode != "explain" {
+		fmt.Fprintln(stderr, "config parse error: unknown configuration mode")
+		return ExitRuntimeError
+	}
+
+	fs := flag.NewFlagSet("config "+mode, flag.ContinueOnError)
+	fs.SetOutput(io.Discard)
+	var path, overlay, pointer string
+	registerConfigSelectionFlags(fs, &path, &overlay)
+	registerConfigurationOverrides(fs)
+	if mode == "explain" {
+		fs.StringVar(&pointer, "path", "", "configuration path as a JSON Pointer")
+	}
+	if err := fs.Parse(args[1:]); err != nil {
+		writeApplicationError(stderr, "config parse", err)
+		return ExitRuntimeError
+	}
+	if fs.NArg() != 0 {
+		fmt.Fprintln(stderr, "config parse error: positional arguments are not accepted")
+		return ExitRuntimeError
+	}
+	var explanationPath configkit.Path
+	if mode == "explain" {
+		if !flagWasSet(fs, "path") {
+			fmt.Fprintln(stderr, "config parse error: --path is required")
+			return ExitRuntimeError
+		}
+		var err error
+		explanationPath, err = configkit.ParsePath(pointer)
+		if err != nil {
+			fmt.Fprintln(stderr, "config parse error: --path must be a JSON Pointer")
+			return ExitRuntimeError
+		}
+	}
+
+	result, err := appconfig.Load(appconfig.LoadOptions{
+		Path: path, Explicit: flagWasSet(fs, "config"), Overlay: overlay,
+		Overrides: applicationOverrides(fs),
+	})
+	if err != nil {
+		writeApplicationError(stderr, "config", err)
+		return ExitRuntimeError
+	}
+	if err := writeApplicationWarnings(stderr, result.Diagnostics()); err != nil {
+		return ExitRuntimeError
+	}
+	switch mode {
+	case "validate":
+		return ExitOK
+	case "show":
+		err = result.WriteEffective(stdout)
+	case "sources":
+		sources := make([]configurationSource, 0, len(result.Sources()))
+		for _, source := range result.Sources() {
+			sources = append(sources, configurationSource{
+				ID: source.ID, Label: source.Label, Layer: source.Layer, Path: source.Path,
+			})
+		}
+		err = json.NewEncoder(stdout).Encode(sources)
+	case "explain":
+		err = result.WriteExplanation(stdout, explanationPath)
+	}
+	if err != nil {
+		fmt.Fprintf(stderr, "config output error: %v\n", err)
+		return ExitRuntimeError
+	}
+	return ExitOK
+}
+
+func registerConfigurationOverrides(fs *flag.FlagSet) {
+	fs.String("prompt-dir", "", "prompt-definition directory override")
+	fs.String("profile-dir", "", "profile directory override")
+	fs.String("schema-dir", "", "schema directory override")
+	fs.String("addr", "", "HTTP listen address override")
+	fs.String("artifact-root", "", "HTTP artifact root override")
+	fs.Int64("max-request-bytes", 0, "HTTP request byte limit override")
+	fs.Int64("max-artifact-bytes", 0, "HTTP artifact byte limit override")
+	fs.Int64("max-response-bytes", 0, "HTTP response byte limit override")
+}

internal/adapter/cli/config_test.go

@@ -0,0 +1,207 @@
+package cli
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"io"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	appconfig "gitea.maximumdirect.net/eric/scriptorium/internal/config"
+)
+
+func TestConfigCommandsInspectApplicationWithoutEngine(t *testing.T) {
+	path := writeAppConfigFile(t, "{}\n")
+	for _, mode := range []string{"validate", "show", "sources"} {
+		code, stdout, stderr := runCLICommand(t, configCommand, []string{mode, "--config", path})
+		if code != ExitOK || stderr != "" {
+			t.Fatalf("%s failed: code=%d stdout=%q stderr=%q", mode, code, stdout, stderr)
+		}
+		switch mode {
+		case "validate":
+			if stdout != "" {
+				t.Fatalf("validation wrote stdout: %q", stdout)
+			}
+		case "show":
+			if !strings.Contains(stdout, "render_format: text") || !strings.Contains(stdout, "max_request_bytes: 16777216") {
+				t.Fatalf("effective defaults missing: %q", stdout)
+			}
+		case "sources":
+			var sources []configurationSource
+			if err := json.Unmarshal([]byte(stdout), &sources); err != nil || len(sources) != 1 || sources[0].Layer != "base" {
+				t.Fatalf("unexpected source records: %q (%v)", stdout, err)
+			}
+		}
+	}
+}
+
+func TestConfigCommandsExposeComposedSourcesAndResolvedValues(t *testing.T) {
+	dir := t.TempDir()
+	if err := os.MkdirAll(filepath.Join(dir, "overlays"), 0o755); err != nil {
+		t.Fatal(err)
+	}
+	base := filepath.Join(dir, "config.yml")
+	for path, body := range map[string]string{
+		base:                               "configkit:\n  imports:\n    - path: fragment.yml\nprompt_dir: ' ./before/../prompts '\nprofile_dir: ./profiles\n",
+		filepath.Join(dir, "fragment.yml"): "server:\n  addr: :8081\n",
+		filepath.Join(dir, "overlays", "local.yml"): "server:\n  addr: :9090\n",
+	} {
+		if err := os.WriteFile(path, []byte(body), 0o600); err != nil {
+			t.Fatal(err)
+		}
+	}
+	args := []string{"--config", base, "--config-overlay", "local", "--profile-dir=", "--addr", ":7070"}
+	code, stdout, stderr := runCLICommand(t, configCommand, append([]string{"show"}, args...))
+	if code != ExitOK || stderr != "" || !strings.Contains(stdout, "prompt_dir: prompts") || !strings.Contains(stdout, "profile_dir: ''") || !strings.Contains(stdout, "addr: :7070") {
+		t.Fatalf("resolved effective output changed: code=%d out=%q err=%q", code, stdout, stderr)
+	}
+	code, stdout, stderr = runCLICommand(t, configCommand, append([]string{"sources"}, args...))
+	if code != ExitOK || stderr != "" || !strings.HasSuffix(stdout, "\n") {
+		t.Fatalf("source command failed: code=%d out=%q err=%q", code, stdout, stderr)
+	}
+	var sources []configurationSource
+	if err := json.Unmarshal([]byte(stdout), &sources); err != nil || len(sources) != 5 {
+		t.Fatalf("unexpected composed source records: %+v (%v)", sources, err)
+	}
+	for i, layer := range []string{"base", "base", "overlay", "cli", "cli"} {
+		if string(sources[i].Layer) != layer {
+			t.Fatalf("source %d layer = %q, want %q", i, sources[i].Layer, layer)
+		}
+	}
+}
+
+func TestConfigInspectionKeepsReferencesAndOrdinaryParameters(t *testing.T) {
+	const name = "SCRIPTORIUM_CONFIG_TEST_API_KEY"
+	const secret = "synthetic-secret-value-should-not-appear"
+	path := writeAppConfigFile(t, `backends:
+  synthetic:
+    endpoint: http://127.0.0.1:1/v1
+    api_key_env: SCRIPTORIUM_CONFIG_TEST_API_KEY
+    extra_params:
+      visible_parameter: synthetic-ordinary-value
+`)
+	t.Setenv(name, secret)
+	for _, present := range []bool{true, false} {
+		if !present {
+			if err := os.Unsetenv(name); err != nil {
+				t.Fatal(err)
+			}
+		}
+		code, stdout, stderr := runCLICommand(t, configCommand, []string{"show", "--config", path})
+		if code != ExitOK || stderr != "" || !strings.Contains(stdout, name) || !strings.Contains(stdout, "synthetic-ordinary-value") || strings.Contains(stdout, secret) {
+			t.Fatalf("credential boundary changed with present=%t: code=%d out=%q err=%q", present, code, stdout, stderr)
+		}
+	}
+	result, err := appconfig.Load(appconfig.LoadOptions{Path: path, Explicit: true})
+	if err != nil {
+		t.Fatal(err)
+	}
+	settings := appconfig.Settings(result.Config)
+	if len(settings.Backends) != 1 || settings.Backends[0].ExtraParams["visible_parameter"] != "synthetic-ordinary-value" {
+		t.Fatalf("runtime backend parameters changed: %+v", settings.Backends)
+	}
+}
+
+type errorOutputWriter struct{}
+
+func (errorOutputWriter) Write([]byte) (int, error) { return 0, errors.New("output unavailable") }
+
+func TestConfigCommandFailuresDoNotProduceSuccessfulOutput(t *testing.T) {
+	path := writeAppConfigFile(t, "server:\n  max_request_bytes: -1\n")
+	for _, mode := range []string{"validate", "show", "sources"} {
+		code, stdout, stderr := runCLICommand(t, configCommand, []string{mode, "--config", path})
+		if code != ExitRuntimeError || stdout != "" || !strings.Contains(stderr, "validate.error") {
+			t.Fatalf("%s misclassified invalid config: code=%d out=%q err=%q", mode, code, stdout, stderr)
+		}
+	}
+	path = writeAppConfigFile(t, "{}\n")
+	var stderr bytes.Buffer
+	if code := configCommand([]string{"show", "--config", path}, errorOutputWriter{}, &stderr); code != ExitRuntimeError || !strings.Contains(stderr.String(), "output error") {
+		t.Fatalf("writer error not reported: code=%d stderr=%q", code, stderr.String())
+	}
+	if code := configCommand([]string{"sources", "--config", path}, errorOutputWriter{}, io.Discard); code != ExitRuntimeError {
+		t.Fatalf("source writer error returned %d", code)
+	}
+}
+
+func TestConfigExplainUsesResolvedHistoryAndKeepsParametersVisible(t *testing.T) {
+	dir := t.TempDir()
+	if err := os.MkdirAll(filepath.Join(dir, "overlays"), 0o755); err != nil {
+		t.Fatal(err)
+	}
+	base := filepath.Join(dir, "config.yml")
+	for path, body := range map[string]string{
+		base: `prompt_dir: " ./prompts/../prompts "
+server:
+  addr: :8080
+backends:
+  "team/blue~1":
+    endpoint: http://127.0.0.1:1/v1
+    api_key_env: SCRIPTORIUM_EXPLAIN_KEY
+    extra_params:
+      "0": base-parameter
+`,
+		filepath.Join(dir, "overlays", "local.yml"): `server:
+  addr: :9090
+backends:
+  "team/blue~1":
+    extra_params:
+      "0": overlay-parameter
+`,
+	} {
+		if err := os.WriteFile(path, []byte(body), 0o600); err != nil {
+			t.Fatal(err)
+		}
+	}
+	const keyValue = "synthetic-key-bytes-must-not-appear"
+	t.Setenv("SCRIPTORIUM_EXPLAIN_KEY", keyValue)
+	args := []string{"--config", base, "--config-overlay", "local", "--addr", ":9090"}
+	for _, tc := range []struct {
+		pointer string
+		want    []string
+	}{
+		{pointer: "/server/addr", want: []string{":8080", ":9090", "--addr"}},
+		{pointer: "/prompt_dir", want: []string{"normalized application path", "prompts"}},
+		{pointer: "/backends/team~1blue~01/extra_params/0", want: []string{"base-parameter", "overlay-parameter"}},
+		{pointer: "/backends/team~1blue~01/api_key_env", want: []string{"SCRIPTORIUM_EXPLAIN_KEY"}},
+		{pointer: "", want: []string{"base-parameter", "overlay-parameter", "SCRIPTORIUM_EXPLAIN_KEY"}},
+	} {
+		code, stdout, stderr := runCLICommand(t, configCommand, append([]string{"explain", "--path", tc.pointer}, args...))
+		if code != ExitOK || stderr != "" || strings.Contains(stdout, keyValue) {
+			t.Fatalf("explain %q failed or disclosed key: code=%d out=%q err=%q", tc.pointer, code, stdout, stderr)
+		}
+		for _, want := range tc.want {
+			if !strings.Contains(stdout, want) {
+				t.Fatalf("explain %q lacks %q: %q", tc.pointer, want, stdout)
+			}
+		}
+	}
+	if err := os.Unsetenv("SCRIPTORIUM_EXPLAIN_KEY"); err != nil {
+		t.Fatal(err)
+	}
+	code, stdout, stderr := runCLICommand(t, configCommand, append([]string{"explain", "--path", "/backends/team~1blue~01/api_key_env"}, args...))
+	if code != ExitOK || stderr != "" || !strings.Contains(stdout, "SCRIPTORIUM_EXPLAIN_KEY") || strings.Contains(stdout, keyValue) {
+		t.Fatalf("absent credential changed reference explanation: code=%d out=%q err=%q", code, stdout, stderr)
+	}
+}
+
+func TestConfigExplainRejectsUnknownOrMalformedPaths(t *testing.T) {
+	path := writeAppConfigFile(t, "{}\n")
+	for _, args := range [][]string{
+		{"explain", "--config", path},
+		{"explain", "--config", path, "--path", "not-a-pointer"},
+		{"explain", "--config", path, "--path", "/missing"},
+	} {
+		code, stdout, stderr := runCLICommand(t, configCommand, args)
+		if code != ExitRuntimeError || stdout != "" || stderr == "" {
+			t.Fatalf("invalid explanation path accepted: args=%q code=%d out=%q err=%q", args, code, stdout, stderr)
+		}
+	}
+	var stderr bytes.Buffer
+	if code := configCommand([]string{"explain", "--config", path, "--path", ""}, errorOutputWriter{}, &stderr); code != ExitRuntimeError || !strings.Contains(stderr.String(), "output error") {
+		t.Fatalf("explanation writer failure not reported: code=%d err=%q", code, stderr.String())
+	}
+}

internal/adapter/cli/inspect.go

@@ -10,21 +10,27 @@ import (
 	"path/filepath"
 	"strings"
 
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
 	"gitea.maximumdirect.net/eric/promptkit"
-	appconfig "gitea.maximumdirect.net/eric/scriptorium/internal/config"
 	appformat "gitea.maximumdirect.net/eric/scriptorium/internal/format"
 )
 
 type promptInspectionConfig struct {
 	configPath, promptDir, promptID, promptVersion, outputPath string
+	configOverlay                                              string
 	configExplicit                                             bool
 	outputFormat                                               appformat.OutputFormat
+	overrides                                                  []configkit.Override
+	diagnostics                                                []configkit.Diagnostic
 }
 
 type profileInspectionConfig struct {
 	configPath, profileDir, profileID, outputPath string
+	configOverlay                                 string
 	configExplicit                                bool
 	outputFormat                                  appformat.OutputFormat
+	overrides                                     []configkit.Override
+	diagnostics                                   []configkit.Diagnostic
 }
 
 // emptyPromptDefinitionFS satisfies Promptkit's engine-level prompt-source
@@ -50,12 +56,15 @@ func inspectCommand(args []string, stdout, stderr io.Writer) int {
 	}
 	cfg, err := parsePromptInspectionArgs(args[1:])
 	if err != nil {
-		fmt.Fprintf(stderr, "inspect parse error: %v\n", err)
+		writeApplicationError(stderr, "inspect parse", err)
 		return ExitRuntimeError
 	}
 	settings, err := resolveAppSettingsForPromptInspection(cfg)
 	if err != nil {
-		fmt.Fprintf(stderr, "inspect error: %v\n", err)
+		writeApplicationError(stderr, "inspect", err)
+		return ExitRuntimeError
+	}
+	if err := writeApplicationWarnings(stderr, cfg.diagnostics); err != nil {
 		return ExitRuntimeError
 	}
 	engine, err := newEngine(settings)
@@ -83,12 +92,15 @@ func inspectCommand(args []string, stdout, stderr io.Writer) int {
 func inspectProfileCommand(args []string, stdout, stderr io.Writer) int {
 	cfg, err := parseProfileInspectionArgs(args)
 	if err != nil {
-		fmt.Fprintf(stderr, "inspect parse error: %v\n", err)
+		writeApplicationError(stderr, "inspect parse", err)
 		return ExitRuntimeError
 	}
 	settings, err := resolveAppSettingsForProfileInspection(cfg)
 	if err != nil {
-		fmt.Fprintf(stderr, "inspect error: %v\n", err)
+		writeApplicationError(stderr, "inspect", err)
+		return ExitRuntimeError
+	}
+	if err := writeApplicationWarnings(stderr, cfg.diagnostics); err != nil {
 		return ExitRuntimeError
 	}
 	engine, err := newEngine(settings, promptkit.WithPromptFS(emptyPromptDefinitionFS{}, "."))
@@ -117,7 +129,7 @@ func parseProfileInspectionArgs(args []string) (*profileInspectionConfig, error)
 	cfg := &profileInspectionConfig{outputFormat: appformat.DefaultOutputFormat}
 	fs := flag.NewFlagSet("inspect profile", flag.ContinueOnError)
 	fs.SetOutput(io.Discard)
-	registerConfigPathFlag(fs, &cfg.configPath)
+	registerConfigSelectionFlags(fs, &cfg.configPath, &cfg.configOverlay)
 	fs.StringVar(&cfg.profileDir, "profile-dir", "", "directory containing execution profiles")
 	fs.StringVar(&cfg.profileID, "profile", "", "profile ID to inspect")
 	rawFormat := ""
@@ -127,6 +139,7 @@ func parseProfileInspectionArgs(args []string) (*profileInspectionConfig, error)
 		return nil, err
 	}
 	cfg.configExplicit = flagWasSet(fs, "config")
+	cfg.overrides = applicationOverrides(fs)
 	if fs.NArg() > 0 {
 		return nil, fmt.Errorf("unexpected positional args: %v", fs.Args())
 	}
@@ -145,20 +158,19 @@ func parseProfileInspectionArgs(args []string) (*profileInspectionConfig, error)
 }
 
 func resolveAppSettingsForProfileInspection(cfg *profileInspectionConfig) (engineSettings, error) {
-	settings, err := resolveAppSettingsWithConfigPresence(cfg.configPath, cfg.configExplicit, appconfig.CLIOverrides{
-		ProfileDir: cfg.profileDir,
-	})
+	settings, diagnostics, err := loadApplicationSettings(cfg.configPath, cfg.configOverlay, cfg.configExplicit, cfg.overrides)
 	if err != nil {
 		return engineSettings{}, err
 	}
-	return engineSettings{profileDir: settings.ProfileDir, backends: settings.Backends}, nil
+	cfg.diagnostics = diagnostics
+	return engineSettings{profileDir: settings.ProfileDir, backends: settings.Backends, transientRetries: settings.TransientRetries}, nil
 }
 
 func parsePromptInspectionArgs(args []string) (*promptInspectionConfig, error) {
 	cfg := &promptInspectionConfig{outputFormat: appformat.DefaultOutputFormat}
 	fs := flag.NewFlagSet("inspect prompt", flag.ContinueOnError)
 	fs.SetOutput(io.Discard)
-	registerConfigPathFlag(fs, &cfg.configPath)
+	registerConfigSelectionFlags(fs, &cfg.configPath, &cfg.configOverlay)
 	fs.StringVar(&cfg.promptDir, "prompt-dir", "", "directory containing prompt definitions")
 	fs.StringVar(&cfg.promptID, "prompt", "", "prompt ID to inspect")
 	fs.StringVar(&cfg.promptVersion, "prompt-version", "", "optional prompt version")
@@ -169,6 +181,7 @@ func parsePromptInspectionArgs(args []string) (*promptInspectionConfig, error) {
 		return nil, err
 	}
 	cfg.configExplicit = flagWasSet(fs, "config")
+	cfg.overrides = applicationOverrides(fs)
 	if fs.NArg() > 0 {
 		return nil, fmt.Errorf("unexpected positional args: %v", fs.Args())
 	}
@@ -187,14 +200,13 @@ func parsePromptInspectionArgs(args []string) (*promptInspectionConfig, error) {
 }
 
 func resolveAppSettingsForPromptInspection(cfg *promptInspectionConfig) (engineSettings, error) {
-	settings, err := resolveAppSettingsWithConfigPresence(cfg.configPath, cfg.configExplicit, appconfig.CLIOverrides{
-		PromptDir: cfg.promptDir,
-	})
+	settings, diagnostics, err := loadApplicationSettings(cfg.configPath, cfg.configOverlay, cfg.configExplicit, cfg.overrides)
 	if err != nil {
 		return engineSettings{}, err
 	}
+	cfg.diagnostics = diagnostics
 	if strings.TrimSpace(settings.PromptDir) == "" {
 		return engineSettings{}, errors.New(errPromptDirRequired)
 	}
-	return engineSettings{promptDir: settings.PromptDir, backends: settings.Backends}, nil
+	return engineSettings{promptDir: settings.PromptDir, backends: settings.Backends, transientRetries: settings.TransientRetries}, nil
 }

internal/adapter/cli/run.go

@@ -2,6 +2,7 @@ package cli
 
 import (
 	"context"
+	"encoding/json"
 	"errors"
 	"flag"
 	"fmt"
@@ -12,11 +13,13 @@ import (
 	"strings"
 	"time"
 
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
 	"gitea.maximumdirect.net/eric/promptkit"
 	httpadapter "gitea.maximumdirect.net/eric/scriptorium/internal/adapter/http"
 	appconfig "gitea.maximumdirect.net/eric/scriptorium/internal/config"
 	"gitea.maximumdirect.net/eric/scriptorium/internal/defaults"
 	renderformat "gitea.maximumdirect.net/eric/scriptorium/internal/format"
+	executionreport "gitea.maximumdirect.net/eric/scriptorium/internal/report"
 )
 
 const (
@@ -32,36 +35,43 @@ const (
 type runConfig struct {
 	configPath string
 
-	promptDir       string
-	profileDir      string
-	promptID        string
-	promptVersion   string
-	profileID       string
-	sessionID       string
-	inputRaw        listFlag
-	varRaw          listFlag
-	outputPath      string
-	llmBaseURL      string
-	apiKeyEnv       string
-	model           string
-	temperature     float64
-	maxTokens       int
-	topP            float64
-	reasoningEffort string
-	schemaDir       string
-	backends        []appconfig.BackendSettings
-	timeout         time.Duration
+	promptDir                  string
+	profileDir                 string
+	promptID                   string
+	promptVersion              string
+	profileID                  string
+	sessionID                  string
+	inputRaw                   listFlag
+	varRaw                     listFlag
+	outputPath                 string
+	executionReportPath        string
+	llmBaseURL                 string
+	apiKeyEnv                  string
+	model                      string
+	temperature                float64
+	maxTokens                  int
+	topP                       float64
+	reasoningEffort            string
+	schemaDir                  string
+	backends                   []appconfig.BackendSettings
+	timeout                    time.Duration
+	runTimeout                 time.Duration
+	transientRetries           int
+	configuredTransientRetries *int
+	diagnostics                []configkit.Diagnostic
+	configOverlay              string
 
 	defaultRenderFormat renderformat.PreparedRunOutputFormat
 
-	llmBaseURLSet      bool
-	apiKeyEnvSet       bool
-	modelSet           bool
-	temperatureSet     bool
-	maxTokensSet       bool
-	topPSet            bool
-	reasoningEffortSet bool
-	timeoutSet         bool
+	llmBaseURLSet       bool
+	apiKeyEnvSet        bool
+	modelSet            bool
+	temperatureSet      bool
+	maxTokensSet        bool
+	topPSet             bool
+	reasoningEffortSet  bool
+	timeoutSet          bool
+	transientRetriesSet bool
 }
 
 type renderConfig struct {
@@ -81,6 +91,9 @@ type serveConfig struct {
 	maxArtifactBytes int64
 	maxResponseBytes int64
 	backends         []appconfig.BackendSettings
+	transientRetries *int
+	diagnostics      []configkit.Diagnostic
+	configOverlay    string
 }
 
 type commonCommandSettings struct {
@@ -94,6 +107,8 @@ type commonCommandSettings struct {
 	maxResponseBytes    int64
 	defaultRenderFormat renderformat.PreparedRunOutputFormat
 	backends            []appconfig.BackendSettings
+	transientRetries    *int
+	diagnostics         []configkit.Diagnostic
 }
 
 type listFlag []string
@@ -122,6 +137,8 @@ func Run(args []string, stdout, stderr io.Writer) int {
 		return serveCommand(args[1:], stderr)
 	case "inspect":
 		return inspectCommand(args[1:], stdout, stderr)
+	case "config":
+		return configCommand(args[1:], stdout, stderr)
 	default:
 		fmt.Fprintf(stderr, "unknown command %q\n", args[0])
 		printUsage(stderr)
@@ -132,7 +149,10 @@ func Run(args []string, stdout, stderr io.Writer) int {
 func runCommand(args []string, stdout, stderr io.Writer) int {
 	cfg, err := parseRunArgs(args)
 	if err != nil {
-		fmt.Fprintf(stderr, "run parse error: %v\n", err)
+		writeApplicationError(stderr, "run parse", err)
+		return ExitRuntimeError
+	}
+	if err := writeApplicationWarnings(stderr, cfg.diagnostics); err != nil {
 		return ExitRuntimeError
 	}
 
@@ -147,15 +167,48 @@ func runCommand(args []string, stdout, stderr io.Writer) int {
 		fmt.Fprintf(stderr, "engine error: %v\n", err)
 		return ExitRuntimeError
 	}
+	return executeRun(cfg, req, engine.Run, stdout, stderr)
+}
 
-	res, runErr := engine.Run(context.Background(), req)
+func executeRun(cfg *runConfig, req promptkit.RunRequest, run func(context.Context, promptkit.RunRequest) (*promptkit.RunResult, error), stdout, stderr io.Writer) int {
+	var collector *executionreport.Collector
+	if cfg.executionReportPath != "" {
+		collector = executionreport.NewCollector()
+		req.Observer = collector.Observe
+	}
+	runContext := context.Background()
+	cancel := func() {}
+	if cfg.runTimeout > 0 {
+		runContext, cancel = context.WithTimeout(runContext, cfg.runTimeout)
+	}
+	res, runErr := run(runContext, req)
+	contextErr := runContext.Err()
+	cancel()
 	if runErr != nil {
-		fmt.Fprintln(stderr, runErrorMessage(runErr))
-		return ExitRuntimeError
+		if errors.Is(contextErr, context.DeadlineExceeded) {
+			fmt.Fprintln(stderr, "run error: execution deadline exceeded")
+		} else {
+			fmt.Fprintln(stderr, runErrorMessage(runErr))
+		}
 	}
 
-	if err := writeOutput(stdout, cfg.outputPath, res.Artifact.Body); err != nil {
-		fmt.Fprintf(stderr, "output write error: %v\n", err)
+	outputFailed := false
+	if runErr == nil {
+		if err := writeOutput(stdout, cfg.outputPath, res.Artifact.Body); err != nil {
+			fmt.Fprintf(stderr, "output write error: %v\n", err)
+			outputFailed = true
+		}
+	}
+	reportFailed := false
+	if collector != nil {
+		if value := collector.Snapshot(); value == nil {
+			fmt.Fprintln(stderr, "execution report not produced: no terminal observation")
+		} else if err := writeExecutionReport(cfg.executionReportPath, value); err != nil {
+			fmt.Fprintf(stderr, "execution report write error: %v\n", err)
+			reportFailed = true
+		}
+	}
+	if runErr != nil || outputFailed || reportFailed {
 		return ExitRuntimeError
 	}
 
@@ -163,10 +216,21 @@ func runCommand(args []string, stdout, stderr io.Writer) int {
 	return determineExitCode(nil, res)
 }
 
+func writeExecutionReport(path string, value *executionreport.Execution) error {
+	data, err := json.Marshal(value)
+	if err != nil {
+		return err
+	}
+	return os.WriteFile(path, append(data, '\n'), 0644)
+}
+
 func renderCommand(args []string, stdout, stderr io.Writer) int {
 	cfg, err := parseRenderArgs(args)
 	if err != nil {
-		fmt.Fprintf(stderr, "render parse error: %v\n", err)
+		writeApplicationError(stderr, "render parse", err)
+		return ExitRuntimeError
+	}
+	if err := writeApplicationWarnings(stderr, cfg.diagnostics); err != nil {
 		return ExitRuntimeError
 	}
 
@@ -204,7 +268,10 @@ func renderCommand(args []string, stdout, stderr io.Writer) int {
 func serveCommand(args []string, stderr io.Writer) int {
 	cfg, err := parseServeArgs(args)
 	if err != nil {
-		fmt.Fprintf(stderr, "serve parse error: %v\n", err)
+		writeApplicationError(stderr, "serve parse", err)
+		return ExitRuntimeError
+	}
+	if err := writeApplicationWarnings(stderr, cfg.diagnostics); err != nil {
 		return ExitRuntimeError
 	}
 
@@ -243,9 +310,11 @@ func parseRunArgs(args []string) (*runConfig, error) {
 	fs := flag.NewFlagSet("run", flag.ContinueOnError)
 	fs.SetOutput(io.Discard)
 
-	registerConfigPathFlag(fs, &cfg.configPath)
+	registerConfigSelectionFlags(fs, &cfg.configPath, &cfg.configOverlay)
 	registerExecutionRequestFlags(fs, cfg)
 	fs.StringVar(&cfg.schemaDir, "schema-dir", "", "base directory for validation schemas")
+	fs.DurationVar(&cfg.runTimeout, "run-timeout", 0, "whole-run execution deadline")
+	fs.StringVar(&cfg.executionReportPath, "execution-report", "", "JSON execution report file path")
 
 	if err := fs.Parse(args); err != nil {
 		return nil, err
@@ -253,11 +322,44 @@ func parseRunArgs(args []string) (*runConfig, error) {
 	if err := finalizeExecutionRequestConfig(fs, cfg); err != nil {
 		return nil, err
 	}
-	cfg.schemaDir = filepath.Clean(cfg.schemaDir)
-
+	if cfg.runTimeout < 0 {
+		return nil, errors.New("--run-timeout must be greater than or equal to 0")
+	}
+	if flagWasSet(fs, "execution-report") {
+		if cfg.executionReportPath == "" {
+			return nil, errors.New("--execution-report requires a nonempty path")
+		}
+		cfg.executionReportPath = filepath.Clean(cfg.executionReportPath)
+		if cfg.outputPath != "" {
+			collision, err := outputPathsCollide(cfg.outputPath, cfg.executionReportPath)
+			if err != nil {
+				return nil, err
+			}
+			if collision {
+				return nil, errors.New("--out and --execution-report select the same file")
+			}
+		}
+	}
 	return cfg, nil
 }
 
+func outputPathsCollide(first, second string) (bool, error) {
+	firstAbs, err := filepath.Abs(first)
+	if err != nil {
+		return false, err
+	}
+	secondAbs, err := filepath.Abs(second)
+	if err != nil {
+		return false, err
+	}
+	if firstAbs == secondAbs {
+		return true, nil
+	}
+	firstInfo, firstErr := os.Stat(firstAbs)
+	secondInfo, secondErr := os.Stat(secondAbs)
+	return firstErr == nil && secondErr == nil && os.SameFile(firstInfo, secondInfo), nil
+}
+
 func parseRenderArgs(args []string) (*renderConfig, error) {
 	cfg := &renderConfig{
 		outputFormat: renderformat.DefaultPreparedRunOutputFormat,
@@ -265,8 +367,9 @@ func parseRenderArgs(args []string) (*renderConfig, error) {
 	fs := flag.NewFlagSet("render", flag.ContinueOnError)
 	fs.SetOutput(io.Discard)
 
-	registerConfigPathFlag(fs, &cfg.configPath)
+	registerConfigSelectionFlags(fs, &cfg.configPath, &cfg.configOverlay)
 	registerExecutionRequestFlags(fs, &cfg.runConfig)
+	fs.StringVar(&cfg.schemaDir, "schema-dir", "", "base directory for validation schemas")
 
 	var rawFormat string
 	fs.StringVar(&rawFormat, "format", "", "render output format (text|json)")
@@ -295,7 +398,7 @@ func parseServeArgs(args []string) (*serveConfig, error) {
 	fs := flag.NewFlagSet("serve", flag.ContinueOnError)
 	fs.SetOutput(io.Discard)
 
-	registerConfigPathFlag(fs, &cfg.configPath)
+	registerConfigSelectionFlags(fs, &cfg.configPath, &cfg.configOverlay)
 	fs.StringVar(&cfg.addr, "addr", "", "HTTP listen address")
 	fs.StringVar(&cfg.promptDir, "prompt-dir", "", "directory containing prompt definition YAML files")
 	fs.StringVar(&cfg.profileDir, "profile-dir", "", "directory containing execution profile YAML files")
@@ -312,16 +415,7 @@ func parseServeArgs(args []string) (*serveConfig, error) {
 		return nil, fmt.Errorf("unexpected positional args: %v", fs.Args())
 	}
 
-	settings, err := resolveCommonSettings(fs, cfg.configPath, appconfig.CLIOverrides{
-		PromptDir:        cfg.promptDirIfSet(fs),
-		ProfileDir:       cfg.profileDirIfSet(fs),
-		SchemaDir:        cfg.schemaDirIfSet(fs),
-		ServerAddr:       cfg.addrIfSet(fs),
-		ArtifactRoot:     cfg.artifactRootIfSet(fs),
-		MaxRequestBytes:  cfg.maxRequestBytesIfSet(fs),
-		MaxArtifactBytes: cfg.maxArtifactBytesIfSet(fs),
-		MaxResponseBytes: cfg.maxResponseBytesIfSet(fs),
-	})
+	settings, err := resolveCommonSettings(fs, cfg.configPath, cfg.configOverlay)
 	if err != nil {
 		return nil, err
 	}
@@ -335,19 +429,13 @@ func parseServeArgs(args []string) (*serveConfig, error) {
 	cfg.maxArtifactBytes = settings.maxArtifactBytes
 	cfg.maxResponseBytes = settings.maxResponseBytes
 	cfg.backends = settings.backends
+	cfg.transientRetries = settings.transientRetries
+	cfg.diagnostics = settings.diagnostics
 
 	if err := validateRequiredLibraryDirs(cfg.promptDir); err != nil {
 		return nil, err
 	}
 
-	cfg.promptDir = filepath.Clean(cfg.promptDir)
-	if strings.TrimSpace(cfg.profileDir) != "" {
-		cfg.profileDir = filepath.Clean(cfg.profileDir)
-	}
-	cfg.schemaDir = filepath.Clean(cfg.schemaDir)
-	if strings.TrimSpace(cfg.artifactRoot) != "" {
-		cfg.artifactRoot = filepath.Clean(cfg.artifactRoot)
-	}
 	return cfg, nil
 }
 
@@ -369,6 +457,7 @@ func registerExecutionRequestFlags(fs *flag.FlagSet, cfg *runConfig) {
 	fs.Float64Var(&cfg.topP, "top-p", 0, "optional top_p override")
 	fs.StringVar(&cfg.reasoningEffort, "reasoning-effort", "", "optional reasoning effort override")
 	fs.DurationVar(&cfg.timeout, "timeout", 0, "LLM request timeout")
+	fs.IntVar(&cfg.transientRetries, "transient-retries", 0, "additional transient generation attempts")
 	fs.StringVar(&cfg.promptID, "prompt-id", "", "deprecated alias for --prompt")
 	fs.StringVar(&cfg.profileID, "profile-id", "", "deprecated alias for --profile")
 }
@@ -378,11 +467,7 @@ func finalizeExecutionRequestConfig(fs *flag.FlagSet, cfg *runConfig) error {
 		return fmt.Errorf("unexpected positional args: %v", fs.Args())
 	}
 
-	settings, err := resolveCommonSettings(fs, cfg.configPath, appconfig.CLIOverrides{
-		PromptDir:  cfg.promptDirIfSet(fs),
-		ProfileDir: cfg.profileDirIfSet(fs),
-		SchemaDir:  cfg.schemaDirIfSet(fs),
-	})
+	settings, err := resolveCommonSettings(fs, cfg.configPath, cfg.configOverlay)
 	if err != nil {
 		return err
 	}
@@ -392,6 +477,8 @@ func finalizeExecutionRequestConfig(fs *flag.FlagSet, cfg *runConfig) error {
 	cfg.schemaDir = settings.schemaDir
 	cfg.defaultRenderFormat = settings.defaultRenderFormat
 	cfg.backends = settings.backends
+	cfg.configuredTransientRetries = settings.transientRetries
+	cfg.diagnostics = settings.diagnostics
 
 	if err := validateRequiredLibraryDirs(cfg.promptDir); err != nil {
 		return err
@@ -399,10 +486,6 @@ func finalizeExecutionRequestConfig(fs *flag.FlagSet, cfg *runConfig) error {
 	if strings.TrimSpace(cfg.promptID) == "" {
 		return errors.New("--prompt is required")
 	}
-	cfg.promptDir = filepath.Clean(cfg.promptDir)
-	if strings.TrimSpace(cfg.profileDir) != "" {
-		cfg.profileDir = filepath.Clean(cfg.profileDir)
-	}
 	if cfg.outputPath != "" {
 		cfg.outputPath = filepath.Clean(cfg.outputPath)
 	}
@@ -414,87 +497,14 @@ func finalizeExecutionRequestConfig(fs *flag.FlagSet, cfg *runConfig) error {
 	cfg.topPSet = flagWasSet(fs, "top-p")
 	cfg.reasoningEffortSet = flagWasSet(fs, "reasoning-effort")
 	cfg.timeoutSet = flagWasSet(fs, "timeout")
-	return nil
-}
-
-func (c *runConfig) promptDirIfSet(fs *flag.FlagSet) string {
-	if flagWasSet(fs, "prompt-dir") {
-		return c.promptDir
-	}
-	return ""
-}
-
-func (c *runConfig) profileDirIfSet(fs *flag.FlagSet) string {
-	if flagWasSet(fs, "profile-dir") {
-		return c.profileDir
-	}
-	return ""
-}
-
-func (c *runConfig) schemaDirIfSet(fs *flag.FlagSet) string {
-	if flagWasSet(fs, "schema-dir") {
-		return c.schemaDir
-	}
-	return ""
-}
-
-func (c *serveConfig) promptDirIfSet(fs *flag.FlagSet) string {
-	if flagWasSet(fs, "prompt-dir") {
-		return c.promptDir
-	}
-	return ""
-}
-
-func (c *serveConfig) profileDirIfSet(fs *flag.FlagSet) string {
-	if flagWasSet(fs, "profile-dir") {
-		return c.profileDir
-	}
-	return ""
-}
-
-func (c *serveConfig) schemaDirIfSet(fs *flag.FlagSet) string {
-	if flagWasSet(fs, "schema-dir") {
-		return c.schemaDir
-	}
-	return ""
-}
-
-func (c *serveConfig) addrIfSet(fs *flag.FlagSet) string {
-	if flagWasSet(fs, "addr") {
-		return c.addr
-	}
-	return ""
-}
-
-func (c *serveConfig) artifactRootIfSet(fs *flag.FlagSet) string {
-	if flagWasSet(fs, "artifact-root") {
-		return c.artifactRoot
-	}
-	return ""
-}
-
-func (c *serveConfig) maxRequestBytesIfSet(fs *flag.FlagSet) *int64 {
-	if flagWasSet(fs, "max-request-bytes") {
-		return &c.maxRequestBytes
+	cfg.transientRetriesSet = flagWasSet(fs, "transient-retries")
+	if cfg.transientRetriesSet && cfg.transientRetries < 0 {
+		return errors.New("--transient-retries must be greater than or equal to 0")
 	}
 	return nil
 }
 
-func (c *serveConfig) maxArtifactBytesIfSet(fs *flag.FlagSet) *int64 {
-	if flagWasSet(fs, "max-artifact-bytes") {
-		return &c.maxArtifactBytes
-	}
-	return nil
-}
-
-func (c *serveConfig) maxResponseBytesIfSet(fs *flag.FlagSet) *int64 {
-	if flagWasSet(fs, "max-response-bytes") {
-		return &c.maxResponseBytes
-	}
-	return nil
-}
-
-func registerConfigPathFlag(fs *flag.FlagSet, target *string) {
+func registerConfigSelectionFlags(fs *flag.FlagSet, target, overlay *string) {
 	fs.StringVar(
 		target,
 		"config",
@@ -505,28 +515,11 @@ func registerConfigPathFlag(fs *flag.FlagSet, target *string) {
 			appconfig.DefaultConfigPath,
 		),
 	)
+	fs.StringVar(overlay, "config-overlay", "", "selected application configuration overlay")
 }
 
-func resolveAppSettings(fs *flag.FlagSet, configPath string, overrides appconfig.CLIOverrides) (appconfig.AppSettings, error) {
-	return resolveAppSettingsWithConfigPresence(configPath, flagWasSet(fs, "config"), overrides)
-}
-
-func resolveAppSettingsWithConfigPresence(configPath string, configExplicit bool, overrides appconfig.CLIOverrides) (appconfig.AppSettings, error) {
-	settings, err := appconfig.LoadConfig(configPath, configExplicit)
-	if err != nil {
-		return appconfig.AppSettings{}, fmt.Errorf("application config error: %w", err)
-	}
-
-	merged, err := appconfig.ApplyCLIOverrides(settings, overrides)
-	if err != nil {
-		return appconfig.AppSettings{}, fmt.Errorf("application config error: %w", err)
-	}
-
-	return merged, nil
-}
-
-func resolveCommonSettings(fs *flag.FlagSet, configPath string, overrides appconfig.CLIOverrides) (commonCommandSettings, error) {
-	settings, err := resolveAppSettings(fs, configPath, overrides)
+func resolveCommonSettings(fs *flag.FlagSet, configPath, overlay string) (commonCommandSettings, error) {
+	settings, diagnostics, err := loadApplicationSettings(configPath, overlay, flagWasSet(fs, "config"), applicationOverrides(fs))
 	if err != nil {
 		return commonCommandSettings{}, err
 	}
@@ -541,6 +534,8 @@ func resolveCommonSettings(fs *flag.FlagSet, configPath string, overrides appcon
 		maxResponseBytes:    settings.MaxResponseBytes,
 		defaultRenderFormat: settings.DefaultRenderFormat,
 		backends:            settings.Backends,
+		transientRetries:    settings.TransientRetries,
+		diagnostics:         diagnostics,
 	}, nil
 }
 
@@ -552,18 +547,19 @@ func validateRequiredLibraryDirs(promptDir string) error {
 }
 
 type engineSettings struct {
-	promptDir  string
-	profileDir string
-	schemaDir  string
-	backends   []appconfig.BackendSettings
+	promptDir        string
+	profileDir       string
+	schemaDir        string
+	backends         []appconfig.BackendSettings
+	transientRetries *int
 }
 
 func (c runConfig) engineSettings() engineSettings {
-	return engineSettings{promptDir: c.promptDir, profileDir: c.profileDir, schemaDir: c.schemaDir, backends: c.backends}
+	return engineSettings{promptDir: c.promptDir, profileDir: c.profileDir, schemaDir: c.schemaDir, backends: c.backends, transientRetries: c.configuredTransientRetries}
 }
 
 func (c serveConfig) engineSettings() engineSettings {
-	return engineSettings{promptDir: c.promptDir, profileDir: c.profileDir, schemaDir: c.schemaDir, backends: c.backends}
+	return engineSettings{promptDir: c.promptDir, profileDir: c.profileDir, schemaDir: c.schemaDir, backends: c.backends, transientRetries: c.transientRetries}
 }
 
 func newEngine(settings engineSettings, options ...promptkit.Option) (*promptkit.Engine, error) {
@@ -580,9 +576,10 @@ func newEngine(settings engineSettings, options ...promptkit.Option) (*promptkit
 	}
 	engineOptions = append(engineOptions, options...)
 	engine, err := promptkit.NewEngine(promptkit.Config{
-		PromptDir:  settings.promptDir,
-		ProfileDir: settings.profileDir,
-		SchemaDir:  settings.schemaDir,
+		PromptDir:        settings.promptDir,
+		ProfileDir:       settings.profileDir,
+		SchemaDir:        settings.schemaDir,
+		TransientRetries: settings.transientRetries,
 	}, engineOptions...)
 	if err != nil {
 		return nil, fmt.Errorf("engine initialization from application configuration: %w", err)
@@ -641,14 +638,19 @@ func buildRunRequestFromConfig(cfg *runConfig) (promptkit.RunRequest, error) {
 		}
 	}
 
+	var transientRetries *int
+	if cfg.transientRetriesSet {
+		transientRetries = &cfg.transientRetries
+	}
 	return promptkit.RunRequest{
-		PromptID:      cfg.promptID,
-		PromptVersion: cfg.promptVersion,
-		ProfileID:     cfg.profileID,
-		SessionID:     cfg.sessionID,
-		Inputs:        inputs,
-		Vars:          varMappings,
-		Execution:     modelOverride,
+		PromptID:         cfg.promptID,
+		PromptVersion:    cfg.promptVersion,
+		ProfileID:        cfg.profileID,
+		SessionID:        cfg.sessionID,
+		Inputs:           inputs,
+		Vars:             varMappings,
+		Execution:        modelOverride,
+		TransientRetries: transientRetries,
 	}, nil
 }
 
@@ -752,10 +754,11 @@ func runErrorMessage(err error) string {
 }
 
 func printUsage(w io.Writer) {
-	fmt.Fprintln(w, "usage: scriptorium <run|render|serve|inspect> ...")
-	fmt.Fprintln(w, "  run:   scriptorium run [--config PATH] [--prompt-dir DIR] [--profile-dir DIR] --prompt ID [--prompt-version VERSION] [--input name=path] [--profile ID] [--session-id ID] [--llm-base-url URL] [--model NAME] [--api-key-env ENV] [--temperature N] [--max-tokens N] [--top-p N] [--reasoning-effort VALUE] [--var k=v] [--out path] [--timeout 10m]")
-	fmt.Fprintln(w, "  render: scriptorium render [--config PATH] [--prompt-dir DIR] [--profile-dir DIR] --prompt ID [--prompt-version VERSION] [--input name=path] [--profile ID] [--session-id ID] [--llm-base-url URL] [--model NAME] [--api-key-env ENV] [--temperature N] [--max-tokens N] [--top-p N] [--reasoning-effort VALUE] [--var k=v] [--format text|json] [--out path] [--timeout 10m]")
-	fmt.Fprintf(w, "  serve: scriptorium serve [--config PATH] [--addr %s] [--prompt-dir DIR] [--profile-dir DIR] [--schema-dir DIR] [--artifact-root DIR] [--max-request-bytes N] [--max-artifact-bytes N] [--max-response-bytes N]\n", defaults.HTTPAddrDefault)
-	fmt.Fprintln(w, "  inspect prompt: scriptorium inspect prompt --prompt ID [--prompt-version VERSION] [--config PATH] [--prompt-dir DIR] [--format text|json] [--out path]")
-	fmt.Fprintln(w, "  inspect profile: scriptorium inspect profile --profile ID [--config PATH] [--profile-dir DIR] [--format text|json] [--out path]")
+	fmt.Fprintln(w, "usage: scriptorium <run|render|serve|inspect|config> ...")
+	fmt.Fprintln(w, "  run:   scriptorium run [--config PATH] [--config-overlay NAME] [--prompt-dir DIR] [--profile-dir DIR] [--schema-dir DIR] --prompt ID [--prompt-version VERSION] [--input name=path] [--profile ID] [--session-id ID] [--llm-base-url URL] [--model NAME] [--api-key-env ENV] [--temperature N] [--max-tokens N] [--top-p N] [--reasoning-effort VALUE] [--var k=v] [--out path] [--execution-report path] [--timeout 10m] [--run-timeout 10m] [--transient-retries N]")
+	fmt.Fprintln(w, "  render: scriptorium render [--config PATH] [--config-overlay NAME] [--prompt-dir DIR] [--profile-dir DIR] [--schema-dir DIR] --prompt ID [--prompt-version VERSION] [--input name=path] [--profile ID] [--session-id ID] [--llm-base-url URL] [--model NAME] [--api-key-env ENV] [--temperature N] [--max-tokens N] [--top-p N] [--reasoning-effort VALUE] [--var k=v] [--format text|json] [--out path] [--timeout 10m] [--transient-retries N]")
+	fmt.Fprintf(w, "  serve: scriptorium serve [--config PATH] [--config-overlay NAME] [--addr %s] [--prompt-dir DIR] [--profile-dir DIR] [--schema-dir DIR] [--artifact-root DIR] [--max-request-bytes N] [--max-artifact-bytes N] [--max-response-bytes N]\n", defaults.HTTPAddrDefault)
+	fmt.Fprintln(w, "  inspect prompt: scriptorium inspect prompt --prompt ID [--prompt-version VERSION] [--config PATH] [--config-overlay NAME] [--prompt-dir DIR] [--format text|json] [--out path]")
+	fmt.Fprintln(w, "  inspect profile: scriptorium inspect profile --profile ID [--config PATH] [--config-overlay NAME] [--profile-dir DIR] [--format text|json] [--out path]")
+	fmt.Fprintln(w, "  config: scriptorium config <validate|show|sources|explain> [--config PATH] [--config-overlay NAME] [application overrides] [--path /pointer for explain]")
 }

internal/adapter/cli/run_test.go

@@ -2,9 +2,9 @@ package cli
 
 import (
 	"bytes"
+	"context"
 	"encoding/json"
 	"errors"
-	"flag"
 	"fmt"
 	"io"
 	"net/http"
@@ -17,8 +17,8 @@ import (
 	"testing"
 	"time"
 
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
 	"gitea.maximumdirect.net/eric/promptkit"
-	appconfig "gitea.maximumdirect.net/eric/scriptorium/internal/config"
 	"gitea.maximumdirect.net/eric/scriptorium/internal/defaults"
 	renderformat "gitea.maximumdirect.net/eric/scriptorium/internal/format"
 )
@@ -64,7 +64,7 @@ func TestParseMappingsMalformed(t *testing.T) {
 }
 
 func TestParseRunArgsRequiredFlags(t *testing.T) {
-	configPath := writeAppConfigFile(t, "")
+	configPath := writeAppConfigFile(t, "{}\n")
 
 	_, err := parseRunArgs([]string{"--config", configPath, "--profile-dir", "./profiles", "--prompt", "p", "--input", "a=b"})
 	if err == nil {
@@ -160,7 +160,7 @@ func TestParseRunArgsRejectsRawLLMAPIKeyFlag(t *testing.T) {
 }
 
 func TestParseServeArgsRequiredFlags(t *testing.T) {
-	configPath := writeAppConfigFile(t, "")
+	configPath := writeAppConfigFile(t, "{}\n")
 
 	_, err := parseServeArgs([]string{"--config", configPath, "--profile-dir", "./profiles"})
 	if err == nil {
@@ -363,22 +363,9 @@ func TestParseRunArgsInvalidExplicitConfigReturnsError(t *testing.T) {
 	if err == nil {
 		t.Fatal("expected invalid explicit config error")
 	}
-	if !strings.Contains(err.Error(), "application config error") {
-		t.Fatalf("expected application config context, got %v", err)
-	}
-}
-
-func TestResolveAppSettingsMissingImplicitConfigDoesNotError(t *testing.T) {
-	fs := flag.NewFlagSet("test", flag.ContinueOnError)
-	settings, err := resolveAppSettings(fs, filepath.Join(t.TempDir(), "missing.yml"), appconfig.CLIOverrides{})
-	if err != nil {
-		t.Fatalf("expected no error, got %v", err)
-	}
-	if settings.SchemaDir != defaults.SchemaDirDefault {
-		t.Fatalf("expected built-in schema dir, got %q", settings.SchemaDir)
-	}
-	if settings.ServerAddr != defaults.HTTPAddrDefault {
-		t.Fatalf("expected built-in server addr, got %q", settings.ServerAddr)
+	var loadError *configkit.LoadError
+	if !errors.As(err, &loadError) {
+		t.Fatalf("expected structured application config error, got %v", err)
 	}
 }
 
@@ -412,6 +399,85 @@ schema_dir: ./from-config/schemas
 	}
 }
 
+func TestApplicationFlagsPreserveExplicitEmptyValues(t *testing.T) {
+	configPath := writeAppConfigFile(t, `prompt_dir: ./configured/prompts
+profile_dir: ./configured/profiles
+schema_dir: ./configured/schemas
+server:
+  addr: ":9090"
+  artifact_root: ./configured/artifacts
+`)
+	if _, err := parseRunArgs([]string{"--config", configPath, "--prompt", "p", "--prompt-dir="}); err == nil || !strings.Contains(err.Error(), "prompt directory is required") {
+		t.Fatalf("explicit empty prompt directory restored lower value: %v", err)
+	}
+	render, err := parseRenderArgs([]string{"--config", configPath, "--prompt", "p", "--profile-dir=", "--schema-dir="})
+	if err != nil || render.profileDir != "" || render.schemaDir != "" {
+		t.Fatalf("render did not preserve explicit empty sources: cfg=%+v err=%v", render, err)
+	}
+	serve, err := parseServeArgs([]string{"--config", configPath, "--addr=", "--artifact-root="})
+	if err != nil || serve.addr != "" || serve.artifactRoot != "" {
+		t.Fatalf("serve did not preserve explicit empty settings: cfg=%+v err=%v", serve, err)
+	}
+
+	prompt, err := parsePromptInspectionArgs([]string{"--config", configPath, "--prompt", "p", "--prompt-dir="})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if _, err := resolveAppSettingsForPromptInspection(prompt); err == nil || !strings.Contains(err.Error(), "prompt directory is required") {
+		t.Fatalf("prompt inspection restored lower directory: %v", err)
+	}
+	profile, err := parseProfileInspectionArgs([]string{"--config", configPath, "--profile", "p", "--profile-dir="})
+	if err != nil {
+		t.Fatal(err)
+	}
+	settings, err := resolveAppSettingsForProfileInspection(profile)
+	if err != nil || settings.profileDir != "" {
+		t.Fatalf("profile inspection required a source or restored lower value: settings=%+v err=%v", settings, err)
+	}
+}
+
+func TestApplicationLoadUsesNativeYAMLAndSafeDiagnostics(t *testing.T) {
+	for _, body := range []string{
+		"",
+		"server: null\n",
+		"server:\n  addr: null\n",
+		"server:\n  unknown: 1\n",
+		"prompt_dir: &alias ./prompts\n",
+		"prompt_dir: ./prompts\n---\nprofile_dir: ./profiles\n",
+	} {
+		path := writeAppConfigFile(t, body)
+		_, err := parseRunArgs([]string{"--config", path, "--prompt-dir", "./prompts", "--prompt", "p"})
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) {
+			t.Fatalf("native YAML failure %q lost structured diagnostics: %v", body, err)
+		}
+	}
+
+	const rejected = "sensitive-synthetic-marker"
+	path := writeAppConfigFile(t, "defaults:\n  render_format: "+rejected+"\n")
+	code, stdout, stderr := runCLICommand(t, renderCommand, []string{"--config", path, "--prompt-dir", "./prompts", "--prompt", "p"})
+	if code != ExitRuntimeError || stdout != "" || !strings.Contains(stderr, "validate.error") || strings.Contains(stderr, rejected) {
+		t.Fatalf("unsafe or misclassified load diagnostic: code=%d stdout=%q stderr=%q", code, stdout, stderr)
+	}
+}
+
+type failingDiagnosticWriter struct{}
+
+func (failingDiagnosticWriter) Write([]byte) (int, error) {
+	return 0, errors.New("diagnostic sink unavailable")
+}
+
+func TestApplicationDiagnosticWriterFailure(t *testing.T) {
+	diagnostics := []configkit.Diagnostic{{Code: "test.error", Message: "safe diagnostic"}}
+	if err := writeApplicationWarnings(failingDiagnosticWriter{}, diagnostics); err == nil {
+		t.Fatal("expected diagnostic writer failure")
+	}
+	path := writeAppConfigFile(t, "unknown_field: secret-value\n")
+	if code := runCommand([]string{"--config", path, "--prompt", "sample"}, io.Discard, failingDiagnosticWriter{}); code != ExitRuntimeError {
+		t.Fatalf("diagnostic writer failure must retain runtime error exit status, got %d", code)
+	}
+}
+
 func TestParseRenderArgsWithExplicitConfigLoadsDirectoriesAndFormat(t *testing.T) {
 	configPath := writeAppConfigFile(t, `
 prompt_dir: ./from-config/prompts
@@ -1005,6 +1071,41 @@ profile_dir: %s
 	}
 }
 
+func TestRenderCommandLoadsImportsAndOverlayWithoutChangingExecutionProfile(t *testing.T) {
+	lib := newCLITestLibrary(t)
+	inputPath := lib.writeInputFile(t, "transcript.md", "hello transcript")
+	writePromptFile(t, lib.promptDir, "prompt.render", "local-default")
+	writeProfileFile(t, lib.profileDir, "local-default", "http://127.0.0.1:1/v1", "profile-model")
+	if err := os.MkdirAll(filepath.Join(lib.rootDir, "fragments"), 0o755); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.MkdirAll(filepath.Join(lib.rootDir, "overlays"), 0o755); err != nil {
+		t.Fatal(err)
+	}
+	base := filepath.Join(lib.rootDir, "config.yml")
+	for path, body := range map[string]string{
+		base: "configkit:\n  imports:\n    - path: fragments/sources.yml\n",
+		filepath.Join(lib.rootDir, "fragments", "sources.yml"): fmt.Sprintf("prompt_dir: %q\nprofile_dir: %q\n", lib.promptDir, lib.profileDir),
+		filepath.Join(lib.rootDir, "overlays", "local.yml"):    "defaults:\n  render_format: json\n",
+	} {
+		if err := os.WriteFile(path, []byte(body), 0o644); err != nil {
+			t.Fatal(err)
+		}
+	}
+	args := []string{"--config", base, "--config-overlay", "local", "--profile", "local-default", "--prompt", "prompt.render", "--input", "transcript=" + inputPath}
+	cfg, err := parseRenderArgs(args)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if cfg.profileID != "local-default" || cfg.outputFormat != renderformat.PreparedRunFormatJSON {
+		t.Fatalf("overlay changed execution profile or output selection: %+v", cfg)
+	}
+	code, stdout, stderr := runCLICommand(t, renderCommand, args)
+	if code != ExitOK || !strings.Contains(stdout, `"selected_profile_id": "local-default"`) {
+		t.Fatalf("render did not use composed config and independent execution profile: code=%d out=%q err=%q", code, stdout, stderr)
+	}
+}
+
 func TestRenderCommandUsesConfiguredCustomBackend(t *testing.T) {
 	lib := newCLITestLibrary(t)
 	writePromptDefinition(t, lib.promptDir, "custom.yaml", `id: custom
@@ -1047,6 +1148,64 @@ backends:
 	}
 }
 
+func TestConfiguredBackendPreservesIntegerParameterAndOptionalCredential(t *testing.T) {
+	lib := newCLITestLibrary(t)
+	writePromptDefinition(t, lib.promptDir, "custom.yaml", `id: custom
+version: "1"
+default_profile: local-gpu
+messages:
+  - role: user
+    content: "hello"
+output:
+  format: text
+  validation_mode: none
+`)
+	if err := os.WriteFile(filepath.Join(lib.profileDir, "local-gpu.yaml"), []byte("id: local-gpu\nbackend: local-gpu\nmodel: local-model\n"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	const credentialName = "SCRIPTORIUM_OPTIONAL_BACKEND_KEY"
+	t.Setenv(credentialName, "")
+	type observedRequest struct {
+		number        json.Number
+		authorization string
+	}
+	observed := make(chan observedRequest, 1)
+	provider := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		request := observedRequest{authorization: r.Header.Get("Authorization")}
+		var body map[string]any
+		decoder := json.NewDecoder(r.Body)
+		decoder.UseNumber()
+		if err := decoder.Decode(&body); err == nil {
+			request.number, _ = body["synthetic_count"].(json.Number)
+		}
+		observed <- request
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = w.Write([]byte(`{"choices":[{"message":{"role":"assistant","content":"generated"}}]}`))
+	}))
+	defer provider.Close()
+	configPath := writeAppConfigFile(t, fmt.Sprintf(`prompt_dir: %q
+profile_dir: %q
+backends:
+  local-gpu:
+    endpoint: %q
+    api_key_env: %s
+    extra_params:
+      synthetic_count: 7
+`, lib.promptDir, lib.profileDir, provider.URL+"/v1", credentialName))
+	cfg, err := parseRunArgs([]string{"--config", configPath, "--prompt", "custom"})
+	if err != nil || len(cfg.backends) != 1 || cfg.backends[0].ExtraParams["synthetic_count"] != int64(7) {
+		t.Fatalf("ConfigKit integer did not reach runtime settings: cfg=%+v err=%v", cfg, err)
+	}
+	code, stdout, stderr := runCLICommand(t, runCommand, []string{"--config", configPath, "--prompt", "custom"})
+	if code != ExitOK {
+		t.Fatalf("backend run failed: code=%d out=%q err=%q", code, stdout, stderr)
+	}
+	request := <-observed
+	if !strings.Contains(stdout, "generated") || request.number != "7" || request.authorization != "" {
+		t.Fatalf("backend request changed: code=%d out=%q err=%q count=%q auth=%q", code, stdout, stderr, request.number, request.authorization)
+	}
+}
+
 func TestRenderCommandMapsReasoningEffortAndSessionID(t *testing.T) {
 	lib := newCLITestLibrary(t)
 	writePromptDefinition(t, lib.promptDir, "session.yaml", `id: session
@@ -1201,9 +1360,9 @@ func TestInspectionParsersPreserveExplicitEmptyConfigPath(t *testing.T) {
 	}
 }
 
-func TestPromptkitV09DefinitionsRenderThroughCLI(t *testing.T) {
-	fixtureRoot := promptkitV09FixtureRoot(t)
-	configPath := writePromptkitV09Config(t, fixtureRoot, true)
+func TestPromptkitDefinitionsRenderThroughCLI(t *testing.T) {
+	fixtureRoot := definitionFixtureRoot(t)
+	configPath := writeDefinitionFixtureConfig(t, fixtureRoot, true)
 	inputPath := filepath.Join(fixtureRoot, "inputs", "source.md")
 
 	code, stdout, stderr := runCLICommand(t, renderCommand, []string{
@@ -1258,9 +1417,290 @@ func TestPromptkitV09DefinitionsRenderThroughCLI(t *testing.T) {
 	}
 }
 
-func TestPromptkitV09PromptInspectionSelectsVersionsAndContracts(t *testing.T) {
-	fixtureRoot := promptkitV09FixtureRoot(t)
-	configPath := writePromptkitV09Config(t, fixtureRoot, true)
+func TestTransientRetrySettingsAndCLIOverride(t *testing.T) {
+	lib := newCLITestLibrary(t)
+	inputPath := lib.writeInputFile(t, "transcript.md", "hello")
+	writePromptFile(t, lib.promptDir, "retry.prompt", "local-default")
+	var hits int32
+	provider := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if atomic.AddInt32(&hits, 1) == 1 {
+			http.Error(w, "temporary", http.StatusServiceUnavailable)
+			return
+		}
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = w.Write([]byte(`{"choices":[{"message":{"role":"assistant","content":"recovered"}}]}`))
+	}))
+	defer provider.Close()
+	writeProfileFile(t, lib.profileDir, "local-default", provider.URL+"/v1", "local-model")
+	configPath := writeAppConfigFile(t, fmt.Sprintf("prompt_dir: %q\nprofile_dir: %q\ntransient_retries: 1\n", lib.promptDir, lib.profileDir))
+	args := []string{"--config", configPath, "--prompt", "retry.prompt", "--input", "transcript=" + inputPath}
+
+	serve, err := parseServeArgs([]string{"--config", configPath})
+	if err != nil || serve.transientRetries == nil || *serve.transientRetries != 1 {
+		t.Fatalf("server did not inherit retry setting: cfg=%+v err=%v", serve, err)
+	}
+	if _, err := newEngine(serve.engineSettings()); err != nil {
+		t.Fatalf("server engine configuration: %v", err)
+	}
+
+	code, stdout, stderr := runCLICommand(t, runCommand, args)
+	if code != ExitOK || stdout != "recovered" || atomic.LoadInt32(&hits) != 2 {
+		t.Fatalf("configured retry was not applied: code=%d stdout=%q stderr=%q hits=%d", code, stdout, stderr, atomic.LoadInt32(&hits))
+	}
+	atomic.StoreInt32(&hits, 0)
+	code, stdout, stderr = runCLICommand(t, runCommand, append(append([]string{}, args...), "--transient-retries", "0"))
+	if code != ExitRuntimeError || stdout != "" || atomic.LoadInt32(&hits) != 1 {
+		t.Fatalf("explicit zero did not override configuration: code=%d stdout=%q stderr=%q hits=%d", code, stdout, stderr, atomic.LoadInt32(&hits))
+	}
+
+	for _, tc := range []struct {
+		name string
+		args []string
+		want *int
+	}{
+		{name: "omitted", args: args},
+		{name: "zero", args: append(append([]string{}, args...), "--transient-retries", "0"), want: intPointer(0)},
+		{name: "positive", args: append(append([]string{}, args...), "--transient-retries", "3"), want: intPointer(3)},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			cfg, err := parseRunArgs(tc.args)
+			if err != nil {
+				t.Fatal(err)
+			}
+			req, err := buildRunRequestFromConfig(cfg)
+			if err != nil {
+				t.Fatal(err)
+			}
+			if !reflect.DeepEqual(req.TransientRetries, tc.want) {
+				t.Fatalf("request retry override: got %v, want %v", req.TransientRetries, tc.want)
+			}
+		})
+	}
+	for _, invalid := range []string{"-1", "wrong"} {
+		if _, err := parseRunArgs(append(append([]string{}, args...), "--transient-retries", invalid)); err == nil {
+			t.Fatalf("run accepted invalid retry count %q", invalid)
+		}
+		if _, err := parseRenderArgs(append(append([]string{}, args...), "--transient-retries", invalid)); err == nil {
+			t.Fatalf("render accepted invalid retry count %q", invalid)
+		}
+	}
+	code, stdout, stderr = runCLICommand(t, renderCommand, append(append([]string{}, args...), "--transient-retries", "0", "--format", "json"))
+	if code != ExitOK || strings.Contains(stdout, "transient_retries") {
+		t.Fatalf("render retry override changed prepared output: code=%d stdout=%q stderr=%q", code, stdout, stderr)
+	}
+}
+
+func TestRunTimeoutParsingAndCancellation(t *testing.T) {
+	lib := newCLITestLibrary(t)
+	inputPath := lib.writeInputFile(t, "transcript.md", "hello")
+	args := []string{"--prompt-dir", lib.promptDir, "--profile-dir", lib.profileDir, "--prompt", "timed", "--input", "transcript=" + inputPath}
+	for _, tc := range []struct {
+		name string
+		flag []string
+		want time.Duration
+	}{
+		{name: "omitted"},
+		{name: "zero", flag: []string{"--run-timeout", "0s"}},
+		{name: "subsecond", flag: []string{"--run-timeout", "125ms"}, want: 125 * time.Millisecond},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			cfg, err := parseRunArgs(append(append([]string{}, args...), tc.flag...))
+			if err != nil || cfg.runTimeout != tc.want {
+				t.Fatalf("run timeout: cfg=%+v err=%v", cfg, err)
+			}
+		})
+	}
+	if _, err := parseRunArgs(append(append([]string{}, args...), "--run-timeout", "-1ms")); err == nil {
+		t.Fatal("negative run timeout accepted")
+	}
+	if _, err := parseRenderArgs(append(append([]string{}, args...), "--run-timeout", "1s")); err == nil {
+		t.Fatal("render accepted run-only timeout")
+	}
+
+	started := make(chan struct{}, 1)
+	release := make(chan struct{})
+	provider := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		started <- struct{}{}
+		<-release
+	}))
+	defer provider.Close()
+	defer close(release)
+	writePromptFile(t, lib.promptDir, "timed", "local-default")
+	writeProfileFile(t, lib.profileDir, "local-default", provider.URL+"/v1", "local-model")
+	code, stdout, stderr := runCLICommand(t, runCommand, append(append([]string{}, args...), "--run-timeout", "500ms", "--transient-retries", "0"))
+	if code != ExitRuntimeError || stdout != "" || !strings.Contains(stderr, "execution deadline exceeded") {
+		t.Fatalf("deadline outcome: code=%d stdout=%q stderr=%q", code, stdout, stderr)
+	}
+	select {
+	case <-started:
+	default:
+		t.Fatal("provider call did not start before deadline")
+	}
+}
+
+func TestRunCommandWritesSeparateExecutionReport(t *testing.T) {
+	lib := newCLITestLibrary(t)
+	inputPath := lib.writeInputFile(t, "transcript.md", "hello")
+	var hits int32
+	provider := newTestLLMServer("generated", &hits)
+	defer provider.Close()
+	writePromptFile(t, lib.promptDir, "reported", "local-default")
+	writeProfileFile(t, lib.profileDir, "local-default", provider.URL+"/v1", "local-model")
+	path := filepath.Join(t.TempDir(), "execution.json")
+	code, stdout, stderr := runCLICommand(t, runCommand, []string{
+		"--prompt-dir", lib.promptDir, "--profile-dir", lib.profileDir,
+		"--prompt", "reported", "--input", "transcript=" + inputPath,
+		"--execution-report", path,
+	})
+	if code != ExitOK || stdout != "generated" || atomic.LoadInt32(&hits) != 1 {
+		t.Fatalf("run with report: code=%d stdout=%q stderr=%q hits=%d", code, stdout, stderr, atomic.LoadInt32(&hits))
+	}
+	data, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !bytes.HasSuffix(data, []byte("\n")) {
+		t.Fatalf("report is not newline-terminated: %q", data)
+	}
+	var report map[string]any
+	if err := json.Unmarshal(data, &report); err != nil {
+		t.Fatal(err)
+	}
+	if report["prompt_id"] != "reported" || report["outcome"] != "success" || report["attempts"] != float64(1) {
+		t.Fatalf("unexpected execution report: %#v", report)
+	}
+}
+
+func TestExecutionReportFailureAndOutputHandling(t *testing.T) {
+	terminal := func(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+		if req.Observer != nil {
+			req.Observer(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, PromptID: "p", Outcome: promptkit.ExecutionOutcomeSuccess, Attempts: 1})
+		}
+		return &promptkit.RunResult{Artifact: promptkit.Artifact{Body: []byte("artifact")}, Validation: promptkit.ValidationResult{Status: promptkit.ValidationPassed}}, nil
+	}
+	for _, tc := range []struct {
+		name           string
+		modify         func(*runConfig, string)
+		run            func(context.Context, promptkit.RunRequest) (*promptkit.RunResult, error)
+		wantCode       int
+		wantOutput     string
+		wantDiagnostic string
+		wantReport     bool
+	}{
+		{name: "default streams", run: terminal, wantCode: ExitOK, wantOutput: "artifact"},
+		{name: "validation failure", run: func(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+			req.Observer(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, Outcome: promptkit.ExecutionOutcomeInvalidOutput})
+			return &promptkit.RunResult{Artifact: promptkit.Artifact{Body: []byte("invalid")}, Validation: promptkit.ValidationResult{Status: promptkit.ValidationFailed}}, nil
+		}, wantCode: ExitValidationFailed, wantOutput: "invalid", wantReport: true},
+		{name: "execution failure", run: func(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+			req.Observer(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, Outcome: promptkit.ExecutionOutcomeError, Failure: promptkit.ExecutionFailureProviderStatus})
+			return nil, promptkit.ErrLLMGenerate
+		}, wantCode: ExitRuntimeError, wantDiagnostic: "run error", wantReport: true},
+		{name: "artifact write failure still writes report", run: terminal, modify: func(cfg *runConfig, root string) {
+			cfg.outputPath = filepath.Join(root, "missing", "artifact.txt")
+		}, wantCode: ExitRuntimeError, wantDiagnostic: "output write error", wantReport: true},
+		{name: "report write failure preserves artifact", run: terminal, modify: func(cfg *runConfig, root string) {
+			cfg.executionReportPath = filepath.Join(root, "missing", "execution.json")
+		}, wantCode: ExitRuntimeError, wantOutput: "artifact", wantDiagnostic: "execution report write error"},
+		{name: "execution and report failures are both visible", run: func(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+			req.Observer(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, Outcome: promptkit.ExecutionOutcomeError})
+			return nil, promptkit.ErrLLMGenerate
+		}, modify: func(cfg *runConfig, root string) {
+			cfg.executionReportPath = filepath.Join(root, "missing", "execution.json")
+		}, wantCode: ExitRuntimeError, wantDiagnostic: "execution report write error"},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			root := t.TempDir()
+			cfg := &runConfig{executionReportPath: filepath.Join(root, "execution.json")}
+			if tc.name == "default streams" {
+				cfg.executionReportPath = ""
+			}
+			if tc.modify != nil {
+				tc.modify(cfg, root)
+			}
+			var stdout, stderr bytes.Buffer
+			code := executeRun(cfg, promptkit.RunRequest{PromptID: "p"}, tc.run, &stdout, &stderr)
+			if code != tc.wantCode || stdout.String() != tc.wantOutput || (tc.wantDiagnostic != "" && !strings.Contains(stderr.String(), tc.wantDiagnostic)) {
+				t.Fatalf("unexpected output: code=%d stdout=%q stderr=%q", code, stdout.String(), stderr.String())
+			}
+			if tc.wantReport {
+				if _, err := os.Stat(cfg.executionReportPath); err != nil {
+					t.Fatalf("terminal report missing: %v", err)
+				}
+			}
+			if tc.name == "execution and report failures are both visible" && !strings.Contains(stderr.String(), "run error") {
+				t.Fatalf("execution failure diagnostic was lost: %s", stderr.String())
+			}
+		})
+	}
+}
+
+func TestExecutionReportLeavesExistingFileWhenNoTerminalEvent(t *testing.T) {
+	path := filepath.Join(t.TempDir(), "execution.json")
+	if err := os.WriteFile(path, []byte("old report"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	cfg := &runConfig{executionReportPath: path}
+	var stdout, stderr bytes.Buffer
+	code := executeRun(cfg, promptkit.RunRequest{PromptID: "p"}, func(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+		if req.Observer == nil {
+			t.Fatal("requested report did not attach observer")
+		}
+		return &promptkit.RunResult{Artifact: promptkit.Artifact{Body: []byte("artifact")}, Validation: promptkit.ValidationResult{Status: promptkit.ValidationPassed}}, nil
+	}, &stdout, &stderr)
+	data, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if code != ExitOK || stdout.String() != "artifact" || string(data) != "old report" || !strings.Contains(stderr.String(), "no terminal observation") {
+		t.Fatalf("no-terminal handling: code=%d stdout=%q stderr=%q report=%q", code, stdout.String(), stderr.String(), data)
+	}
+}
+
+func TestExecutionReportPathValidationAndPreInvocationFailure(t *testing.T) {
+	root := t.TempDir()
+	args := []string{"--prompt-dir", root, "--prompt", "p"}
+	path := filepath.Join(root, "execution.json")
+	for _, tail := range [][]string{
+		{"--execution-report", ""},
+		{"--out", path, "--execution-report", path},
+		{"--out", path, "--execution-report", filepath.Join(root, "subdir", "..", "execution.json")},
+	} {
+		if _, err := parseRunArgs(append(append([]string{}, args...), tail...)); err == nil {
+			t.Fatalf("accepted invalid report destination: %q", tail)
+		}
+	}
+	if err := os.WriteFile(path, []byte("old report"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	alias := filepath.Join(root, "alias.json")
+	if err := os.Link(path, alias); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := parseRunArgs(append(append([]string{}, args...), "--out", path, "--execution-report", alias)); err == nil {
+		t.Fatal("accepted two names for the same output file")
+	}
+	if cfg, err := parseRunArgs(append(append([]string{}, args...), "--execution-report", "-")); err != nil || cfg.executionReportPath != "-" {
+		t.Fatalf("ordinary dash path was rejected: cfg=%+v err=%v", cfg, err)
+	}
+	if _, err := parseRenderArgs(append(append([]string{}, args...), "--execution-report", path)); err == nil {
+		t.Fatal("render accepted run-only report flag")
+	}
+	code, stdout, stderr := runCLICommand(t, runCommand, []string{"--config", filepath.Join(root, "missing.yml"), "--prompt", "p", "--execution-report", path})
+	data, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if code != ExitRuntimeError || stdout != "" || !strings.Contains(stderr, "source.missing") || string(data) != "old report" {
+		t.Fatalf("pre-invocation failure changed report: code=%d stdout=%q stderr=%q report=%q", code, stdout, stderr, data)
+	}
+}
+
+func intPointer(value int) *int { return &value }
+
+func TestPromptkitPromptInspectionSelectsVersionsAndContracts(t *testing.T) {
+	fixtureRoot := definitionFixtureRoot(t)
+	configPath := writeDefinitionFixtureConfig(t, fixtureRoot, true)
 	tests := []struct {
 		promptID string
 		version  string
@@ -1296,12 +1736,12 @@ func TestPromptkitV09PromptInspectionSelectsVersionsAndContracts(t *testing.T) {
 	}
 }
 
-func TestPromptkitV09ProfileInspectionResolvesSupportedTargets(t *testing.T) {
+func TestPromptkitProfileInspectionResolvesSupportedTargets(t *testing.T) {
 	const secret = "sentinel-profile-secret"
 	t.Setenv("FIXTURE_PROFILE_API_KEY", secret)
 
-	fixtureRoot := promptkitV09FixtureRoot(t)
-	configPath := writePromptkitV09Config(t, fixtureRoot, true)
+	fixtureRoot := definitionFixtureRoot(t)
+	configPath := writeDefinitionFixtureConfig(t, fixtureRoot, true)
 	tests := []struct {
 		name          string
 		profileID     string
@@ -1311,7 +1751,7 @@ func TestPromptkitV09ProfileInspectionResolvesSupportedTargets(t *testing.T) {
 	}{
 		{name: "inherited custom backend", profileID: "custom-derived", wantBackend: "fixture-custom", wantModel: "fixture-derived-model", wantAPIKeyEnv: "FIXTURE_PROFILE_API_KEY"},
 		{name: "endpoint only", profileID: "endpoint-only", wantModel: "fixture-endpoint-model"},
-		{name: "built in", profileID: "deepseek-4-flash", wantBackend: "openrouter", wantModel: "deepseek/deepseek-v4-flash", wantAPIKeyEnv: "OPENROUTER_API_KEY"},
+		{name: "built in", profileID: "promptkit-openrouter-auto", wantBackend: "openrouter", wantModel: "openrouter/auto", wantAPIKeyEnv: "OPENROUTER_API_KEY"},
 	}
 
 	for _, tc := range tests {
@@ -1340,7 +1780,7 @@ func TestPromptkitV09ProfileInspectionResolvesSupportedTargets(t *testing.T) {
 		})
 	}
 
-	configWithoutPromptDir := writePromptkitV09Config(t, fixtureRoot, false)
+	configWithoutPromptDir := writeDefinitionFixtureConfig(t, fixtureRoot, false)
 	code, _, stderr := runCLICommand(t, inspectCommand, []string{
 		"profile",
 		"--config", configWithoutPromptDir,
@@ -1352,9 +1792,92 @@ func TestPromptkitV09ProfileInspectionResolvesSupportedTargets(t *testing.T) {
 	}
 }
 
+func TestMaintainedProfilesResolveAcrossSelectionAndSourcePrecedence(t *testing.T) {
+	promptDir := t.TempDir()
+	profileDir := t.TempDir()
+	writePromptDefinition(t, promptDir, "maintained.yaml", `id: maintained
+version: "1.0.0"
+default_profile: promptkit-openrouter-standard
+messages:
+  - role: user
+    content: Hello
+output:
+  format: text
+  validation_mode: none
+`)
+	configPath := writeAppConfigFile(t, fmt.Sprintf("prompt_dir: %q\nprofile_dir: %q\n", promptDir, profileDir))
+
+	for _, profileID := range []string{"promptkit-openrouter-standard", "promptkit-openrouter-auto"} {
+		t.Run(profileID+" inspection", func(t *testing.T) {
+			code, stdout, stderr := runCLICommand(t, inspectCommand, []string{"profile", "--config", configPath, "--profile", profileID, "--format", "json"})
+			if code != ExitOK {
+				t.Fatalf("inspect %s: code=%d stderr=%q", profileID, code, stderr)
+			}
+			var inspection renderformat.ProfileInspection
+			if err := json.Unmarshal([]byte(stdout), &inspection); err != nil {
+				t.Fatal(err)
+			}
+			if inspection.ProfileID != profileID || inspection.EffectiveModelParams.BackendID != "openrouter" || inspection.EffectiveModelParams.Model == "" {
+				t.Fatalf("maintained profile did not resolve: %+v", inspection)
+			}
+			if profileID == "promptkit-openrouter-auto" && inspection.EffectiveModelParams.Model != "openrouter/auto" {
+				t.Fatalf("unexpected auto target: %+v", inspection.EffectiveModelParams)
+			}
+		})
+	}
+
+	for _, tc := range []struct {
+		name      string
+		selection []string
+		wantID    string
+	}{
+		{name: "prompt default", wantID: "promptkit-openrouter-standard"},
+		{name: "explicit selection", selection: []string{"--profile", "promptkit-openrouter-auto"}, wantID: "promptkit-openrouter-auto"},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			args := append([]string{"--config", configPath, "--prompt", "maintained", "--format", "json"}, tc.selection...)
+			code, stdout, stderr := runCLICommand(t, renderCommand, args)
+			if code != ExitOK {
+				t.Fatalf("render: code=%d stderr=%q", code, stderr)
+			}
+			var rendered struct {
+				SelectedProfileID string `json:"selected_profile_id"`
+				SelectedBackendID string `json:"selected_backend_id"`
+			}
+			if err := json.Unmarshal([]byte(stdout), &rendered); err != nil {
+				t.Fatal(err)
+			}
+			if rendered.SelectedProfileID != tc.wantID || rendered.SelectedBackendID != "openrouter" {
+				t.Fatalf("unexpected selection: %+v", rendered)
+			}
+		})
+	}
+
+	writePromptDefinition(t, profileDir, "derived.yaml", "id: derived-maintained\nbase_profile: promptkit-openrouter-standard\nmodel: fixture-inherited-model\n")
+	writePromptDefinition(t, profileDir, "replacement.yaml", "id: promptkit-openrouter-auto\nbackend: openrouter\nmodel: fixture-override-model\n")
+	for _, tc := range []struct{ id, model string }{
+		{id: "derived-maintained", model: "fixture-inherited-model"},
+		{id: "promptkit-openrouter-auto", model: "fixture-override-model"},
+	} {
+		t.Run(tc.id+" source", func(t *testing.T) {
+			code, stdout, stderr := runCLICommand(t, inspectCommand, []string{"profile", "--config", configPath, "--profile", tc.id, "--format", "json"})
+			if code != ExitOK {
+				t.Fatalf("inspect: code=%d stderr=%q", code, stderr)
+			}
+			var inspection renderformat.ProfileInspection
+			if err := json.Unmarshal([]byte(stdout), &inspection); err != nil {
+				t.Fatal(err)
+			}
+			if inspection.EffectiveModelParams.BackendID != "openrouter" || inspection.EffectiveModelParams.Model != tc.model {
+				t.Fatalf("unexpected effective target: %+v", inspection)
+			}
+		})
+	}
+}
+
 func TestProfileInspectionHonorsDirectoryPrecedenceOutputAndFailures(t *testing.T) {
-	fixtureRoot := promptkitV09FixtureRoot(t)
-	configPath := writePromptkitV09Config(t, fixtureRoot, true)
+	fixtureRoot := definitionFixtureRoot(t)
+	configPath := writeDefinitionFixtureConfig(t, fixtureRoot, true)
 	overrideDir := t.TempDir()
 	writeProfileFile(t, overrideDir, "custom-derived", "http://127.0.0.1:9000/v1", "override-model")
 	outPath := filepath.Join(t.TempDir(), "inspection.json")
@@ -1971,7 +2494,7 @@ func runCLICommand(t *testing.T, command func([]string, io.Writer, io.Writer) in
 	return code, stdout.String(), stderr.String()
 }
 
-func promptkitV09FixtureRoot(t *testing.T) string {
+func definitionFixtureRoot(t *testing.T) string {
 	t.Helper()
 	root, err := filepath.Abs(filepath.Join("..", "..", "..", "testdata", "promptkit-v0.9"))
 	if err != nil {
@@ -1983,7 +2506,7 @@ func promptkitV09FixtureRoot(t *testing.T) string {
 	return root
 }
 
-func writePromptkitV09Config(t *testing.T, fixtureRoot string, includeSources bool) string {
+func writeDefinitionFixtureConfig(t *testing.T, fixtureRoot string, includeSources bool) string {
 	t.Helper()
 	sources := ""
 	if includeSources {

internal/adapter/http/dto.go

@@ -2,17 +2,22 @@ package httpadapter
 
 import (
 	"time"
+
+	executionreport "gitea.maximumdirect.net/eric/scriptorium/internal/report"
 )
 
 type runRequestDTO struct {
-	PromptID         string                   `json:"prompt_id"`
-	PromptVersion    string                   `json:"prompt_version,omitempty"`
-	ProfileID        string                   `json:"profile_id,omitempty"`
-	SessionID        string                   `json:"session_id,omitempty"`
-	Inputs           map[string]inputRefDTO   `json:"inputs"`
-	Vars             map[string]string        `json:"vars,omitempty"`
-	Model            *modelOverrideRequestDTO `json:"model,omitempty"`
-	IncludeRawOutput bool                     `json:"include_raw_output,omitempty"`
+	PromptID               string                   `json:"prompt_id"`
+	PromptVersion          string                   `json:"prompt_version,omitempty"`
+	ProfileID              string                   `json:"profile_id,omitempty"`
+	SessionID              string                   `json:"session_id,omitempty"`
+	Inputs                 map[string]inputRefDTO   `json:"inputs"`
+	Vars                   map[string]string        `json:"vars,omitempty"`
+	Model                  *modelOverrideRequestDTO `json:"model,omitempty"`
+	IncludeRawOutput       bool                     `json:"include_raw_output,omitempty"`
+	TransientRetries       *int                     `json:"transient_retries,omitempty"`
+	RunTimeoutSeconds      *int64                   `json:"run_timeout_seconds,omitempty"`
+	IncludeExecutionReport bool                     `json:"include_execution_report,omitempty"`
 }
 
 type inputRefDTO struct {
@@ -35,10 +40,11 @@ type modelOverrideRequestDTO struct {
 }
 
 type runResponseDTO struct {
-	Artifact       artifactDTO   `json:"artifact"`
-	Validation     validationDTO `json:"validation"`
-	Metadata       metadataDTO   `json:"metadata"`
-	RawModelOutput *string       `json:"raw_model_output,omitempty"`
+	Artifact       artifactDTO                `json:"artifact"`
+	Validation     validationDTO              `json:"validation"`
+	Metadata       metadataDTO                `json:"metadata"`
+	RawModelOutput *string                    `json:"raw_model_output,omitempty"`
+	Execution      *executionreport.Execution `json:"execution,omitempty"`
 }
 
 type artifactDTO struct {
@@ -104,7 +110,8 @@ type validationDTO struct {
 }
 
 type errorResponse struct {
-	Error errorBody `json:"error"`
+	Error     errorBody                  `json:"error"`
+	Execution *executionreport.Execution `json:"execution,omitempty"`
 }
 
 type errorBody struct {

internal/adapter/http/handler.go

@@ -5,11 +5,14 @@ import (
 	"encoding/json"
 	"errors"
 	"io"
+	"math"
 	"net/http"
 	"strings"
+	"time"
 
 	"gitea.maximumdirect.net/eric/promptkit"
 	"gitea.maximumdirect.net/eric/scriptorium/internal/defaults"
+	executionreport "gitea.maximumdirect.net/eric/scriptorium/internal/report"
 )
 
 type Runner interface {
@@ -76,6 +79,16 @@ func (h *Handler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
 		writeError(w, http.StatusBadRequest, "invalid_request", "prompt_id is required")
 		return
 	}
+	if req.TransientRetries != nil && *req.TransientRetries < 0 {
+		writeError(w, http.StatusBadRequest, "invalid_request", "transient_retries must be greater than or equal to 0")
+		return
+	}
+	if req.RunTimeoutSeconds != nil {
+		if *req.RunTimeoutSeconds < 0 || *req.RunTimeoutSeconds > math.MaxInt64/int64(time.Second) {
+			writeError(w, http.StatusBadRequest, "invalid_request", "run_timeout_seconds is out of range")
+			return
+		}
+	}
 	var mappedInputs map[string]promptkit.ArtifactRef
 	if len(req.Inputs) > 0 {
 		mappedInputs = make(map[string]promptkit.ArtifactRef, len(req.Inputs))
@@ -93,18 +106,43 @@ func (h *Handler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
 		model = executionTargetOverrideFromModelOverrideDTO(req.Model)
 	}
 
-	res, err := h.runner.Run(r.Context(), promptkit.RunRequest{
-		PromptID:      req.PromptID,
-		PromptVersion: req.PromptVersion,
-		ProfileID:     req.ProfileID,
-		SessionID:     req.SessionID,
-		Inputs:        mappedInputs,
-		Vars:          req.Vars,
-		Execution:     model,
-	})
+	runContext := r.Context()
+	if req.RunTimeoutSeconds != nil && *req.RunTimeoutSeconds > 0 {
+		var cancel context.CancelFunc
+		runContext, cancel = context.WithTimeout(runContext, time.Duration(*req.RunTimeoutSeconds)*time.Second)
+		defer cancel()
+	}
+	mappedRequest := promptkit.RunRequest{
+		PromptID:         req.PromptID,
+		PromptVersion:    req.PromptVersion,
+		ProfileID:        req.ProfileID,
+		SessionID:        req.SessionID,
+		Inputs:           mappedInputs,
+		Vars:             req.Vars,
+		Execution:        model,
+		TransientRetries: req.TransientRetries,
+	}
+	var collector *executionreport.Collector
+	if req.IncludeExecutionReport {
+		collector = executionreport.NewCollector()
+		mappedRequest.Observer = collector.Observe
+	}
+	res, err := h.runner.Run(runContext, mappedRequest)
+	var execution *executionreport.Execution
+	if collector != nil {
+		execution = collector.Snapshot()
+	}
 	if err != nil {
+		switch runContext.Err() {
+		case context.DeadlineExceeded:
+			writeRunFailure(w, http.StatusGatewayTimeout, "deadline_exceeded", "run execution deadline exceeded", execution, h.options.MaxResponseBytes)
+			return
+		case context.Canceled:
+			writeRunFailure(w, http.StatusRequestTimeout, "request_canceled", "run request canceled", execution, h.options.MaxResponseBytes)
+			return
+		}
 		status, code, message := mapRunError(err)
-		writeError(w, status, code, message)
+		writeRunFailure(w, status, code, message, execution, h.options.MaxResponseBytes)
 		return
 	}
 
@@ -145,6 +183,7 @@ func (h *Handler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
 			ValidationStatus:   string(res.Validation.Status),
 			RepairAttemptsUsed: res.Validation.RepairAttempts,
 		},
+		Execution: execution,
 	}
 	if req.IncludeRawOutput {
 		raw := res.RawOutput
@@ -153,6 +192,14 @@ func (h *Handler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
 	writeLimitedJSON(w, http.StatusOK, resp, h.options.MaxResponseBytes)
 }
 
+func writeRunFailure(w http.ResponseWriter, status int, code, message string, execution *executionreport.Execution, maxBytes int64) {
+	if execution == nil {
+		writeError(w, status, code, message)
+		return
+	}
+	writeLimitedJSON(w, status, errorResponse{Error: errorBody{Code: code, Message: message}, Execution: execution}, maxBytes)
+}
+
 func executionTargetOverrideFromModelOverrideDTO(dto *modelOverrideRequestDTO) *promptkit.ExecutionTargetOverride {
 	if dto == nil {
 		return nil
@@ -224,7 +271,7 @@ func mapRunError(err error) (int, string, string) {
 		return http.StatusBadRequest, "prompt_render_failed", "failed to render prompt"
 	case errors.Is(err, promptkit.ErrCapacityExceeded):
 		return http.StatusServiceUnavailable, "capacity_exceeded", "model backend capacity is exhausted"
-	case errors.Is(err, promptkit.ErrLLMGenerate):
+	case errors.Is(err, promptkit.ErrLLMGenerate), errors.Is(err, promptkit.ErrTransport), errors.Is(err, promptkit.ErrMalformedResponse), errors.Is(err, promptkit.ErrRetriesExhausted):
 		return http.StatusBadGateway, "llm_failed", "model generation request failed"
 	case errors.Is(err, promptkit.ErrValidation):
 		return http.StatusInternalServerError, "validation_runtime_failed", "validation runtime failed"

internal/adapter/http/handler_test.go

@@ -24,6 +24,12 @@ type fakeRunner struct {
 	last   promptkit.RunRequest
 }
 
+type runnerFunc func(context.Context, promptkit.RunRequest) (*promptkit.RunResult, error)
+
+func (run runnerFunc) Run(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+	return run(ctx, req)
+}
+
 func (f *fakeRunner) Run(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
 	f.last = req
 	if f.err != nil {
@@ -32,6 +38,208 @@ func (f *fakeRunner) Run(ctx context.Context, req promptkit.RunRequest) (*prompt
 	return f.result, nil
 }
 
+func TestHandlerExecutionReportOptInAndResultShapes(t *testing.T) {
+	for _, tc := range []struct {
+		name, body string
+		wantReport bool
+		invalid    bool
+	}{
+		{"omitted", `{"prompt_id":"p"}`, false, false},
+		{"null", `{"prompt_id":"p","include_execution_report":null}`, false, false},
+		{"false", `{"prompt_id":"p","include_execution_report":false}`, false, false},
+		{"success", `{"prompt_id":"p","include_execution_report":true}`, true, false},
+		{"invalid output", `{"prompt_id":"p","include_execution_report":true}`, true, true},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			runner := runnerFunc(func(_ context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+				if (req.Observer != nil) != tc.wantReport {
+					t.Fatalf("unexpected observer presence: %v", req.Observer != nil)
+				}
+				status := promptkit.ValidationPassed
+				outcome := promptkit.ExecutionOutcomeSuccess
+				if tc.invalid {
+					status = promptkit.ValidationFailed
+					outcome = promptkit.ExecutionOutcomeInvalidOutput
+				}
+				if req.Observer != nil {
+					req.Observer(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, RunID: "observed", Outcome: outcome, Failure: promptkit.ExecutionFailureNone, Attempts: 2})
+				}
+				return &promptkit.RunResult{Artifact: promptkit.Artifact{Body: []byte("body")}, Validation: promptkit.ValidationResult{Status: status}, Usage: promptkit.TokenUsage{PromptTokens: 7}}, nil
+			})
+			w := httptest.NewRecorder()
+			NewHandler(runner).ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(tc.body)))
+			if w.Code != http.StatusOK {
+				t.Fatalf("status=%d body=%s", w.Code, w.Body.String())
+			}
+			var response map[string]any
+			if err := json.Unmarshal(w.Body.Bytes(), &response); err != nil {
+				t.Fatal(err)
+			}
+			if (response["execution"] != nil) != tc.wantReport {
+				t.Fatalf("unexpected report: %#v", response["execution"])
+			}
+			if tc.wantReport {
+				report := response["execution"].(map[string]any)
+				wantOutcome := "success"
+				if tc.invalid {
+					wantOutcome = "invalid_output"
+				}
+				if report["outcome"] != wantOutcome || report["attempts"] != float64(2) {
+					t.Fatalf("unexpected report: %#v", report)
+				}
+			}
+			if response["metadata"].(map[string]any)["usage"].(map[string]any)["prompt_tokens"] != float64(7) {
+				t.Fatalf("metadata usage changed: %#v", response["metadata"])
+			}
+		})
+	}
+
+	w := httptest.NewRecorder()
+	NewHandler(&fakeRunner{}).ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","include_execution_report":"true"}`)))
+	assertHTTPErrorCode(t, w, http.StatusBadRequest, "invalid_json")
+}
+
+func TestHandlerExecutionReportFailureAndAbsence(t *testing.T) {
+	secret := "private provider diagnostic"
+	runner := runnerFunc(func(_ context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+		req.Observer(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, Outcome: promptkit.ExecutionOutcomeError, Failure: promptkit.ExecutionFailureProviderStatus, StatusCode: 503, RetriesExhausted: true, Attempts: 3})
+		return nil, fmt.Errorf("%w: %s", promptkit.ErrLLMGenerate, secret)
+	})
+	w := httptest.NewRecorder()
+	NewHandler(runner).ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","include_execution_report":true}`)))
+	assertHTTPErrorCode(t, w, http.StatusBadGateway, "llm_failed")
+	var response map[string]any
+	if err := json.Unmarshal(w.Body.Bytes(), &response); err != nil {
+		t.Fatal(err)
+	}
+	report := response["execution"].(map[string]any)
+	if report["failure"] != "provider_status" || report["retries_exhausted"] != true || report["attempts"] != float64(3) {
+		t.Fatalf("unexpected failure report: %#v", report)
+	}
+	if strings.Contains(w.Body.String(), secret) {
+		t.Fatalf("response leaked provider diagnostic: %s", w.Body.String())
+	}
+
+	noTerminal := runnerFunc(func(_ context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+		if req.Observer == nil {
+			t.Fatal("missing opted-in observer")
+		}
+		return nil, promptkit.ErrPromptLoad
+	})
+	w = httptest.NewRecorder()
+	NewHandler(noTerminal).ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","include_execution_report":true}`)))
+	assertHTTPErrorCode(t, w, http.StatusBadRequest, "prompt_load_failed")
+	if strings.Contains(w.Body.String(), "execution") {
+		t.Fatalf("unexpected absent report: %s", w.Body.String())
+	}
+
+	w = httptest.NewRecorder()
+	NewHandler(noTerminal).ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"include_execution_report":true}`)))
+	assertHTTPErrorCode(t, w, http.StatusBadRequest, "invalid_request")
+	if strings.Contains(w.Body.String(), "execution") {
+		t.Fatalf("pre-invocation failure has report: %s", w.Body.String())
+	}
+}
+
+func TestHandlerExecutionReportResponseLimit(t *testing.T) {
+	for _, failure := range []bool{false, true} {
+		runner := runnerFunc(func(_ context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+			if req.Observer != nil {
+				req.Observer(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, Outcome: promptkit.ExecutionOutcomeError, Failure: promptkit.ExecutionFailureGeneration})
+			}
+			if failure {
+				return nil, promptkit.ErrLLMGenerate
+			}
+			return &promptkit.RunResult{}, nil
+		})
+		body := `{"prompt_id":"p"}`
+		base := httptest.NewRecorder()
+		NewHandler(runner).ServeHTTP(base, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(body)))
+		body = `{"prompt_id":"p","include_execution_report":true}`
+		limited := httptest.NewRecorder()
+		NewHandlerWithOptions(runner, HandlerOptions{MaxResponseBytes: int64(base.Body.Len())}).ServeHTTP(limited, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(body)))
+		assertHTTPErrorCode(t, limited, http.StatusRequestEntityTooLarge, "response_too_large")
+		if strings.Contains(limited.Body.String(), "execution") {
+			t.Fatalf("overflow fallback included report: %s", limited.Body.String())
+		}
+	}
+}
+
+func TestHandlerRealEngineFailureIncludesTerminalReport(t *testing.T) {
+	provider := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		http.Error(w, "private provider detail", http.StatusServiceUnavailable)
+	}))
+	defer provider.Close()
+	h := newRetryHandler(t, provider.URL+"/v1", intPointer(0), 0, nil)
+	w := httptest.NewRecorder()
+	h.ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","include_execution_report":true}`)))
+	assertHTTPErrorCode(t, w, http.StatusBadGateway, "llm_failed")
+	var response map[string]any
+	if err := json.Unmarshal(w.Body.Bytes(), &response); err != nil {
+		t.Fatal(err)
+	}
+	report := response["execution"].(map[string]any)
+	if report["outcome"] != "error" || report["failure"] != "provider_status" || report["attempts"] != float64(1) {
+		t.Fatalf("unexpected terminal report: %#v", report)
+	}
+	if strings.Contains(w.Body.String(), "private provider detail") {
+		t.Fatalf("provider diagnostic leaked: %s", w.Body.String())
+	}
+}
+
+func TestHandlerConcurrentExecutionReportsAreIsolated(t *testing.T) {
+	started := make(chan struct{}, 2)
+	release := make(chan struct{})
+	defer func() {
+		select {
+		case <-release:
+		default:
+			close(release)
+		}
+	}()
+	runner := runnerFunc(func(_ context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+		started <- struct{}{}
+		<-release
+		req.Observer(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, RunID: req.SessionID, Outcome: promptkit.ExecutionOutcomeSuccess, Attempts: uint64(len(req.SessionID))})
+		return &promptkit.RunResult{}, nil
+	})
+	h := NewHandler(runner)
+	responses := make([]*httptest.ResponseRecorder, 2)
+	done := make(chan struct{}, 2)
+	for i, id := range []string{"one", "second"} {
+		responses[i] = httptest.NewRecorder()
+		go func(w *httptest.ResponseRecorder, session string) {
+			h.ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(fmt.Sprintf(`{"prompt_id":"p","session_id":%q,"include_execution_report":true}`, session))))
+			done <- struct{}{}
+		}(responses[i], id)
+	}
+	for range 2 {
+		select {
+		case <-started:
+		case <-time.After(3 * time.Second):
+			t.Fatal("concurrent runs did not enter runner")
+		}
+	}
+	close(release)
+	for range 2 {
+		select {
+		case <-done:
+		case <-time.After(3 * time.Second):
+			t.Fatal("concurrent run did not complete")
+		}
+	}
+	for i, id := range []string{"one", "second"} {
+		var response map[string]any
+		if err := json.Unmarshal(responses[i].Body.Bytes(), &response); err != nil {
+			t.Fatal(err)
+		}
+		report := response["execution"].(map[string]any)
+		if report["run_id"] != id || report["attempts"] != float64(len(id)) {
+			t.Fatalf("report crossed requests: %#v", report)
+		}
+	}
+}
+
 func TestMaintainedHTTPRunExampleMatchesRequestContract(t *testing.T) {
 	body, err := os.ReadFile(filepath.Join("..", "..", "..", "examples", "http-run.json"))
 	if err != nil {
@@ -64,6 +272,129 @@ func TestMaintainedHTTPRunExampleMatchesRequestContract(t *testing.T) {
 	assertHTTPErrorCode(t, invalidW, http.StatusBadRequest, "invalid_json")
 }
 
+func TestHandlerTransientRetryOverridePresenceAndStrictness(t *testing.T) {
+	for _, tc := range []struct {
+		name   string
+		body   string
+		want   *int
+		status int
+		code   string
+	}{
+		{name: "omitted", body: `{"prompt_id":"p"}`},
+		{name: "null", body: `{"prompt_id":"p","transient_retries":null}`},
+		{name: "zero", body: `{"prompt_id":"p","transient_retries":0}`, want: intPointer(0)},
+		{name: "positive", body: `{"prompt_id":"p","transient_retries":3}`, want: intPointer(3)},
+		{name: "negative", body: `{"prompt_id":"p","transient_retries":-1}`, status: http.StatusBadRequest, code: "invalid_request"},
+		{name: "string", body: `{"prompt_id":"p","transient_retries":"2"}`, status: http.StatusBadRequest, code: "invalid_json"},
+		{name: "fractional", body: `{"prompt_id":"p","transient_retries":1.5}`, status: http.StatusBadRequest, code: "invalid_json"},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			runner := &fakeRunner{result: &promptkit.RunResult{}}
+			w := httptest.NewRecorder()
+			NewHandler(runner).ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(tc.body)))
+			if tc.status != 0 {
+				assertHTTPErrorCode(t, w, tc.status, tc.code)
+				if runner.last.PromptID != "" {
+					t.Fatalf("runner invoked for invalid request: %+v", runner.last)
+				}
+				return
+			}
+			if w.Code != http.StatusOK || !reflect.DeepEqual(runner.last.TransientRetries, tc.want) {
+				t.Fatalf("retry mapping: status=%d got=%v want=%v body=%s", w.Code, runner.last.TransientRetries, tc.want, w.Body.String())
+			}
+		})
+	}
+}
+
+func TestHandlerRunTimeoutValidationAndContextMapping(t *testing.T) {
+	for _, tc := range []struct {
+		name, body, code string
+		status           int
+	}{
+		{name: "negative", body: `{"prompt_id":"p","run_timeout_seconds":-1}`, status: http.StatusBadRequest, code: "invalid_request"},
+		{name: "duration overflow", body: `{"prompt_id":"p","run_timeout_seconds":9223372037}`, status: http.StatusBadRequest, code: "invalid_request"},
+		{name: "string", body: `{"prompt_id":"p","run_timeout_seconds":"1"}`, status: http.StatusBadRequest, code: "invalid_json"},
+		{name: "fractional", body: `{"prompt_id":"p","run_timeout_seconds":1.5}`, status: http.StatusBadRequest, code: "invalid_json"},
+		{name: "int64 overflow", body: `{"prompt_id":"p","run_timeout_seconds":9223372036854775808}`, status: http.StatusBadRequest, code: "invalid_json"},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			runner := &fakeRunner{result: &promptkit.RunResult{}}
+			w := httptest.NewRecorder()
+			NewHandler(runner).ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(tc.body)))
+			assertHTTPErrorCode(t, w, tc.status, tc.code)
+			if runner.last.PromptID != "" {
+				t.Fatalf("runner invoked for invalid timeout: %+v", runner.last)
+			}
+		})
+	}
+
+	parent, cancel := context.WithDeadline(context.Background(), time.Now().Add(5*time.Second))
+	defer cancel()
+	parentDeadline, _ := parent.Deadline()
+	for _, tc := range []struct {
+		name, body   string
+		parent       context.Context
+		wantDeadline time.Time
+	}{
+		{name: "omitted inherits deadline", body: `{"prompt_id":"p"}`, parent: parent, wantDeadline: parentDeadline},
+		{name: "null inherits deadline", body: `{"prompt_id":"p","run_timeout_seconds":null}`, parent: parent, wantDeadline: parentDeadline},
+		{name: "zero inherits deadline", body: `{"prompt_id":"p","run_timeout_seconds":0}`, parent: parent, wantDeadline: parentDeadline},
+		{name: "positive preserves earlier deadline", body: `{"prompt_id":"p","run_timeout_seconds":10}`, parent: parent, wantDeadline: parentDeadline},
+		{name: "omitted adds no deadline", body: `{"prompt_id":"p"}`, parent: context.Background()},
+		{name: "null adds no deadline", body: `{"prompt_id":"p","run_timeout_seconds":null}`, parent: context.Background()},
+		{name: "zero adds no deadline", body: `{"prompt_id":"p","run_timeout_seconds":0}`, parent: context.Background()},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			called := false
+			h := NewHandler(runnerFunc(func(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+				called = true
+				deadline, ok := ctx.Deadline()
+				if ok != !tc.wantDeadline.IsZero() || (ok && !deadline.Equal(tc.wantDeadline)) {
+					t.Fatalf("execution deadline: got %v (present=%t), want %v", deadline, ok, tc.wantDeadline)
+				}
+				return &promptkit.RunResult{}, nil
+			}))
+			w := httptest.NewRecorder()
+			h.ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(tc.body)).WithContext(tc.parent))
+			if !called || w.Code != http.StatusOK {
+				t.Fatalf("context mapping: called=%t status=%d body=%s", called, w.Code, w.Body.String())
+			}
+		})
+	}
+
+	canceled, cancelParent := context.WithCancel(context.Background())
+	cancelParent()
+	h := NewHandler(runnerFunc(func(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+		if ctx.Err() != context.Canceled {
+			t.Fatalf("parent cancellation did not reach execution: %v", ctx.Err())
+		}
+		return nil, ctx.Err()
+	}))
+	w := httptest.NewRecorder()
+	h.ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","run_timeout_seconds":10}`)).WithContext(canceled))
+	assertHTTPErrorCode(t, w, http.StatusRequestTimeout, "request_canceled")
+
+	h = NewHandler(runnerFunc(func(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+		return nil, fmt.Errorf("attempt: %w: %w", promptkit.ErrLLMGenerate, context.DeadlineExceeded)
+	}))
+	w = httptest.NewRecorder()
+	h.ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","run_timeout_seconds":10}`)))
+	assertHTTPErrorCode(t, w, http.StatusBadGateway, "llm_failed")
+
+	successContext, cancelSuccess := context.WithCancel(context.Background())
+	h = NewHandler(runnerFunc(func(ctx context.Context, req promptkit.RunRequest) (*promptkit.RunResult, error) {
+		cancelSuccess()
+		return &promptkit.RunResult{}, nil
+	}))
+	w = httptest.NewRecorder()
+	h.ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","run_timeout_seconds":10}`)).WithContext(successContext))
+	if w.Code != http.StatusOK {
+		t.Fatalf("successful result was reclassified after cancellation: %d %s", w.Code, w.Body.String())
+	}
+}
+
+func intPointer(value int) *int { return &value }
+
 type handlerLLMClient struct{}
 
 func (handlerLLMClient) Generate(ctx context.Context, req promptkit.GenerateRequest) (*promptkit.GenerateResponse, error) {
@@ -774,6 +1105,179 @@ model: test
 	}
 }
 
+func TestHandlerBuiltInTransientRetriesInheritAndAllowExplicitZero(t *testing.T) {
+	var hits int32
+	provider := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if atomic.AddInt32(&hits, 1) == 1 {
+			http.Error(w, "private provider detail", http.StatusServiceUnavailable)
+			return
+		}
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = w.Write([]byte(`{"choices":[{"message":{"role":"assistant","content":"recovered"}}]}`))
+	}))
+	defer provider.Close()
+	h := newRetryHandler(t, provider.URL+"/v1", intPointer(1), 0, nil)
+
+	w := httptest.NewRecorder()
+	h.ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p"}`)))
+	if w.Code != http.StatusOK || atomic.LoadInt32(&hits) != 2 {
+		t.Fatalf("inherited retry: status=%d hits=%d body=%s", w.Code, atomic.LoadInt32(&hits), w.Body.String())
+	}
+	atomic.StoreInt32(&hits, 0)
+	w = httptest.NewRecorder()
+	h.ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","transient_retries":0}`)))
+	assertHTTPErrorCode(t, w, http.StatusBadGateway, "llm_failed")
+	if atomic.LoadInt32(&hits) != 1 || strings.Contains(w.Body.String(), "private provider detail") {
+		t.Fatalf("explicit zero or safe error failed: hits=%d body=%s", atomic.LoadInt32(&hits), w.Body.String())
+	}
+}
+
+func TestHandlerRetainsSharedBackendLimitDuringRetry(t *testing.T) {
+	started := make(chan struct{}, 1)
+	release := make(chan struct{})
+	var hits, active, peak int32
+	provider := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		current := atomic.AddInt32(&active, 1)
+		defer atomic.AddInt32(&active, -1)
+		for {
+			old := atomic.LoadInt32(&peak)
+			if current <= old || atomic.CompareAndSwapInt32(&peak, old, current) {
+				break
+			}
+		}
+		if atomic.AddInt32(&hits, 1) == 1 {
+			started <- struct{}{}
+			<-release
+			http.Error(w, "temporary", http.StatusServiceUnavailable)
+			return
+		}
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = w.Write([]byte(`{"choices":[{"message":{"role":"assistant","content":"ok"}}]}`))
+	}))
+	defer provider.Close()
+	defer func() {
+		select {
+		case <-release:
+		default:
+			close(release)
+		}
+	}()
+	queueCapacity := 0
+	h := newRetryHandler(t, provider.URL+"/v1", intPointer(1), 1, &queueCapacity)
+	first := httptest.NewRecorder()
+	done := make(chan struct{})
+	go func() {
+		defer close(done)
+		h.ServeHTTP(first, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p"}`)))
+	}()
+	select {
+	case <-started:
+	case <-time.After(3 * time.Second):
+		t.Fatal("first provider call did not start")
+	}
+	second := httptest.NewRecorder()
+	h.ServeHTTP(second, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","transient_retries":0}`)))
+	assertHTTPErrorCode(t, second, http.StatusServiceUnavailable, "capacity_exceeded")
+	close(release)
+	select {
+	case <-done:
+	case <-time.After(5 * time.Second):
+		t.Fatal("retrying request did not complete")
+	}
+	if first.Code != http.StatusOK || atomic.LoadInt32(&hits) != 2 || atomic.LoadInt32(&peak) > 1 {
+		t.Fatalf("shared backend limit failed: first=%d hits=%d peak=%d body=%s", first.Code, atomic.LoadInt32(&hits), atomic.LoadInt32(&peak), first.Body.String())
+	}
+}
+
+func TestHandlerDeadlineCancelsBuiltInRetryWait(t *testing.T) {
+	var hits int32
+	provider := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		atomic.AddInt32(&hits, 1)
+		w.Header().Set("Retry-After", "3600")
+		http.Error(w, "temporary", http.StatusServiceUnavailable)
+	}))
+	defer provider.Close()
+	h := newRetryHandler(t, provider.URL+"/v1", intPointer(1), 0, nil)
+	w := httptest.NewRecorder()
+	h.ServeHTTP(w, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","run_timeout_seconds":1}`)))
+	assertHTTPErrorCode(t, w, http.StatusGatewayTimeout, "deadline_exceeded")
+	if atomic.LoadInt32(&hits) != 1 {
+		t.Fatalf("expected one provider call before canceled retry wait, got %d", atomic.LoadInt32(&hits))
+	}
+}
+
+func TestHandlerDeadlineCancelsCapacityWaitAndReleasesAdmission(t *testing.T) {
+	started := make(chan struct{}, 1)
+	release := make(chan struct{})
+	var hits int32
+	provider := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if atomic.AddInt32(&hits, 1) == 1 {
+			started <- struct{}{}
+			<-release
+		}
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = w.Write([]byte(`{"choices":[{"message":{"role":"assistant","content":"ok"}}]}`))
+	}))
+	defer provider.Close()
+	defer func() {
+		select {
+		case <-release:
+		default:
+			close(release)
+		}
+	}()
+	queue := 1
+	h := newRetryHandler(t, provider.URL+"/v1", intPointer(0), 1, &queue)
+	first := httptest.NewRecorder()
+	done := make(chan struct{})
+	go func() {
+		defer close(done)
+		h.ServeHTTP(first, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p"}`)))
+	}()
+	select {
+	case <-started:
+	case <-time.After(3 * time.Second):
+		t.Fatal("first provider call did not start")
+	}
+	waiting := httptest.NewRecorder()
+	h.ServeHTTP(waiting, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p","run_timeout_seconds":1}`)))
+	assertHTTPErrorCode(t, waiting, http.StatusGatewayTimeout, "deadline_exceeded")
+	if atomic.LoadInt32(&hits) != 1 {
+		t.Fatalf("capacity waiter reached provider: hits=%d", atomic.LoadInt32(&hits))
+	}
+	close(release)
+	select {
+	case <-done:
+	case <-time.After(3 * time.Second):
+		t.Fatal("first request did not finish")
+	}
+	if first.Code != http.StatusOK {
+		t.Fatalf("first request: status=%d body=%s", first.Code, first.Body.String())
+	}
+	third := httptest.NewRecorder()
+	h.ServeHTTP(third, httptest.NewRequest(http.MethodPost, "/v1/runs", strings.NewReader(`{"prompt_id":"p"}`)))
+	if third.Code != http.StatusOK || atomic.LoadInt32(&hits) != 2 {
+		t.Fatalf("admission was not released: status=%d hits=%d body=%s", third.Code, atomic.LoadInt32(&hits), third.Body.String())
+	}
+}
+
+func newRetryHandler(t *testing.T, endpoint string, retries *int, concurrency int, queue *int) *Handler {
+	t.Helper()
+	promptDir := t.TempDir()
+	profileDir := t.TempDir()
+	if err := os.WriteFile(filepath.Join(promptDir, "prompt.yaml"), []byte("id: p\nversion: \"1\"\ndefault_profile: limited\nmessages:\n  - role: user\n    content: hi\noutput:\n  format: text\n  validation_mode: none\n"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.WriteFile(filepath.Join(profileDir, "profile.yaml"), []byte("id: limited\nbackend: limited\nmodel: test\n"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	engine, err := promptkit.NewEngine(promptkit.Config{PromptDir: promptDir, ProfileDir: profileDir, TransientRetries: retries}, promptkit.WithBackend(promptkit.Backend{ID: "limited", Endpoint: endpoint, ConcurrencyLimit: concurrency, QueueCapacity: queue}))
+	if err != nil {
+		t.Fatal(err)
+	}
+	return NewHandler(engine)
+}
+
 func TestHandlerRejectsOverlongSessionIDAndNonStringReasoningEffort(t *testing.T) {
 	engine := newHandlerEngine(t)
 	for _, body := range []string{
@@ -1126,6 +1630,10 @@ func TestHandlerPublicErrorMapping(t *testing.T) {
 		{name: "prompt render", err: wrap(promptkit.ErrPromptRender, fmt.Errorf("render failed")), status: http.StatusBadRequest, code: "prompt_render_failed", message: "failed to render prompt", avoidCause: "render failed"},
 		{name: "capacity", err: &promptkit.CapacityError{BackendID: "private-backend"}, status: http.StatusServiceUnavailable, code: "capacity_exceeded", message: "model backend capacity is exhausted", avoidCause: "private-backend"},
 		{name: "llm", err: wrap(promptkit.ErrLLMGenerate, fmt.Errorf("llm failed")), status: http.StatusBadGateway, code: "llm_failed", message: "model generation request failed", avoidCause: "llm failed"},
+		{name: "transport", err: wrap(promptkit.ErrTransport, fmt.Errorf("private transport detail")), status: http.StatusBadGateway, code: "llm_failed", message: "model generation request failed", avoidCause: "private transport detail"},
+		{name: "malformed response", err: wrap(promptkit.ErrMalformedResponse, fmt.Errorf("private response detail")), status: http.StatusBadGateway, code: "llm_failed", message: "model generation request failed", avoidCause: "private response detail"},
+		{name: "retry exhaustion", err: wrap(promptkit.ErrRetriesExhausted, fmt.Errorf("private provider detail")), status: http.StatusBadGateway, code: "llm_failed", message: "model generation request failed", avoidCause: "private provider detail"},
+		{name: "retry exhaustion value", err: fmt.Errorf("generation: %w", &promptkit.RetryExhaustedError{}), status: http.StatusBadGateway, code: "llm_failed", message: "model generation request failed"},
 		{name: "validation runtime", err: wrap(promptkit.ErrValidation, fmt.Errorf("validator broke")), status: http.StatusInternalServerError, code: "validation_runtime_failed", message: "validation runtime failed", avoidCause: "validator broke"},
 	}
 

internal/config/config.go

@@ -1,289 +0,0 @@
-package config
-
-import (
-	"bytes"
-	"errors"
-	"fmt"
-	"io"
-	"os"
-	"path/filepath"
-	"sort"
-	"strings"
-
-	"gitea.maximumdirect.net/eric/scriptorium/internal/defaults"
-	renderformat "gitea.maximumdirect.net/eric/scriptorium/internal/format"
-	"gopkg.in/yaml.v3"
-)
-
-const (
-	DefaultConfigPath      = "/etc/scriptorium/config.yml"
-	DefaultConfigPathLocal = "/usr/local/etc/scriptorium/config.yml"
-)
-
-var defaultConfigSearchPaths = []string{
-	DefaultConfigPathLocal,
-	DefaultConfigPath,
-}
-
-var (
-	ErrConfigNotFound    = errors.New("config file not found")
-	ErrInvalidConfigYAML = errors.New("invalid config YAML")
-	ErrInvalidConfig     = errors.New("invalid config")
-)
-
-// Config is the on-disk YAML shape for application-level settings.
-type Config struct {
-	PromptDir  string                   `yaml:"prompt_dir"`
-	ProfileDir string                   `yaml:"profile_dir"`
-	SchemaDir  string                   `yaml:"schema_dir"`
-	Server     ServerConfig             `yaml:"server"`
-	Defaults   DefaultsConfig           `yaml:"defaults"`
-	Backends   map[string]BackendConfig `yaml:"backends"`
-}
-
-type BackendConfig struct {
-	Endpoint         string         `yaml:"endpoint"`
-	APIKeyEnv        string         `yaml:"api_key_env"`
-	ExtraParams      map[string]any `yaml:"extra_params"`
-	ConcurrencyLimit int            `yaml:"concurrency_limit"`
-	QueueCapacity    *int           `yaml:"queue_capacity"`
-}
-
-type BackendSettings struct {
-	ID               string
-	Endpoint         string
-	APIKeyEnv        string
-	ExtraParams      map[string]any
-	ConcurrencyLimit int
-	QueueCapacity    *int
-}
-
-type ServerConfig struct {
-	Addr             string `yaml:"addr"`
-	ArtifactRoot     string `yaml:"artifact_root"`
-	MaxRequestBytes  *int64 `yaml:"max_request_bytes"`
-	MaxArtifactBytes *int64 `yaml:"max_artifact_bytes"`
-	MaxResponseBytes *int64 `yaml:"max_response_bytes"`
-}
-
-type DefaultsConfig struct {
-	RenderFormat string `yaml:"render_format"`
-}
-
-// AppSettings is the resolved application settings used by adapters.
-type AppSettings struct {
-	PromptDir           string
-	ProfileDir          string
-	SchemaDir           string
-	ServerAddr          string
-	ArtifactRoot        string
-	MaxRequestBytes     int64
-	MaxArtifactBytes    int64
-	MaxResponseBytes    int64
-	DefaultRenderFormat renderformat.PreparedRunOutputFormat
-	Backends            []BackendSettings
-}
-
-// CLIOverrides can be applied after config load to enforce precedence.
-type CLIOverrides struct {
-	PromptDir        string
-	ProfileDir       string
-	SchemaDir        string
-	ServerAddr       string
-	ArtifactRoot     string
-	MaxRequestBytes  *int64
-	MaxArtifactBytes *int64
-	MaxResponseBytes *int64
-	RenderFormat     string
-}
-
-// BuiltInDefaults returns compile-time application defaults.
-func BuiltInDefaults() AppSettings {
-	return AppSettings{
-		SchemaDir:           defaults.SchemaDirDefault,
-		ServerAddr:          defaults.HTTPAddrDefault,
-		MaxRequestBytes:     defaults.HTTPMaxRequestBytesDefault,
-		MaxArtifactBytes:    defaults.HTTPMaxArtifactBytesDefault,
-		MaxResponseBytes:    defaults.HTTPMaxResponseBytesDefault,
-		DefaultRenderFormat: renderformat.DefaultPreparedRunOutputFormat,
-	}
-}
-
-// LoadConfig loads and resolves app settings from YAML.
-//
-// If explicit is false and the file does not exist, built-in defaults are returned.
-// If explicit is true, missing file is an error.
-func LoadConfig(path string, explicit bool) (AppSettings, error) {
-	resolved := BuiltInDefaults()
-	configPath := strings.TrimSpace(path)
-	if configPath == "" {
-		if explicit {
-			configPath = DefaultConfigPath
-		} else {
-			for _, candidate := range defaultConfigSearchPaths {
-				if _, err := os.Stat(candidate); err == nil {
-					configPath = candidate
-					break
-				} else if !errors.Is(err, os.ErrNotExist) {
-					return AppSettings{}, fmt.Errorf("failed to stat config file %q: %w", candidate, err)
-				}
-			}
-		}
-	}
-	if configPath == "" {
-		if explicit {
-			return AppSettings{}, fmt.Errorf("%w: %s", ErrConfigNotFound, DefaultConfigPath)
-		}
-		return resolved, nil
-	}
-	configPath = filepath.Clean(configPath)
-
-	raw, err := os.ReadFile(configPath)
-	if err != nil {
-		if errors.Is(err, os.ErrNotExist) {
-			if explicit {
-				return AppSettings{}, fmt.Errorf("%w: %s", ErrConfigNotFound, configPath)
-			}
-			return resolved, nil
-		}
-		return AppSettings{}, fmt.Errorf("failed to read config file %q: %w", configPath, err)
-	}
-
-	cfg, err := decodeConfig(raw)
-	if err != nil {
-		return AppSettings{}, err
-	}
-
-	resolved, err = applyConfig(resolved, cfg)
-	if err != nil {
-		return AppSettings{}, err
-	}
-	return resolved, nil
-}
-
-// ApplyCLIOverrides applies CLI-provided overrides over resolved config settings.
-func ApplyCLIOverrides(base AppSettings, overrides CLIOverrides) (AppSettings, error) {
-	out := base
-
-	if v := strings.TrimSpace(overrides.PromptDir); v != "" {
-		out.PromptDir = filepath.Clean(v)
-	}
-	if v := strings.TrimSpace(overrides.ProfileDir); v != "" {
-		out.ProfileDir = filepath.Clean(v)
-	}
-	if v := strings.TrimSpace(overrides.SchemaDir); v != "" {
-		out.SchemaDir = filepath.Clean(v)
-	}
-	if v := strings.TrimSpace(overrides.ServerAddr); v != "" {
-		out.ServerAddr = v
-	}
-	if v := strings.TrimSpace(overrides.ArtifactRoot); v != "" {
-		out.ArtifactRoot = filepath.Clean(v)
-	}
-	if overrides.MaxRequestBytes != nil {
-		if *overrides.MaxRequestBytes < 0 {
-			return AppSettings{}, fmt.Errorf("%w: server.max_request_bytes must be greater than or equal to 0", ErrInvalidConfig)
-		}
-		out.MaxRequestBytes = *overrides.MaxRequestBytes
-	}
-	if overrides.MaxArtifactBytes != nil {
-		if *overrides.MaxArtifactBytes < 0 {
-			return AppSettings{}, fmt.Errorf("%w: server.max_artifact_bytes must be greater than or equal to 0", ErrInvalidConfig)
-		}
-		out.MaxArtifactBytes = *overrides.MaxArtifactBytes
-	}
-	if overrides.MaxResponseBytes != nil {
-		if *overrides.MaxResponseBytes < 0 {
-			return AppSettings{}, fmt.Errorf("%w: server.max_response_bytes must be greater than or equal to 0", ErrInvalidConfig)
-		}
-		out.MaxResponseBytes = *overrides.MaxResponseBytes
-	}
-	if rawFormat := strings.TrimSpace(overrides.RenderFormat); rawFormat != "" {
-		parsed, err := renderformat.ParsePreparedRunOutputFormat(rawFormat)
-		if err != nil {
-			return AppSettings{}, fmt.Errorf("%w: defaults.render_format: %v", ErrInvalidConfig, err)
-		}
-		out.DefaultRenderFormat = parsed
-	}
-
-	return out, nil
-}
-
-func decodeConfig(raw []byte) (Config, error) {
-	var cfg Config
-	decoder := yaml.NewDecoder(bytes.NewReader(raw))
-	decoder.KnownFields(true)
-	if err := decoder.Decode(&cfg); err != nil {
-		if errors.Is(err, io.EOF) {
-			return Config{}, nil
-		}
-		return Config{}, fmt.Errorf("%w: %v", ErrInvalidConfigYAML, err)
-	}
-	return cfg, nil
-}
-
-func applyConfig(base AppSettings, cfg Config) (AppSettings, error) {
-	out := base
-
-	if v := strings.TrimSpace(cfg.PromptDir); v != "" {
-		out.PromptDir = filepath.Clean(v)
-	}
-	if v := strings.TrimSpace(cfg.ProfileDir); v != "" {
-		out.ProfileDir = filepath.Clean(v)
-	}
-	if v := strings.TrimSpace(cfg.SchemaDir); v != "" {
-		out.SchemaDir = filepath.Clean(v)
-	}
-	if v := strings.TrimSpace(cfg.Server.Addr); v != "" {
-		out.ServerAddr = v
-	}
-	if v := strings.TrimSpace(cfg.Server.ArtifactRoot); v != "" {
-		out.ArtifactRoot = filepath.Clean(v)
-	}
-	if cfg.Server.MaxRequestBytes != nil {
-		if *cfg.Server.MaxRequestBytes < 0 {
-			return AppSettings{}, fmt.Errorf("%w: server.max_request_bytes must be greater than or equal to 0", ErrInvalidConfig)
-		}
-		out.MaxRequestBytes = *cfg.Server.MaxRequestBytes
-	}
-	if cfg.Server.MaxArtifactBytes != nil {
-		if *cfg.Server.MaxArtifactBytes < 0 {
-			return AppSettings{}, fmt.Errorf("%w: server.max_artifact_bytes must be greater than or equal to 0", ErrInvalidConfig)
-		}
-		out.MaxArtifactBytes = *cfg.Server.MaxArtifactBytes
-	}
-	if cfg.Server.MaxResponseBytes != nil {
-		if *cfg.Server.MaxResponseBytes < 0 {
-			return AppSettings{}, fmt.Errorf("%w: server.max_response_bytes must be greater than or equal to 0", ErrInvalidConfig)
-		}
-		out.MaxResponseBytes = *cfg.Server.MaxResponseBytes
-	}
-	if rawFormat := strings.TrimSpace(cfg.Defaults.RenderFormat); rawFormat != "" {
-		parsed, err := renderformat.ParsePreparedRunOutputFormat(rawFormat)
-		if err != nil {
-			return AppSettings{}, fmt.Errorf("%w: defaults.render_format: %v", ErrInvalidConfig, err)
-		}
-		out.DefaultRenderFormat = parsed
-	}
-	if len(cfg.Backends) > 0 {
-		ids := make([]string, 0, len(cfg.Backends))
-		for id := range cfg.Backends {
-			ids = append(ids, id)
-		}
-		sort.Strings(ids)
-		out.Backends = make([]BackendSettings, 0, len(ids))
-		for _, id := range ids {
-			backend := cfg.Backends[id]
-			out.Backends = append(out.Backends, BackendSettings{
-				ID:               id,
-				Endpoint:         backend.Endpoint,
-				APIKeyEnv:        backend.APIKeyEnv,
-				ExtraParams:      backend.ExtraParams,
-				ConcurrencyLimit: backend.ConcurrencyLimit,
-				QueueCapacity:    backend.QueueCapacity,
-			})
-		}
-	}
-
-	return out, nil
-}

internal/config/config_test.go

@@ -1,423 +0,0 @@
-package config
-
-import (
-	"errors"
-	"os"
-	"path/filepath"
-	"reflect"
-	"testing"
-
-	"gitea.maximumdirect.net/eric/scriptorium/internal/defaults"
-	renderformat "gitea.maximumdirect.net/eric/scriptorium/internal/format"
-)
-
-func TestLoadConfigMissingImplicitPathUsesBuiltInDefaults(t *testing.T) {
-	tmp := t.TempDir()
-	missing := filepath.Join(tmp, "missing.yml")
-
-	got, err := LoadConfig(missing, false)
-	if err != nil {
-		t.Fatalf("expected no error, got %v", err)
-	}
-
-	want := BuiltInDefaults()
-	if !reflect.DeepEqual(got, want) {
-		t.Fatalf("unexpected settings: got=%+v want=%+v", got, want)
-	}
-}
-
-func TestBuiltInDefaultsIncludeHTTPSizeLimits(t *testing.T) {
-	got := BuiltInDefaults()
-
-	if got.MaxRequestBytes != defaults.HTTPMaxRequestBytesDefault {
-		t.Fatalf("unexpected max request bytes: %d", got.MaxRequestBytes)
-	}
-	if got.MaxArtifactBytes != defaults.HTTPMaxArtifactBytesDefault {
-		t.Fatalf("unexpected max artifact bytes: %d", got.MaxArtifactBytes)
-	}
-	if got.MaxResponseBytes != defaults.HTTPMaxResponseBytesDefault {
-		t.Fatalf("unexpected max response bytes: %d", got.MaxResponseBytes)
-	}
-}
-
-func TestLoadConfigMissingExplicitPathReturnsError(t *testing.T) {
-	tmp := t.TempDir()
-	missing := filepath.Join(tmp, "missing.yml")
-
-	_, err := LoadConfig(missing, true)
-	if err == nil {
-		t.Fatal("expected error for missing explicit config")
-	}
-	if !errors.Is(err, ErrConfigNotFound) {
-		t.Fatalf("expected ErrConfigNotFound, got %v", err)
-	}
-}
-
-func TestLoadConfigInvalidYAMLReturnsError(t *testing.T) {
-	path := writeConfigFile(t, "config.yml", "prompt_dir: [")
-
-	_, err := LoadConfig(path, true)
-	if err == nil {
-		t.Fatal("expected invalid YAML error")
-	}
-	if !errors.Is(err, ErrInvalidConfigYAML) {
-		t.Fatalf("expected ErrInvalidConfigYAML, got %v", err)
-	}
-}
-
-func TestLoadConfigInvalidYAMLImplicitPathReturnsError(t *testing.T) {
-	path := writeConfigFile(t, "config.yml", "prompt_dir: [")
-
-	_, err := LoadConfig(path, false)
-	if err == nil {
-		t.Fatal("expected invalid YAML error")
-	}
-	if !errors.Is(err, ErrInvalidConfigYAML) {
-		t.Fatalf("expected ErrInvalidConfigYAML, got %v", err)
-	}
-}
-
-func TestLoadConfigUnknownFieldReturnsError(t *testing.T) {
-	path := writeConfigFile(t, "config.yml", "unknown_field: true\n")
-
-	_, err := LoadConfig(path, true)
-	if err == nil {
-		t.Fatal("expected unknown field error")
-	}
-	if !errors.Is(err, ErrInvalidConfigYAML) {
-		t.Fatalf("expected ErrInvalidConfigYAML, got %v", err)
-	}
-}
-
-func TestLoadConfigAPIKeyFieldIsRejectedAsUnknown(t *testing.T) {
-	path := writeConfigFile(t, "config.yml", "api_key: secret\n")
-
-	_, err := LoadConfig(path, true)
-	if err == nil {
-		t.Fatal("expected unknown field error for api_key")
-	}
-	if !errors.Is(err, ErrInvalidConfigYAML) {
-		t.Fatalf("expected ErrInvalidConfigYAML, got %v", err)
-	}
-}
-
-func TestLoadConfigBackendsRetainsResolvedSettingsInSortedOrder(t *testing.T) {
-	path := writeConfigFile(t, "config.yml", `
-backends:
-  zebra:
-    endpoint: https://zebra.example/v1
-    api_key_env: ZEBRA_API_KEY
-    extra_params:
-      provider_option: enabled
-      nested:
-        enabled: true
-        attempts: 2
-    concurrency_limit: 2
-  alpha:
-    endpoint: http://alpha.example/v1
-    queue_capacity: 0
-`)
-
-	got, err := LoadConfig(path, true)
-	if err != nil {
-		t.Fatalf("expected no error, got %v", err)
-	}
-	if len(got.Backends) != 2 {
-		t.Fatalf("expected two backends, got %#v", got.Backends)
-	}
-	if got.Backends[0].ID != "alpha" || got.Backends[1].ID != "zebra" {
-		t.Fatalf("expected sorted backend IDs, got %#v", got.Backends)
-	}
-	if got.Backends[0].QueueCapacity == nil || *got.Backends[0].QueueCapacity != 0 {
-		t.Fatalf("expected explicit zero queue capacity, got %#v", got.Backends[0].QueueCapacity)
-	}
-	if got.Backends[1].QueueCapacity != nil {
-		t.Fatalf("expected omitted queue capacity to remain nil, got %#v", got.Backends[1].QueueCapacity)
-	}
-	wantParams := map[string]any{
-		"provider_option": "enabled",
-		"nested": map[string]any{
-			"enabled":  true,
-			"attempts": 2,
-		},
-	}
-	if !reflect.DeepEqual(got.Backends[1].ExtraParams, wantParams) {
-		t.Fatalf("unexpected extra params: got=%#v want=%#v", got.Backends[1].ExtraParams, wantParams)
-	}
-	if got.Backends[1].Endpoint != "https://zebra.example/v1" || got.Backends[1].APIKeyEnv != "ZEBRA_API_KEY" || got.Backends[1].ConcurrencyLimit != 2 {
-		t.Fatalf("unexpected zebra backend: %#v", got.Backends[1])
-	}
-}
-
-func TestLoadConfigRejectsUnknownOrSecretBackendFields(t *testing.T) {
-	for name, body := range map[string]string{
-		"unknown": "backends:\n  local:\n    endpoint: http://localhost:11434/v1\n    unexpected: value\n",
-		"secret":  "backends:\n  local:\n    endpoint: http://localhost:11434/v1\n    api_key: secret\n",
-	} {
-		t.Run(name, func(t *testing.T) {
-			path := writeConfigFile(t, "config.yml", body)
-			_, err := LoadConfig(path, true)
-			if !errors.Is(err, ErrInvalidConfigYAML) {
-				t.Fatalf("expected ErrInvalidConfigYAML, got %v", err)
-			}
-		})
-	}
-}
-
-func TestLoadConfigRejectsInvalidBackendFieldTypes(t *testing.T) {
-	path := writeConfigFile(t, "config.yml", "backends:\n  local:\n    endpoint: http://localhost:11434/v1\n    queue_capacity: not-a-number\n")
-
-	_, err := LoadConfig(path, true)
-	if !errors.Is(err, ErrInvalidConfigYAML) {
-		t.Fatalf("expected ErrInvalidConfigYAML, got %v", err)
-	}
-}
-
-func TestLoadConfigValidConfigSetsDirectoriesAndServerAddr(t *testing.T) {
-	path := writeConfigFile(t, "config.yml", `
-prompt_dir: ./prompts
-profile_dir: ./profiles
-schema_dir: ./schemas
-server:
-  addr: 127.0.0.1:9090
-  artifact_root: ./artifacts
-  max_request_bytes: 1024
-  max_artifact_bytes: 2048
-  max_response_bytes: 4096
-defaults:
-  render_format: json
-`)
-
-	got, err := LoadConfig(path, true)
-	if err != nil {
-		t.Fatalf("expected no error, got %v", err)
-	}
-
-	if got.PromptDir != filepath.Clean("./prompts") {
-		t.Fatalf("unexpected prompt_dir: %q", got.PromptDir)
-	}
-	if got.ProfileDir != filepath.Clean("./profiles") {
-		t.Fatalf("unexpected profile_dir: %q", got.ProfileDir)
-	}
-	if got.SchemaDir != filepath.Clean("./schemas") {
-		t.Fatalf("unexpected schema_dir: %q", got.SchemaDir)
-	}
-	if got.ServerAddr != "127.0.0.1:9090" {
-		t.Fatalf("unexpected server.addr: %q", got.ServerAddr)
-	}
-	if got.ArtifactRoot != filepath.Clean("./artifacts") {
-		t.Fatalf("unexpected server.artifact_root: %q", got.ArtifactRoot)
-	}
-	if got.MaxRequestBytes != 1024 {
-		t.Fatalf("unexpected server.max_request_bytes: %d", got.MaxRequestBytes)
-	}
-	if got.MaxArtifactBytes != 2048 {
-		t.Fatalf("unexpected server.max_artifact_bytes: %d", got.MaxArtifactBytes)
-	}
-	if got.MaxResponseBytes != 4096 {
-		t.Fatalf("unexpected server.max_response_bytes: %d", got.MaxResponseBytes)
-	}
-	if got.DefaultRenderFormat != renderformat.PreparedRunFormatJSON {
-		t.Fatalf("unexpected defaults.render_format: %q", got.DefaultRenderFormat)
-	}
-}
-
-func TestLoadConfigAcceptsZeroHTTPSizeLimits(t *testing.T) {
-	path := writeConfigFile(t, "config.yml", `
-server:
-  max_request_bytes: 0
-  max_artifact_bytes: 0
-  max_response_bytes: 0
-`)
-
-	got, err := LoadConfig(path, true)
-	if err != nil {
-		t.Fatalf("expected no error, got %v", err)
-	}
-	if got.MaxRequestBytes != 0 || got.MaxArtifactBytes != 0 || got.MaxResponseBytes != 0 {
-		t.Fatalf("expected zero limits to be preserved, got request=%d artifact=%d response=%d", got.MaxRequestBytes, got.MaxArtifactBytes, got.MaxResponseBytes)
-	}
-}
-
-func TestLoadConfigRejectsNegativeHTTPSizeLimits(t *testing.T) {
-	tests := []struct {
-		name string
-		body string
-	}{
-		{name: "request", body: "server:\n  max_request_bytes: -1\n"},
-		{name: "artifact", body: "server:\n  max_artifact_bytes: -1\n"},
-		{name: "response", body: "server:\n  max_response_bytes: -1\n"},
-	}
-
-	for _, tc := range tests {
-		t.Run(tc.name, func(t *testing.T) {
-			path := writeConfigFile(t, "config.yml", tc.body)
-			_, err := LoadConfig(path, true)
-			if !errors.Is(err, ErrInvalidConfig) {
-				t.Fatalf("expected ErrInvalidConfig, got %v", err)
-			}
-		})
-	}
-}
-
-func TestLoadConfigEmptyFileResolvesToBuiltInDefaults(t *testing.T) {
-	path := writeConfigFile(t, "config.yml", "")
-
-	got, err := LoadConfig(path, true)
-	if err != nil {
-		t.Fatalf("expected no error, got %v", err)
-	}
-
-	want := BuiltInDefaults()
-	if !reflect.DeepEqual(got, want) {
-		t.Fatalf("unexpected settings: got=%+v want=%+v", got, want)
-	}
-}
-
-func TestLoadConfigImplicitSearchPrefersUsrLocalEtcOverEtc(t *testing.T) {
-	tmp := t.TempDir()
-	localPath := filepath.Join(tmp, "usr-local.yml")
-	etcPath := filepath.Join(tmp, "etc.yml")
-
-	if err := os.WriteFile(localPath, []byte("prompt_dir: ./from-usr-local\n"), 0o644); err != nil {
-		t.Fatal(err)
-	}
-	if err := os.WriteFile(etcPath, []byte("prompt_dir: ./from-etc\n"), 0o644); err != nil {
-		t.Fatal(err)
-	}
-
-	orig := defaultConfigSearchPaths
-	defaultConfigSearchPaths = []string{localPath, etcPath}
-	t.Cleanup(func() { defaultConfigSearchPaths = orig })
-
-	got, err := LoadConfig("", false)
-	if err != nil {
-		t.Fatalf("expected no error, got %v", err)
-	}
-	if got.PromptDir != filepath.Clean("./from-usr-local") {
-		t.Fatalf("expected usr-local config to win, got prompt_dir=%q", got.PromptDir)
-	}
-}
-
-func TestLoadConfigImplicitSearchFallsBackToEtcWhenUsrLocalMissing(t *testing.T) {
-	tmp := t.TempDir()
-	missingLocal := filepath.Join(tmp, "missing-local.yml")
-	etcPath := filepath.Join(tmp, "etc.yml")
-
-	if err := os.WriteFile(etcPath, []byte("prompt_dir: ./from-etc\n"), 0o644); err != nil {
-		t.Fatal(err)
-	}
-
-	orig := defaultConfigSearchPaths
-	defaultConfigSearchPaths = []string{missingLocal, etcPath}
-	t.Cleanup(func() { defaultConfigSearchPaths = orig })
-
-	got, err := LoadConfig("", false)
-	if err != nil {
-		t.Fatalf("expected no error, got %v", err)
-	}
-	if got.PromptDir != filepath.Clean("./from-etc") {
-		t.Fatalf("expected etc fallback config, got prompt_dir=%q", got.PromptDir)
-	}
-}
-
-func TestApplyCLIOverridesAppliesPrecedence(t *testing.T) {
-	base := AppSettings{
-		PromptDir:           "/from/config/prompts",
-		ProfileDir:          "/from/config/profiles",
-		SchemaDir:           "/from/config/schemas",
-		ServerAddr:          ":1234",
-		ArtifactRoot:        "/from/config/artifacts",
-		MaxRequestBytes:     111,
-		MaxArtifactBytes:    222,
-		MaxResponseBytes:    333,
-		DefaultRenderFormat: renderformat.PreparedRunFormatJSON,
-	}
-	maxRequestBytes := int64(0)
-	maxArtifactBytes := int64(444)
-	maxResponseBytes := int64(555)
-
-	got, err := ApplyCLIOverrides(base, CLIOverrides{
-		PromptDir:        "./prompts-cli",
-		ProfileDir:       "./profiles-cli",
-		SchemaDir:        "./schemas-cli",
-		ServerAddr:       ":8081",
-		ArtifactRoot:     "./artifacts-cli",
-		MaxRequestBytes:  &maxRequestBytes,
-		MaxArtifactBytes: &maxArtifactBytes,
-		MaxResponseBytes: &maxResponseBytes,
-		RenderFormat:     "text",
-	})
-	if err != nil {
-		t.Fatalf("expected no error, got %v", err)
-	}
-
-	if got.PromptDir != filepath.Clean("./prompts-cli") {
-		t.Fatalf("unexpected prompt dir: %q", got.PromptDir)
-	}
-	if got.ProfileDir != filepath.Clean("./profiles-cli") {
-		t.Fatalf("unexpected profile dir: %q", got.ProfileDir)
-	}
-	if got.SchemaDir != filepath.Clean("./schemas-cli") {
-		t.Fatalf("unexpected schema dir: %q", got.SchemaDir)
-	}
-	if got.ServerAddr != ":8081" {
-		t.Fatalf("unexpected server addr: %q", got.ServerAddr)
-	}
-	if got.ArtifactRoot != filepath.Clean("./artifacts-cli") {
-		t.Fatalf("unexpected artifact root: %q", got.ArtifactRoot)
-	}
-	if got.MaxRequestBytes != 0 {
-		t.Fatalf("unexpected max request bytes: %d", got.MaxRequestBytes)
-	}
-	if got.MaxArtifactBytes != 444 {
-		t.Fatalf("unexpected max artifact bytes: %d", got.MaxArtifactBytes)
-	}
-	if got.MaxResponseBytes != 555 {
-		t.Fatalf("unexpected max response bytes: %d", got.MaxResponseBytes)
-	}
-	if got.DefaultRenderFormat != renderformat.PreparedRunFormatText {
-		t.Fatalf("unexpected render format: %q", got.DefaultRenderFormat)
-	}
-}
-
-func TestApplyCLIOverridesRejectsNegativeHTTPSizeLimits(t *testing.T) {
-	negative := int64(-1)
-
-	tests := []struct {
-		name      string
-		overrides CLIOverrides
-	}{
-		{name: "request", overrides: CLIOverrides{MaxRequestBytes: &negative}},
-		{name: "artifact", overrides: CLIOverrides{MaxArtifactBytes: &negative}},
-		{name: "response", overrides: CLIOverrides{MaxResponseBytes: &negative}},
-	}
-
-	for _, tc := range tests {
-		t.Run(tc.name, func(t *testing.T) {
-			_, err := ApplyCLIOverrides(BuiltInDefaults(), tc.overrides)
-			if !errors.Is(err, ErrInvalidConfig) {
-				t.Fatalf("expected ErrInvalidConfig, got %v", err)
-			}
-		})
-	}
-}
-
-func TestApplyCLIOverridesInvalidRenderFormatReturnsError(t *testing.T) {
-	_, err := ApplyCLIOverrides(BuiltInDefaults(), CLIOverrides{RenderFormat: "yaml"})
-	if err == nil {
-		t.Fatal("expected invalid render format error")
-	}
-	if !errors.Is(err, ErrInvalidConfig) {
-		t.Fatalf("expected ErrInvalidConfig, got %v", err)
-	}
-}
-
-func writeConfigFile(t *testing.T, name, content string) string {
-	t.Helper()
-	path := filepath.Join(t.TempDir(), name)
-	if err := os.WriteFile(path, []byte(content), 0o644); err != nil {
-		t.Fatalf("failed to write config file: %v", err)
-	}
-	return path
-}

internal/config/document.go

@@ -0,0 +1,109 @@
+package config
+
+import (
+	"path/filepath"
+	"strings"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+	"gitea.maximumdirect.net/eric/scriptorium/internal/defaults"
+	renderformat "gitea.maximumdirect.net/eric/scriptorium/internal/format"
+)
+
+// Document is Scriptorium's ConfigKit-native application configuration shape.
+type Document struct {
+	PromptDir        string                     `yaml:"prompt_dir"`
+	ProfileDir       string                     `yaml:"profile_dir"`
+	SchemaDir        string                     `yaml:"schema_dir"`
+	TransientRetries *int                       `yaml:"transient_retries"`
+	Server           ServerDocument             `yaml:"server"`
+	Defaults         DefaultsDocument           `yaml:"defaults"`
+	Backends         map[string]BackendDocument `yaml:"backends"`
+}
+
+type BackendDocument struct {
+	Endpoint         string         `yaml:"endpoint"`
+	APIKeyEnv        string         `yaml:"api_key_env"`
+	ExtraParams      map[string]any `yaml:"extra_params"`
+	ConcurrencyLimit int            `yaml:"concurrency_limit"`
+	QueueCapacity    *int           `yaml:"queue_capacity"`
+}
+
+type ServerDocument struct {
+	Addr             string `yaml:"addr"`
+	ArtifactRoot     string `yaml:"artifact_root"`
+	MaxRequestBytes  int64  `yaml:"max_request_bytes"`
+	MaxArtifactBytes int64  `yaml:"max_artifact_bytes"`
+	MaxResponseBytes int64  `yaml:"max_response_bytes"`
+}
+
+type DefaultsDocument struct {
+	RenderFormat string `yaml:"render_format"`
+}
+
+func documentDefinition() configkit.Definition[Document] {
+	return configkit.Definition[Document]{
+		Defaults: documentDefaults,
+		Resolve:  resolveDocument,
+		Validate: validateDocument,
+	}
+}
+
+func documentDefaults() Document {
+	return Document{
+		SchemaDir: defaults.SchemaDirDefault,
+		Server: ServerDocument{
+			Addr:             defaults.HTTPAddrDefault,
+			MaxRequestBytes:  defaults.HTTPMaxRequestBytesDefault,
+			MaxArtifactBytes: defaults.HTTPMaxArtifactBytesDefault,
+			MaxResponseBytes: defaults.HTTPMaxResponseBytesDefault,
+		},
+		Defaults: DefaultsDocument{RenderFormat: string(renderformat.DefaultPreparedRunOutputFormat)},
+	}
+}
+
+func resolveDocument(document *Document, resolution *configkit.Resolution) error {
+	normalizePath := func(path configkit.Path, value *string) {
+		cleaned := strings.TrimSpace(*value)
+		if cleaned != "" {
+			cleaned = filepath.Clean(cleaned)
+		}
+		if cleaned != *value {
+			configkit.Normalize(resolution, path, value, cleaned, "normalized application path")
+		}
+	}
+	normalizeString := func(path configkit.Path, value *string, normalized, reason string) {
+		if normalized != *value {
+			configkit.Normalize(resolution, path, value, normalized, reason)
+		}
+	}
+
+	normalizePath(configkit.Keys("prompt_dir"), &document.PromptDir)
+	normalizePath(configkit.Keys("profile_dir"), &document.ProfileDir)
+	normalizePath(configkit.Keys("schema_dir"), &document.SchemaDir)
+	normalizePath(configkit.Keys("server", "artifact_root"), &document.Server.ArtifactRoot)
+	normalizeString(configkit.Keys("server", "addr"), &document.Server.Addr,
+		strings.TrimSpace(document.Server.Addr), "trimmed server address")
+	normalizeString(configkit.Keys("defaults", "render_format"), &document.Defaults.RenderFormat,
+		strings.ToLower(strings.TrimSpace(document.Defaults.RenderFormat)), "normalized render format")
+	return nil
+}
+
+func validateDocument(document *Document, validation *configkit.Validation) {
+	if document.TransientRetries != nil && *document.TransientRetries < 0 {
+		validation.Error(configkit.Keys("transient_retries"), "must be greater than or equal to zero")
+	}
+	if document.Server.MaxRequestBytes < 0 {
+		validation.Error(configkit.Keys("server", "max_request_bytes"), "must be greater than or equal to zero")
+	}
+	if document.Server.MaxArtifactBytes < 0 {
+		validation.Error(configkit.Keys("server", "max_artifact_bytes"), "must be greater than or equal to zero")
+	}
+	if document.Server.MaxResponseBytes < 0 {
+		validation.Error(configkit.Keys("server", "max_response_bytes"), "must be greater than or equal to zero")
+	}
+	switch document.Defaults.RenderFormat {
+	case string(renderformat.OutputFormatText), string(renderformat.OutputFormatJSON):
+	default:
+		validation.Error(configkit.Keys("defaults", "render_format"), "must be text or json")
+	}
+}

internal/config/document_test.go

@@ -0,0 +1,163 @@
+package config
+
+import (
+	"bytes"
+	"errors"
+	"reflect"
+	"strings"
+	"testing"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+func loadDocumentBytes(content string) (*configkit.Result[Document], error) {
+	return configkit.Load(configkit.Bytes("application.yml", []byte(content)), configkit.Options[Document]{
+		Definition: documentDefinition(),
+	})
+}
+
+func TestDocumentDefaultsAndExplicitValues(t *testing.T) {
+	defaultsOnly, err := configkit.Load(configkit.NoSource(), configkit.Options[Document]{Definition: documentDefinition()})
+	if err != nil {
+		t.Fatalf("defaults-only load: %v", err)
+	}
+	if !reflect.DeepEqual(defaultsOnly.Config, documentDefaults()) || len(defaultsOnly.Sources()) != 0 {
+		t.Fatalf("unexpected defaults-only result: %+v sources=%+v", defaultsOnly.Config, defaultsOnly.Sources())
+	}
+
+	result, err := loadDocumentBytes(`prompt_dir: ""
+profile_dir: ""
+schema_dir: ""
+transient_retries: 0
+server:
+  addr: ""
+  artifact_root: ""
+  max_request_bytes: 0
+  max_artifact_bytes: 0
+  max_response_bytes: 0
+defaults:
+  render_format: " JSON "
+backends:
+  local:
+    endpoint: "http://fixture.invalid/v1"
+    queue_capacity: 0
+    extra_params:
+      sample_count: 23
+`)
+	if err != nil {
+		t.Fatalf("explicit-value load: %v", err)
+	}
+	config := result.Config
+	if config.SchemaDir != "" || config.Server.Addr != "" || config.Server.MaxRequestBytes != 0 || config.Server.MaxArtifactBytes != 0 || config.Server.MaxResponseBytes != 0 {
+		t.Fatalf("explicit empty and zero values did not replace defaults: %+v", config)
+	}
+	if config.TransientRetries == nil || *config.TransientRetries != 0 {
+		t.Fatalf("explicit zero retry budget lost: %+v", config.TransientRetries)
+	}
+	if backend := config.Backends["local"]; backend.QueueCapacity == nil || *backend.QueueCapacity != 0 || backend.ExtraParams["sample_count"] != int64(23) {
+		t.Fatalf("backend presence or dynamic integer changed: %+v", backend)
+	}
+	if config.Defaults.RenderFormat != "json" {
+		t.Fatalf("render format normalization failed: %q", config.Defaults.RenderFormat)
+	}
+
+	nullable, err := loadDocumentBytes("transient_retries: null\nbackends:\n  local:\n    queue_capacity: null\n")
+	if err != nil {
+		t.Fatalf("nullable-value load: %v", err)
+	}
+	if nullable.Config.TransientRetries != nil || nullable.Config.Backends["local"].QueueCapacity != nil {
+		t.Fatalf("nullable fields did not remain nil: %+v", nullable.Config)
+	}
+
+	nullBackends, err := loadDocumentBytes("backends: null\n")
+	if err != nil || nullBackends.Config.Backends != nil {
+		t.Fatalf("nullable backend map did not remain nil: %+v error=%v", nullBackends, err)
+	}
+}
+
+func TestDocumentRecordsEffectiveNormalization(t *testing.T) {
+	result, err := loadDocumentBytes(`prompt_dir: "  ./one/../two  "
+profile_dir: "  "
+server:
+  artifact_root: " ./data/../artifacts "
+  addr: "  :9000  "
+defaults:
+  render_format: " JSON "
+`)
+	if err != nil {
+		t.Fatalf("load normalized document: %v", err)
+	}
+	if result.Config.PromptDir != "two" || result.Config.ProfileDir != "" || result.Config.Server.ArtifactRoot != "artifacts" || result.Config.Server.Addr != ":9000" {
+		t.Fatalf("unexpected normalized values: %+v", result.Config)
+	}
+	var effective bytes.Buffer
+	if err := result.WriteEffective(&effective); err != nil {
+		t.Fatalf("write effective configuration: %v", err)
+	}
+	for _, fragment := range []string{"prompt_dir: two", "profile_dir: ''", "artifact_root: artifacts", "addr: :9000", "render_format: json"} {
+		if !strings.Contains(effective.String(), fragment) {
+			t.Fatalf("effective output omitted %q: %s", fragment, effective.String())
+		}
+	}
+	explanation, known := result.Explain(configkit.Keys("prompt_dir"))
+	if !known || explanation.Effective.YAML != "two\n" {
+		t.Fatalf("explanation does not agree with effective value: %+v known=%t", explanation, known)
+	}
+	found := false
+	for _, change := range explanation.History {
+		if change.Origin.Operation == configkit.ChangeNormalization {
+			found = true
+		}
+	}
+	if !found {
+		t.Fatalf("normalization was not recorded: %+v", explanation.History)
+	}
+}
+
+func TestDocumentValidationIsSafeAndPathSpecific(t *testing.T) {
+	for _, tc := range []struct{ name, content, path string }{
+		{"negative retry", "transient_retries: -9173\n", "/transient_retries"},
+		{"negative request limit", "server:\n  max_request_bytes: -9173\n", "/server/max_request_bytes"},
+		{"negative artifact limit", "server:\n  max_artifact_bytes: -9173\n", "/server/max_artifact_bytes"},
+		{"negative response limit", "server:\n  max_response_bytes: -9173\n", "/server/max_response_bytes"},
+		{"empty format", "defaults:\n  render_format: \"\"\n", "/defaults/render_format"},
+		{"unknown format", "defaults:\n  render_format: private-invalid-value\n", "/defaults/render_format"},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			_, err := loadDocumentBytes(tc.content)
+			var loadError *configkit.LoadError
+			if !errors.As(err, &loadError) {
+				t.Fatalf("expected typed validation error, got %v", err)
+			}
+			found := false
+			for _, diagnostic := range loadError.Diagnostics() {
+				if diagnostic.Code == configkit.CodeValidateError && diagnostic.Path.String() == tc.path {
+					found = true
+				}
+				if strings.Contains(diagnostic.Message, "9173") || strings.Contains(diagnostic.Message, "private-invalid-value") {
+					t.Fatalf("diagnostic disclosed rejected value: %+v", diagnostic)
+				}
+			}
+			if !found {
+				t.Fatalf("expected validation at %s, got %+v", tc.path, loadError.Diagnostics())
+			}
+		})
+	}
+}
+
+func TestDocumentRejectsNativeShapeErrors(t *testing.T) {
+	for _, content := range []string{
+		"server: null\n",
+		"defaults: null\n",
+		"server:\n  addr: null\n",
+		"server:\n  max_request_bytes: null\n",
+		"defaults:\n  render_format: null\n",
+		"server:\n  unknown_field: 1\n",
+		"backends:\n  local:\n    unknown_field: 1\n",
+	} {
+		_, err := loadDocumentBytes(content)
+		if err == nil {
+			t.Fatalf("unexpected acceptance of %q", content)
+		}
+	}
+}

internal/config/load.go

@@ -0,0 +1,96 @@
+package config
+
+import (
+	"errors"
+	"fmt"
+	"os"
+	"sort"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+	renderformat "gitea.maximumdirect.net/eric/scriptorium/internal/format"
+)
+
+// LoadOptions selects one application source and explicit configuration values.
+type LoadOptions struct {
+	Path      string
+	Explicit  bool
+	Overlay   string
+	Overrides []configkit.Override
+}
+
+// Load resolves one ConfigKit document and retains its inspection snapshot.
+func Load(options LoadOptions) (*configkit.Result[Document], error) {
+	var source configkit.Source
+	fileBacked := false
+	if options.Explicit {
+		source = configkit.File(options.Path)
+		fileBacked = true
+	} else {
+		selected := ""
+		for _, candidate := range defaultConfigSearchPaths {
+			if _, err := os.Stat(candidate); err == nil {
+				selected = candidate
+				break
+			} else if !errors.Is(err, os.ErrNotExist) {
+				reason := "filesystem error"
+				if errors.Is(err, os.ErrPermission) {
+					reason = "permission denied"
+				}
+				return nil, fmt.Errorf("cannot inspect configuration path %q: %s", candidate, reason)
+			}
+		}
+		if selected == "" {
+			source = configkit.NoSource()
+		} else {
+			source = configkit.File(selected)
+			fileBacked = true
+		}
+	}
+
+	result, err := configkit.Load(source, configkit.Options[Document]{
+		Definition: documentDefinition(),
+		Imports:    fileBacked,
+		Overlay:    options.Overlay,
+		Overrides:  options.Overrides,
+	})
+	if err != nil {
+		return nil, fmt.Errorf("load application configuration: %w", err)
+	}
+	return result, nil
+}
+
+// Settings converts a validated document without another merge or normalization.
+func Settings(document Document) AppSettings {
+	settings := AppSettings{
+		PromptDir:           document.PromptDir,
+		ProfileDir:          document.ProfileDir,
+		SchemaDir:           document.SchemaDir,
+		ServerAddr:          document.Server.Addr,
+		ArtifactRoot:        document.Server.ArtifactRoot,
+		MaxRequestBytes:     document.Server.MaxRequestBytes,
+		MaxArtifactBytes:    document.Server.MaxArtifactBytes,
+		MaxResponseBytes:    document.Server.MaxResponseBytes,
+		DefaultRenderFormat: renderformat.PreparedRunOutputFormat(document.Defaults.RenderFormat),
+		TransientRetries:    document.TransientRetries,
+	}
+	if document.Backends != nil {
+		ids := make([]string, 0, len(document.Backends))
+		for id := range document.Backends {
+			ids = append(ids, id)
+		}
+		sort.Strings(ids)
+		settings.Backends = make([]BackendSettings, 0, len(ids))
+		for _, id := range ids {
+			backend := document.Backends[id]
+			settings.Backends = append(settings.Backends, BackendSettings{
+				ID:               id,
+				Endpoint:         backend.Endpoint,
+				APIKeyEnv:        backend.APIKeyEnv,
+				ExtraParams:      backend.ExtraParams,
+				ConcurrencyLimit: backend.ConcurrencyLimit,
+				QueueCapacity:    backend.QueueCapacity,
+			})
+		}
+	}
+	return settings
+}

internal/config/load_test.go

@@ -0,0 +1,233 @@
+package config
+
+import (
+	"errors"
+	"os"
+	"path/filepath"
+	"reflect"
+	"strings"
+	"testing"
+
+	configkit "gitea.maximumdirect.net/eric/ConfigKit"
+)
+
+func useConfigSearchPaths(t *testing.T, paths ...string) {
+	t.Helper()
+	previous := defaultConfigSearchPaths
+	defaultConfigSearchPaths = paths
+	t.Cleanup(func() { defaultConfigSearchPaths = previous })
+}
+
+func writeApplicationFile(t *testing.T, path, body string) {
+	t.Helper()
+	if err := os.WriteFile(path, []byte(body), 0o600); err != nil {
+		t.Fatal(err)
+	}
+}
+
+func TestLoadSelectsOrderedSourceAndKeepsItsSnapshot(t *testing.T) {
+	dir := t.TempDir()
+	first := filepath.Join(dir, "first.yml")
+	second := filepath.Join(dir, "second.yml")
+	writeApplicationFile(t, first, "prompt_dir: ./first\n")
+	writeApplicationFile(t, second, "prompt_dir: ./second\n")
+	useConfigSearchPaths(t, first, second)
+
+	result, err := Load(LoadOptions{Path: filepath.Join(dir, "ignored.yml")})
+	if err != nil {
+		t.Fatalf("load first implicit source: %v", err)
+	}
+	if result.Config.PromptDir != "first" || len(result.Sources()) != 1 || result.Sources()[0].Path != first {
+		t.Fatalf("unexpected implicit source: config=%+v sources=%+v", result.Config, result.Sources())
+	}
+	writeApplicationFile(t, first, "prompt_dir: ./changed\n")
+	if result.Config.PromptDir != "first" {
+		t.Fatalf("result did not retain resolved snapshot: %+v", result.Config)
+	}
+
+	if err := os.Remove(first); err != nil {
+		t.Fatal(err)
+	}
+	result, err = Load(LoadOptions{})
+	if err != nil || result.Config.PromptDir != "second" || result.Sources()[0].Path != second {
+		t.Fatalf("second implicit source not selected: result=%+v err=%v", result, err)
+	}
+
+	result, err = Load(LoadOptions{Path: first, Explicit: true})
+	if result != nil || err == nil {
+		t.Fatalf("missing explicit source was accepted: result=%+v err=%v", result, err)
+	}
+	var loadError *configkit.LoadError
+	if !errors.As(err, &loadError) {
+		t.Fatalf("typed load error was lost: %v", err)
+	}
+}
+
+func TestLoadRejectsExplicitEmptyAndDistinguishesEmptyFile(t *testing.T) {
+	useConfigSearchPaths(t, filepath.Join(t.TempDir(), "absent.yml"))
+	if _, err := Load(LoadOptions{Explicit: true, Path: ""}); err == nil {
+		t.Fatal("explicit empty path was accepted")
+	} else {
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) {
+			t.Fatalf("expected upstream source diagnostic, got %v", err)
+		}
+	}
+
+	dir := t.TempDir()
+	empty := filepath.Join(dir, "empty.yml")
+	writeApplicationFile(t, empty, "")
+	if _, err := Load(LoadOptions{Path: empty, Explicit: true}); err == nil {
+		t.Fatal("empty file was treated as defaults-only")
+	}
+	writeApplicationFile(t, empty, "{}\n")
+	result, err := Load(LoadOptions{Path: empty, Explicit: true})
+	if err != nil || !reflect.DeepEqual(result.Config, documentDefaults()) {
+		t.Fatalf("empty mapping did not load defaults: result=%+v err=%v", result, err)
+	}
+}
+
+func TestLoadDefaultsOnlyAndOverridesBeforeValidation(t *testing.T) {
+	useConfigSearchPaths(t, filepath.Join(t.TempDir(), "absent.yml"))
+	result, err := Load(LoadOptions{})
+	if err != nil || !reflect.DeepEqual(result.Config, documentDefaults()) || len(result.Sources()) != 0 {
+		t.Fatalf("defaults-only load: result=%+v err=%v", result, err)
+	}
+	var effective strings.Builder
+	if err := result.WriteEffective(&effective); err != nil || !strings.Contains(effective.String(), "render_format: text") || len(result.Sources()) != 0 {
+		t.Fatalf("defaults-only inspection changed: effective=%q err=%v sources=%+v", effective.String(), err, result.Sources())
+	}
+
+	result, err = Load(LoadOptions{Overrides: []configkit.Override{{
+		Path: configkit.Keys("server", "max_request_bytes"), Value: int64(0), Flag: "--max-request-bytes",
+	}}})
+	if err != nil || result.Config.Server.MaxRequestBytes != 0 || len(result.Sources()) != 1 || result.Sources()[0].Layer != configkit.LayerCLI {
+		t.Fatalf("defaults-only override lost: result=%+v err=%v", result, err)
+	}
+
+	path := filepath.Join(t.TempDir(), "invalid.yml")
+	writeApplicationFile(t, path, "defaults:\n  render_format: invalid-choice\n")
+	result, err = Load(LoadOptions{Path: path, Explicit: true, Overrides: []configkit.Override{{
+		Path: configkit.Keys("defaults", "render_format"), Value: "json", Flag: "--render-format",
+	}}})
+	if err != nil || result.Config.Defaults.RenderFormat != "json" {
+		t.Fatalf("override did not precede validation: result=%+v err=%v", result, err)
+	}
+	if len(result.Sources()) != 2 || result.Sources()[0].Layer != configkit.LayerBase || result.Sources()[1].Layer != configkit.LayerCLI {
+		t.Fatalf("source order changed: %+v", result.Sources())
+	}
+	writeApplicationFile(t, path, "defaults: [\n")
+	if _, err := Load(LoadOptions{Path: path, Explicit: true, Overrides: []configkit.Override{{
+		Path: configkit.Keys("defaults", "render_format"), Value: "json", Flag: "--render-format",
+	}}}); err == nil {
+		t.Fatal("override bypassed malformed lower-level YAML")
+	}
+}
+
+func TestSettingsCopiesResolvedValuesAndSortsBackends(t *testing.T) {
+	result, err := loadDocumentBytes(`schema_dir: ""
+transient_retries: 0
+server:
+  addr: ""
+  artifact_root: ""
+  max_request_bytes: 0
+defaults:
+  render_format: json
+backends:
+  zed:
+    endpoint: http://zed.invalid/v1
+    extra_params:
+      count: 4
+  alpha:
+    queue_capacity: 0
+`)
+	if err != nil {
+		t.Fatal(err)
+	}
+	settings := Settings(result.Config)
+	if settings.SchemaDir != "" || settings.ServerAddr != "" || settings.MaxRequestBytes != 0 || settings.DefaultRenderFormat != "json" {
+		t.Fatalf("resolved scalar values changed: %+v", settings)
+	}
+	if settings.TransientRetries == nil || *settings.TransientRetries != 0 {
+		t.Fatalf("retry presence lost: %+v", settings.TransientRetries)
+	}
+	if len(settings.Backends) != 2 || settings.Backends[0].ID != "alpha" || settings.Backends[1].ID != "zed" {
+		t.Fatalf("backend order changed: %+v", settings.Backends)
+	}
+	if settings.Backends[0].QueueCapacity == nil || *settings.Backends[0].QueueCapacity != 0 || settings.Backends[1].QueueCapacity != nil {
+		t.Fatalf("queue presence changed: %+v", settings.Backends)
+	}
+	if value := settings.Backends[1].ExtraParams["count"]; value != int64(4) {
+		t.Fatalf("dynamic number changed shape: %T %v", value, value)
+	}
+
+	if got := Settings(documentDefaults()); got.TransientRetries != nil || got.Backends != nil {
+		t.Fatalf("default nullable values changed: %+v", got)
+	}
+}
+
+func TestLoadDiscoveryErrorQuotesPath(t *testing.T) {
+	dir := t.TempDir()
+	loop := filepath.Join(dir, "loop-\n.yml")
+	if err := os.Symlink(loop, loop); err != nil {
+		t.Skipf("symlink unavailable: %v", err)
+	}
+	useConfigSearchPaths(t, loop)
+	_, err := Load(LoadOptions{})
+	if err == nil || !strings.Contains(err.Error(), `\n`) || strings.Contains(err.Error(), "loop-\n.yml") {
+		t.Fatalf("discovery error did not quote path safely: %v", err)
+	}
+}
+
+func TestLoadComposesExplicitImportsAndSelectedOverlay(t *testing.T) {
+	dir := t.TempDir()
+	if err := os.MkdirAll(filepath.Join(dir, "fragments"), 0o700); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.MkdirAll(filepath.Join(dir, "overlays"), 0o700); err != nil {
+		t.Fatal(err)
+	}
+	base := filepath.Join(dir, "config.yml")
+	writeApplicationFile(t, base, "configkit:\n  imports:\n    - glob: fragments/*.yml\nprompt_dir: ./base-prompts\nserver:\n  addr: :8081\n")
+	writeApplicationFile(t, filepath.Join(dir, "fragments", "a.yml"), "backends:\n  alpha:\n    endpoint: http://alpha.invalid/v1\n")
+	writeApplicationFile(t, filepath.Join(dir, "fragments", "b.yml"), "backends:\n  beta:\n    endpoint: http://beta.invalid/v1\n")
+	writeApplicationFile(t, filepath.Join(dir, "overlays", "local.yml"), "prompt_dir: ./overlay-prompts\nserver:\n  addr: :9090\n")
+
+	result, err := Load(LoadOptions{Path: base, Explicit: true})
+	if err != nil || result.Config.PromptDir != "base-prompts" || len(result.Config.Backends) != 2 || len(result.Sources()) != 3 {
+		t.Fatalf("disjoint imports did not compose: result=%+v err=%v", result, err)
+	}
+	result, err = Load(LoadOptions{Path: base, Explicit: true, Overlay: "local", Overrides: []configkit.Override{{
+		Path: configkit.Keys("server", "addr"), Value: ":7070", Flag: "--addr",
+	}}})
+	if err != nil || result.Config.PromptDir != "overlay-prompts" || result.Config.Server.Addr != ":7070" || len(result.Sources()) != 5 {
+		t.Fatalf("overlay or explicit override changed: result=%+v err=%v", result, err)
+	}
+	if _, err := Load(LoadOptions{Path: base, Explicit: true, Overlay: "missing"}); err == nil {
+		t.Fatal("missing selected overlay accepted")
+	}
+	writeApplicationFile(t, filepath.Join(dir, "fragments", "b.yml"), "backends:\n  alpha:\n    endpoint: http://conflict.invalid/v1\n")
+	if _, err := Load(LoadOptions{Path: base, Explicit: true}); err == nil {
+		t.Fatal("conflicting import accepted")
+	} else {
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) {
+			t.Fatalf("conflict lost structured diagnostic: %v", err)
+		}
+	}
+}
+
+func TestLoadRejectsOverlayWithoutBase(t *testing.T) {
+	useConfigSearchPaths(t, filepath.Join(t.TempDir(), "absent.yml"))
+	if _, err := Load(LoadOptions{Overlay: "local"}); err == nil {
+		t.Fatal("overlay without base accepted")
+	} else {
+		var loadError *configkit.LoadError
+		if !errors.As(err, &loadError) {
+			t.Fatalf("no-base overlay lost structured diagnostic: %v", err)
+		}
+	}
+	if _, err := Load(LoadOptions{Overlay: "default"}); err != nil {
+		t.Fatalf("default must select no overlay: %v", err)
+	}
+}

internal/config/settings.go

@@ -0,0 +1,37 @@
+package config
+
+import renderformat "gitea.maximumdirect.net/eric/scriptorium/internal/format"
+
+const (
+	DefaultConfigPath      = "/etc/scriptorium/config.yml"
+	DefaultConfigPathLocal = "/usr/local/etc/scriptorium/config.yml"
+)
+
+var defaultConfigSearchPaths = []string{
+	DefaultConfigPathLocal,
+	DefaultConfigPath,
+}
+
+// AppSettings holds the validated application values used by executable adapters.
+type AppSettings struct {
+	PromptDir           string
+	ProfileDir          string
+	SchemaDir           string
+	ServerAddr          string
+	ArtifactRoot        string
+	MaxRequestBytes     int64
+	MaxArtifactBytes    int64
+	MaxResponseBytes    int64
+	DefaultRenderFormat renderformat.PreparedRunOutputFormat
+	Backends            []BackendSettings
+	TransientRetries    *int
+}
+
+type BackendSettings struct {
+	ID               string
+	Endpoint         string
+	APIKeyEnv        string
+	ExtraParams      map[string]any
+	ConcurrencyLimit int
+	QueueCapacity    *int
+}

internal/report/report.go

@@ -0,0 +1,105 @@
+package report
+
+import "gitea.maximumdirect.net/eric/promptkit"
+
+// Execution is the terminal observation exposed by Scriptorium's opt-in adapters.
+type Execution struct {
+	RunID                string `json:"run_id,omitempty"`
+	PromptID             string `json:"prompt_id,omitempty"`
+	PromptVersion        string `json:"prompt_version,omitempty"`
+	ProfileID            string `json:"profile_id,omitempty"`
+	BackendID            string `json:"backend_id,omitempty"`
+	Model                string `json:"model,omitempty"`
+	Outcome              string `json:"outcome"`
+	Failure              string `json:"failure"`
+	StatusCode           int    `json:"status_code,omitempty"`
+	RetriesExhausted     bool   `json:"retries_exhausted"`
+	Attempts             uint64 `json:"attempts"`
+	TransientRetries     uint64 `json:"transient_retries"`
+	RepairCalls          uint64 `json:"repair_calls"`
+	DurationMS           int64  `json:"duration_ms"`
+	CapacityWaitMS       int64  `json:"capacity_wait_ms"`
+	GenerationDurationMS int64  `json:"generation_duration_ms"`
+	RetryDelayMS         int64  `json:"retry_delay_ms"`
+	Usage                Usage  `json:"usage"`
+}
+
+type Usage struct {
+	PromptTokens     TokenCount `json:"prompt_tokens"`
+	CompletionTokens TokenCount `json:"completion_tokens"`
+	TotalTokens      TokenCount `json:"total_tokens"`
+	CachedTokens     TokenCount `json:"cached_tokens"`
+	CacheWriteTokens TokenCount `json:"cache_write_tokens"`
+}
+
+type TokenCount struct {
+	Value    int  `json:"value"`
+	Known    bool `json:"known"`
+	Complete bool `json:"complete"`
+}
+
+// Collector retains at most one terminal observation from a synchronous run.
+type Collector struct {
+	report *Execution
+}
+
+func NewCollector() *Collector { return &Collector{} }
+
+func (collector *Collector) Observe(event promptkit.ExecutionEvent) {
+	if event.Kind != promptkit.ExecutionFinished {
+		return
+	}
+	count := func(value promptkit.ObservedTokenCount) TokenCount {
+		return TokenCount{Value: value.Value, Known: value.Known, Complete: value.Complete}
+	}
+	collector.report = &Execution{
+		RunID: event.RunID, PromptID: event.PromptID, PromptVersion: event.PromptVersion,
+		ProfileID: event.ProfileID, BackendID: event.BackendID, Model: event.Model,
+		Outcome: outcome(event.Outcome), Failure: failure(event.Failure),
+		StatusCode: event.StatusCode, RetriesExhausted: event.RetriesExhausted,
+		Attempts: event.Attempts, TransientRetries: event.TransientRetries, RepairCalls: event.RepairCalls,
+		DurationMS: event.Duration.Milliseconds(), CapacityWaitMS: event.CapacityWait.Milliseconds(),
+		GenerationDurationMS: event.GenerationDuration.Milliseconds(), RetryDelayMS: event.RetryDelay.Milliseconds(),
+		Usage: Usage{
+			PromptTokens: count(event.Usage.PromptTokens), CompletionTokens: count(event.Usage.CompletionTokens),
+			TotalTokens: count(event.Usage.TotalTokens), CachedTokens: count(event.Usage.CachedTokens),
+			CacheWriteTokens: count(event.Usage.CacheWriteTokens),
+		},
+	}
+}
+
+func (collector *Collector) Snapshot() *Execution {
+	if collector.report == nil {
+		return nil
+	}
+	value := *collector.report
+	return &value
+}
+
+func outcome(value promptkit.ExecutionOutcome) string {
+	switch value {
+	case promptkit.ExecutionOutcomeSuccess, promptkit.ExecutionOutcomeInvalidOutput, promptkit.ExecutionOutcomeError:
+		return string(value)
+	default:
+		return "other"
+	}
+}
+
+func failure(value promptkit.ExecutionFailureKind) string {
+	switch value {
+	case promptkit.ExecutionFailureNone:
+		return "none"
+	case promptkit.ExecutionFailureCanceled, promptkit.ExecutionFailureDeadline,
+		promptkit.ExecutionFailureTransport, promptkit.ExecutionFailureProviderStatus,
+		promptkit.ExecutionFailureMalformedResponse, promptkit.ExecutionFailureCapacity,
+		promptkit.ExecutionFailureInvalidConfig, promptkit.ExecutionFailureInvalidRequest,
+		promptkit.ExecutionFailurePromptNotFound, promptkit.ExecutionFailureProfileNotFound,
+		promptkit.ExecutionFailurePromptLoad, promptkit.ExecutionFailureProfileLoad,
+		promptkit.ExecutionFailureArtifactLoad, promptkit.ExecutionFailurePromptRender,
+		promptkit.ExecutionFailureValidation, promptkit.ExecutionFailureGeneration,
+		promptkit.ExecutionFailureOther:
+		return string(value)
+	default:
+		return "other"
+	}
+}

internal/report/report_test.go

@@ -0,0 +1,126 @@
+package report
+
+import (
+	"context"
+	"encoding/json"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+	"time"
+
+	"gitea.maximumdirect.net/eric/promptkit"
+)
+
+func TestCollectorProjectsOnlyTerminalAggregates(t *testing.T) {
+	collector := NewCollector()
+	if collector.Snapshot() != nil {
+		t.Fatal("empty collector produced a report")
+	}
+	collector.Observe(promptkit.ExecutionEvent{Kind: promptkit.ExecutionAttempt, Attempts: 99})
+	if collector.Snapshot() != nil {
+		t.Fatal("nonterminal event produced a report")
+	}
+	collector.Observe(promptkit.ExecutionEvent{
+		Kind: promptkit.ExecutionFinished, RunID: "run-1", PromptID: "prompt-1",
+		PromptVersion: "1", ProfileID: "profile-1", BackendID: "backend-1", Model: "model-1",
+		Outcome: promptkit.ExecutionOutcomeError, Failure: promptkit.ExecutionFailureProviderStatus,
+		StatusCode: 503, RetriesExhausted: true, Attempts: 3, TransientRetries: 2, RepairCalls: 1,
+		Duration: 1500 * time.Millisecond, CapacityWait: 20 * time.Millisecond,
+		GenerationDuration: 300 * time.Millisecond, RetryDelay: 700 * time.Millisecond,
+		Usage: promptkit.ObservedTokenUsage{
+			PromptTokens:     promptkit.ObservedTokenCount{Value: 0, Known: true, Complete: true},
+			CompletionTokens: promptkit.ObservedTokenCount{Value: 0, Known: false, Complete: false},
+			TotalTokens:      promptkit.ObservedTokenCount{Value: 11, Known: true, Complete: false},
+		},
+	})
+	collector.Observe(promptkit.ExecutionEvent{Kind: promptkit.ExecutionRetryDelay, Attempts: 50})
+	report := collector.Snapshot()
+	if report == nil || report.Attempts != 3 || report.TransientRetries != 2 || report.RepairCalls != 1 || report.StatusCode != 503 || !report.RetriesExhausted {
+		t.Fatalf("terminal aggregates changed: %+v", report)
+	}
+	if report.DurationMS != 1500 || report.CapacityWaitMS != 20 || report.GenerationDurationMS != 300 || report.RetryDelayMS != 700 {
+		t.Fatalf("terminal timings changed: %+v", report)
+	}
+	if report.Usage.PromptTokens != (TokenCount{Known: true, Complete: true}) || report.Usage.CompletionTokens != (TokenCount{}) || report.Usage.TotalTokens != (TokenCount{Value: 11, Known: true}) {
+		t.Fatalf("presence-aware usage changed: %+v", report.Usage)
+	}
+	encoded, err := json.Marshal(report)
+	if err != nil {
+		t.Fatal(err)
+	}
+	var fields map[string]any
+	if err := json.Unmarshal(encoded, &fields); err != nil {
+		t.Fatal(err)
+	}
+	for _, key := range []string{"run_id", "prompt_id", "prompt_version", "profile_id", "backend_id", "model", "outcome", "failure", "status_code", "retries_exhausted", "attempts", "transient_retries", "repair_calls", "duration_ms", "capacity_wait_ms", "generation_duration_ms", "retry_delay_ms", "usage"} {
+		if _, ok := fields[key]; !ok {
+			t.Fatalf("report missing %q: %s", key, encoded)
+		}
+	}
+	for _, key := range []string{"prompt_tokens", "completion_tokens", "total_tokens", "cached_tokens", "cache_write_tokens"} {
+		item := fields["usage"].(map[string]any)[key].(map[string]any)
+		if len(item) != 3 {
+			t.Fatalf("incomplete token count %q: %#v", key, item)
+		}
+	}
+	for _, forbidden := range []string{"session_id", "headers", "endpoint", "api_key", "error", "output", "input"} {
+		if _, present := fields[forbidden]; present {
+			t.Fatalf("report exposed %q: %s", forbidden, encoded)
+		}
+	}
+}
+
+func TestCollectorHandlesUnknownValuesAndKeepsInvocationsSeparate(t *testing.T) {
+	first := NewCollector()
+	second := NewCollector()
+	first.Observe(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, RunID: "first", Outcome: "future", Failure: "future"})
+	second.Observe(promptkit.ExecutionEvent{Kind: promptkit.ExecutionFinished, RunID: "second", Outcome: promptkit.ExecutionOutcomeSuccess})
+	one := first.Snapshot()
+	two := second.Snapshot()
+	if one.RunID != "first" || one.Outcome != "other" || one.Failure != "other" || two.RunID != "second" || two.Failure != "none" {
+		t.Fatalf("collector isolation or classification failed: first=%+v second=%+v", one, two)
+	}
+	one.RunID = "changed"
+	if first.Snapshot().RunID != "first" {
+		t.Fatal("snapshot mutation changed stored report")
+	}
+	encoded, err := json.Marshal(two)
+	if err != nil {
+		t.Fatal(err)
+	}
+	for _, absent := range []string{"prompt_id", "prompt_version", "profile_id", "backend_id", "model", "status_code"} {
+		if strings.Contains(string(encoded), `"`+absent+`"`) {
+			t.Fatalf("unavailable optional field %q was included: %s", absent, encoded)
+		}
+	}
+}
+
+type localClient struct{}
+
+func (localClient) Generate(ctx context.Context, req promptkit.GenerateRequest) (*promptkit.GenerateResponse, error) {
+	return &promptkit.GenerateResponse{Content: "ok"}, nil
+}
+
+func TestCollectorReceivesRealEngineTerminalEvent(t *testing.T) {
+	promptDir := t.TempDir()
+	profileDir := t.TempDir()
+	if err := os.WriteFile(filepath.Join(promptDir, "prompt.yaml"), []byte("id: p\nversion: \"1\"\ndefault_profile: local\nmessages:\n  - role: user\n    content: hi\noutput:\n  format: text\n  validation_mode: none\n"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.WriteFile(filepath.Join(profileDir, "profile.yaml"), []byte("id: local\nendpoint: http://127.0.0.1:1/v1\nmodel: local-model\n"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	engine, err := promptkit.NewEngine(promptkit.Config{PromptDir: promptDir, ProfileDir: profileDir}, promptkit.WithLLMClient(localClient{}))
+	if err != nil {
+		t.Fatal(err)
+	}
+	collector := NewCollector()
+	if _, err := engine.Run(context.Background(), promptkit.RunRequest{PromptID: "p", Observer: collector.Observe}); err != nil {
+		t.Fatal(err)
+	}
+	value := collector.Snapshot()
+	if value == nil || value.PromptID != "p" || value.ProfileID != "local" || value.Outcome != "success" || value.Attempts != 1 {
+		t.Fatalf("unexpected terminal event projection: %+v", value)
+	}
+}

scripts/release.sh

@@ -0,0 +1,122 @@
+#!/bin/sh
+# Publish one validated source release; binary delivery is asynchronous.
+set -eu
+
+die() {
+    printf '%s\n' "release: $*" >&2
+    exit 1
+}
+
+[ "$#" -eq 1 ] || die 'usage: sh scripts/release.sh vMAJOR.MINOR.PATCH'
+version=$1
+case "$version" in
+    *'
+'*) die 'invalid stable release version' ;;
+esac
+printf '%s\n' "$version" |
+    grep -Eq '^v(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)\.(0|[1-9][0-9]*)$' ||
+    die 'invalid stable release version'
+
+cd "$(git rev-parse --show-toplevel)"
+candidate=$(git rev-parse --verify 'HEAD^{commit}')
+tag_ref="refs/tags/$version"
+notes="docs/releases/$version.md"
+
+# Inspect and publish to the same endpoint, including when origin has a
+# separate push URL. A raw URL avoids configured mirror/refspec behavior.
+remote=$(git remote get-url --push --all origin)
+case "$remote" in
+    ''|*'
+'*) die 'origin must have exactly one publication URL' ;;
+esac
+
+check_candidate() {
+    [ "$(git branch --show-current)" = main ] || die 'candidate must be on main'
+    [ -z "$(git status --porcelain --untracked-files=all)" ] ||
+        die 'worktree and index must be clean'
+    [ "$(git rev-parse --verify 'HEAD^{commit}')" = "$candidate" ] ||
+        die 'candidate changed during validation; validate the new candidate'
+    [ "$(git remote get-url --push --all origin)" = "$remote" ] ||
+        die 'origin publication URL changed during validation'
+}
+
+read_remote() {
+    remote_state=$(git ls-remote --refs "$remote" refs/heads/main "$tag_ref") ||
+        return 1
+    remote_main=$(printf '%s\n' "$remote_state" |
+        awk '$2 == "refs/heads/main" { print $1 }')
+    remote_tag=$(printf '%s\n' "$remote_state" |
+        awk -v ref="$tag_ref" '$2 == ref { print $1 }')
+}
+
+check_publication() {
+    check_candidate
+    if git show-ref --verify --quiet "$tag_ref"; then
+        die 'version already exists locally; see docs/release.md for recovery'
+    fi
+    read_remote || die 'cannot establish authoritative publication state'
+    [ "$remote_main" = "$candidate" ] ||
+        die 'candidate must exactly match main on origin; publish main first'
+    [ -z "$remote_tag" ] || die 'version already exists on origin'
+}
+
+check_publication
+[ -z "$(git ls-tree -r --name-only "$candidate" -- go.work go.work.sum vendor)" ] ||
+    die 'candidate must not contain a workspace or vendor tree'
+if grep -Eq '^[[:space:]]*replace([[:space:]]|\()' go.mod; then
+    die 'candidate must use published dependencies without go.mod replacements'
+fi
+git cat-file -e "$candidate:$notes" 2>/dev/null ||
+    die 'matching release note is missing from the candidate'
+
+release_tmp=$(mktemp -d "${TMPDIR:-/tmp}/scriptorium-release.XXXXXX")
+trap 'rm -rf -- "$release_tmp"' EXIT
+trap 'exit 1' HUP INT TERM
+git show "$candidate:$notes" >"$release_tmp/notes.md"
+awk -v version="$version" '
+    NR == 1 {
+        for (i = 2; i <= NF; i++) {
+            if ($i == "Scriptorium") product = 1
+            if ($i == version) matched = 1
+        }
+        valid = ($1 == "#" && product && matched)
+    }
+    END { exit !valid }
+' "$release_tmp/notes.md" || die 'release note title must contain Scriptorium and the exact version'
+
+# Dependencies and the toolchain are provisioned before invocation. Do not
+# discover a workspace, use vendor mode, modify module metadata, or download
+# modules/toolchains during candidate validation.
+export GOWORK=off GOFLAGS=-mod=readonly GOPROXY=off GONOPROXY=none GOSUMDB=off GOTOOLCHAIN=local
+host_os=$(go env GOHOSTOS)
+host_arch=$(go env GOHOSTARCH)
+GOOS="$host_os" GOARCH="$host_arch" CGO_ENABLED=1 go test -race ./...
+for arch in amd64 arm64; do
+    GOOS=linux GOARCH="$arch" CGO_ENABLED=0 go build -trimpath -ldflags '-s -w' \
+        -o "$release_tmp/scriptorium-$version-linux-$arch" ./cmd/scriptorium
+done
+
+check_publication
+git tag --annotate --cleanup=verbatim \
+    --file "$release_tmp/notes.md" "$version" "$candidate"
+tag_object=$(git rev-parse --verify "$tag_ref")
+[ "$(git cat-file -t "$tag_object")" = tag ] || die 'expected an annotated tag'
+[ "$(git rev-parse --verify "$tag_object^{commit}")" = "$candidate" ] ||
+    die 'tag does not identify the validated candidate'
+check_candidate
+
+# Freeze the source object and name only the intended destination ref. Disable
+# automatic following of other annotated tags even if enabled in user config.
+if git -c push.followTags=false push "$remote" "$tag_object:$tag_ref"; then
+    : # Successful pushes do not need another network read-back gate.
+else
+    printf '%s\n' 'release: push failed; resolving authoritative remote state' >&2
+    read_remote || die 'publication status unknown; preserve the local tag and inspect origin before retrying'
+    if [ "$remote_tag" != "$tag_object" ] || [ "$remote_main" != "$candidate" ]; then
+        die 'publication not confirmed; preserve the local tag and follow docs/release.md recovery'
+    fi
+    printf '%s\n' 'release: origin confirms the intended tag despite the push error'
+fi
+
+printf 'Source release %s published: commit %s, tag %s\n' "$version" "$candidate" "$tag_object"
+printf '%s\n' 'Binary delivery: asynchronous, not verified.'

scripts/release_test.go

@@ -0,0 +1,357 @@
+package scripts_test
+
+import (
+	"context"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"runtime"
+	"strings"
+	"testing"
+	"time"
+)
+
+const version = "v0.14.0"
+
+// Git operations use real temporary repositories. Go is the expensive boundary:
+// record validation commands rather than recursively running this test suite.
+const fakeGo = `#!/bin/sh
+set -eu
+if [ "$1" = env ]; then
+    case "$2" in
+        GOHOSTOS) printf 'linux\n' ;;
+        GOHOSTARCH) printf 'amd64\n' ;;
+        *) exit 1 ;;
+    esac
+    exit 0
+fi
+[ "$GONOPROXY" = none ]
+printf '%s|%s|%s|%s|%s|%s|%s|%s|%s\n' "$*" "$GOWORK" "$GOFLAGS" "$GOPROXY" "$GOSUMDB" "$GOTOOLCHAIN" "$GOOS" "$GOARCH" "$CGO_ENABLED" >> "$RELEASE_TEST_LOG"
+if [ "$1" = test ] && [ -n "${RELEASE_TEST_MUTATION:-}" ]; then
+    sh "$RELEASE_TEST_MUTATION"
+fi
+case "${RELEASE_TEST_FAIL:-}" in
+    test) [ "$1" != test ] || exit 1 ;;
+    arm64) [ "$GOARCH" != arm64 ] || exit 1 ;;
+esac
+`
+
+// Simulate the transport losing the reply after a real local push, rejecting
+// the push, or becoming unavailable while recovery queries the remote.
+const gitTransport = `#!/bin/sh
+set -eu
+if [ "$1" = ls-remote ] && [ -f "$RELEASE_TEST_PUSH_MARKER" ]; then
+    if [ "$RELEASE_TEST_PUSH" = unknown ] || [ "$RELEASE_TEST_PUSH" = confirmed ]; then
+        exit 1
+    fi
+fi
+if [ "$1" = -c ] && [ "$3" = push ]; then
+    if [ "$RELEASE_TEST_PUSH" = rejected ]; then exit 1; fi
+    if [ "$RELEASE_TEST_PUSH" = conflict ]; then
+        "$RELEASE_TEST_GIT" push "$4" HEAD:refs/tags/v0.14.0
+    else
+        "$RELEASE_TEST_GIT" "$@"
+    fi
+    touch "$RELEASE_TEST_PUSH_MARKER"
+    if [ "$RELEASE_TEST_PUSH" = confirmed ]; then exit 0; fi
+    exit 1
+fi
+exec "$RELEASE_TEST_GIT" "$@"
+`
+
+type fixture struct {
+	t         *testing.T
+	root      string
+	repo      string
+	remote    string
+	bin       string
+	gitPath   string
+	log       string
+	env       []string
+	candidate string
+}
+
+func newFixture(t *testing.T) *fixture {
+	t.Helper()
+	if runtime.GOOS == "windows" {
+		t.Skip("release command requires a POSIX shell")
+	}
+	gitPath, err := exec.LookPath("git")
+	if err != nil {
+		t.Fatal("release tests require git:", err)
+	}
+	root := t.TempDir()
+	f := &fixture{
+		t: t, root: root, repo: filepath.Join(root, "work"),
+		remote: filepath.Join(root, "origin.git"), bin: filepath.Join(root, "bin"),
+		gitPath: gitPath, log: filepath.Join(root, "validation.log"),
+	}
+	for _, env := range os.Environ() {
+		key, _, _ := strings.Cut(env, "=")
+		if !strings.HasPrefix(key, "GIT_") && !strings.HasPrefix(key, "RELEASE_TEST_") && key != "PATH" {
+			f.env = append(f.env, env)
+		}
+	}
+	f.env = append(f.env,
+		"PATH="+f.bin+string(os.PathListSeparator)+os.Getenv("PATH"),
+		"GIT_CONFIG_NOSYSTEM=1", "GIT_CONFIG_GLOBAL="+os.DevNull,
+		"GIT_ALLOW_PROTOCOL=file", "GIT_TERMINAL_PROMPT=0",
+		"RELEASE_TEST_LOG="+f.log, "RELEASE_TEST_GIT="+gitPath,
+		"RELEASE_TEST_PUSH_MARKER="+filepath.Join(root, "pushed"),
+	)
+	f.run(root, gitPath, "init", "--bare", "--initial-branch=main", f.remote)
+	f.run(root, gitPath, "init", "--initial-branch=main", f.repo)
+	f.git("config", "user.name", "Release Test")
+	f.git("config", "user.email", "release@example.invalid")
+	f.git("remote", "add", "origin", f.remote)
+	f.write(filepath.Join(f.bin, "go"), fakeGo, 0o755)
+	script, err := os.ReadFile("release.sh")
+	if err != nil {
+		t.Fatal(err)
+	}
+	f.write(filepath.Join(f.repo, "scripts", "release.sh"), string(script), 0o644)
+	f.write(filepath.Join(f.repo, "docs", "releases", version+".md"), "# Scriptorium "+version+"\n\nTest release.\n", 0o644)
+	f.write(filepath.Join(f.repo, ".gitignore"), "ignored-output\n", 0o644)
+	f.write(filepath.Join(f.repo, "go.mod"), "module release.test/example\n\ngo 1.25.5\n", 0o644)
+	f.commit()
+	f.candidate = strings.TrimSpace(f.git("rev-parse", "HEAD"))
+	return f
+}
+
+func (f *fixture) write(path, content string, mode os.FileMode) {
+	f.t.Helper()
+	if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
+		f.t.Fatal(err)
+	}
+	if err := os.WriteFile(path, []byte(content), mode); err != nil {
+		f.t.Fatal(err)
+	}
+}
+
+func (f *fixture) run(dir, name string, args ...string) string {
+	f.t.Helper()
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer cancel()
+	cmd := exec.CommandContext(ctx, name, args...)
+	cmd.Dir, cmd.Env = dir, f.env
+	out, err := cmd.CombinedOutput()
+	if err != nil {
+		f.t.Fatalf("%s %v: %v\n%s", name, args, err, out)
+	}
+	return string(out)
+}
+
+func (f *fixture) git(args ...string) string {
+	f.t.Helper()
+	return f.run(f.repo, f.gitPath, args...)
+}
+
+func (f *fixture) commit() {
+	f.t.Helper()
+	f.git("add", ".")
+	f.git("commit", "-m", "Prepare test candidate")
+	f.git("push", "origin", "refs/heads/main:refs/heads/main")
+}
+
+func (f *fixture) publish(wantSuccess bool, args ...string) string {
+	f.t.Helper()
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer cancel()
+	cmd := exec.CommandContext(ctx, "sh", append([]string{"scripts/release.sh"}, args...)...)
+	cmd.Dir, cmd.Env = f.repo, f.env
+	out, err := cmd.CombinedOutput()
+	if (err == nil) != wantSuccess {
+		f.t.Fatalf("release success=%v, want %v: %v\n%s", err == nil, wantSuccess, err, out)
+	}
+	return string(out)
+}
+
+func (f *fixture) noPublishedTag() {
+	f.t.Helper()
+	if out := f.git("ls-remote", "--tags", f.remote, "refs/tags/"+version); strings.TrimSpace(out) != "" {
+		f.t.Fatalf("unexpected published tag: %s", out)
+	}
+}
+
+func TestReleasePublishesOnlyValidatedTag(t *testing.T) {
+	f := newFixture(t)
+	f.git("tag", "-a", "unrelated", "-m", "Do not publish this")
+	f.git("config", "push.followTags", "true")
+	f.git("config", "remote.origin.mirror", "true")
+	f.write(filepath.Join(f.repo, "ignored-output"), "local build", 0o644)
+	out := f.publish(true, version)
+	if !strings.Contains(out, "Source release "+version+" published") || !strings.Contains(out, "asynchronous, not verified") {
+		t.Fatalf("missing source/delivery status: %s", out)
+	}
+	tag := strings.TrimSpace(f.git("rev-parse", "refs/tags/"+version))
+	if got := strings.TrimSpace(f.git("cat-file", "-t", tag)); got != "tag" {
+		t.Fatalf("object type = %s", got)
+	}
+	if got := strings.TrimSpace(f.git("rev-parse", tag+"^{commit}")); got != f.candidate {
+		t.Fatalf("tag commit = %s, want %s", got, f.candidate)
+	}
+	note, err := os.ReadFile(filepath.Join(f.repo, "docs", "releases", version+".md"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if object := f.git("cat-file", "-p", tag); !strings.HasSuffix(object, "\n\n"+string(note)) {
+		t.Fatalf("tag message differs from committed note: %s", object)
+	}
+	refs := f.git("ls-remote", "--refs", f.remote)
+	if strings.Count(strings.TrimSpace(refs), "\n") != 1 || !strings.Contains(refs, tag+"\trefs/tags/"+version) || !strings.Contains(refs, f.candidate+"\trefs/heads/main") {
+		t.Fatalf("unexpected published refs: %s", refs)
+	}
+	log, err := os.ReadFile(f.log)
+	if err != nil {
+		t.Fatal(err)
+	}
+	lines := strings.Split(strings.TrimSpace(string(log)), "\n")
+	if len(lines) != 3 || lines[0] != "test -race ./...|off|-mod=readonly|off|off|local|linux|amd64|1" {
+		t.Fatalf("unexpected validation calls: %s", log)
+	}
+	for i, arch := range []string{"amd64", "arm64"} {
+		if !strings.HasPrefix(lines[i+1], "build -trimpath -ldflags -s -w -o ") ||
+			!strings.Contains(lines[i+1], "scriptorium-"+version+"-linux-"+arch+" ./cmd/scriptorium|") ||
+			!strings.HasSuffix(lines[i+1], "|off|-mod=readonly|off|off|local|linux|"+arch+"|0") {
+			t.Fatalf("unexpected build: %s", lines[i+1])
+		}
+	}
+	if got := f.git("status", "--porcelain"); got != "" {
+		t.Fatalf("release dirtied worktree: %s", got)
+	}
+}
+
+func TestReleaseRejectsInvalidCandidates(t *testing.T) {
+	tests := []struct {
+		name string
+		args []string
+		edit func(*fixture)
+	}{
+		{name: "no version"},
+		{name: "leading zero", args: []string{"v00.14.0"}},
+		{name: "prerelease", args: []string{"v0.14.0-rc.1"}},
+		{name: "multiline version", args: []string{"v0.14.0\ninvalid"}},
+		{name: "local tag", edit: func(f *fixture) { f.git("tag", version) }},
+		{name: "remote tag", edit: func(f *fixture) {
+			f.run(f.root, f.gitPath, "--git-dir="+f.remote, "update-ref", "refs/tags/"+version, f.candidate)
+		}},
+		{name: "dirty", edit: func(f *fixture) { f.write(filepath.Join(f.repo, ".gitignore"), "changed\n", 0o644) }},
+		{name: "untracked", edit: func(f *fixture) { f.write(filepath.Join(f.repo, "untracked"), "new\n", 0o644) }},
+		{name: "staged", edit: func(f *fixture) {
+			f.write(filepath.Join(f.repo, ".gitignore"), "changed\n", 0o644)
+			f.git("add", ".gitignore")
+		}},
+		{name: "wrong branch", edit: func(f *fixture) { f.git("switch", "-c", "topic") }},
+		{name: "unpublished main", edit: func(f *fixture) { f.git("commit", "--allow-empty", "-m", "Unpublished") }},
+		{name: "missing note", edit: func(f *fixture) { f.git("rm", "docs/releases/"+version+".md"); f.commit() }},
+		{name: "mismatched title", edit: func(f *fixture) {
+			f.write(filepath.Join(f.repo, "docs/releases/"+version+".md"), "# Scriptorium v0.14.1\n", 0o644)
+			f.commit()
+		}},
+		{name: "empty note", edit: func(f *fixture) {
+			f.write(filepath.Join(f.repo, "docs/releases/"+version+".md"), "", 0o644)
+			f.commit()
+		}},
+		{name: "committed workspace", edit: func(f *fixture) {
+			f.write(filepath.Join(f.repo, "go.work"), "go 1.25.5\nuse .\n", 0o644)
+			f.commit()
+		}},
+		{name: "committed vendor tree", edit: func(f *fixture) {
+			f.write(filepath.Join(f.repo, "vendor", "modules.txt"), "vendored dependency\n", 0o644)
+			f.commit()
+		}},
+		{name: "module replacement", edit: func(f *fixture) {
+			f.write(filepath.Join(f.repo, "go.mod"), "module release.test/example\n\ngo 1.25.5\nreplace example.org/dependency => ../dependency\n", 0o644)
+			f.commit()
+		}},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			f := newFixture(t)
+			if tt.edit != nil {
+				tt.edit(f)
+			}
+			args := tt.args
+			if args == nil && tt.name != "no version" {
+				args = []string{version}
+			}
+			f.publish(false, args...)
+			if _, err := os.Stat(f.log); !os.IsNotExist(err) {
+				t.Fatalf("invalid candidate reached expensive validation: %v", err)
+			}
+			if tt.name != "remote tag" {
+				f.noPublishedTag()
+			}
+		})
+	}
+}
+
+func TestReleaseStopsAfterFailedValidation(t *testing.T) {
+	for _, step := range []string{"test", "arm64"} {
+		t.Run(step, func(t *testing.T) {
+			f := newFixture(t)
+			f.env = append(f.env, "RELEASE_TEST_FAIL="+step)
+			f.publish(false, version)
+			f.noPublishedTag()
+			if tags := f.git("tag", "--list", version); tags != "" {
+				t.Fatalf("failed validation created tag: %s", tags)
+			}
+		})
+	}
+}
+
+func TestReleaseRechecksCandidateAndRemote(t *testing.T) {
+	for name, mutation := range map[string]string{
+		"candidate changed":   "git commit --allow-empty -m changed\ngit push origin main\n",
+		"worktree changed":    "touch unexpected\n",
+		"remote main changed": "git --git-dir=\"$RELEASE_TEST_REMOTE\" update-ref -d refs/heads/main\n",
+		"remote tag appeared": "git push origin HEAD:refs/tags/v0.14.0\n",
+	} {
+		t.Run(name, func(t *testing.T) {
+			f := newFixture(t)
+			path := filepath.Join(f.root, "mutate.sh")
+			f.write(path, "set -eu\n"+mutation, 0o644)
+			f.env = append(f.env, "RELEASE_TEST_MUTATION="+path, "RELEASE_TEST_REMOTE="+f.remote)
+			f.publish(false, version)
+			if tags := f.git("tag", "--list", version); tags != "" {
+				t.Fatalf("changed candidate created tag: %s", tags)
+			}
+			if name != "remote tag appeared" {
+				f.noPublishedTag()
+			}
+		})
+	}
+}
+
+func TestReleaseHandlesPushOutcomes(t *testing.T) {
+	for _, mode := range []string{"confirmed", "lost reply", "rejected", "unknown", "conflict"} {
+		t.Run(mode, func(t *testing.T) {
+			f := newFixture(t)
+			f.write(filepath.Join(f.bin, "git"), gitTransport, 0o755)
+			f.env = append(f.env, "RELEASE_TEST_PUSH="+mode)
+			out := f.publish(mode == "confirmed" || mode == "lost reply", version)
+			if mode == "unknown" && !strings.Contains(out, "publication status unknown") {
+				t.Fatalf("did not report uncertainty: %s", out)
+			}
+			if tag := f.git("tag", "--list", version); strings.TrimSpace(tag) != version {
+				t.Fatalf("local tag was not preserved: %s", tag)
+			}
+			if mode == "rejected" {
+				f.noPublishedTag()
+			}
+			if mode == "confirmed" || mode == "lost reply" || mode == "unknown" {
+				tag := strings.TrimSpace(f.git("rev-parse", "refs/tags/"+version))
+				if remote := f.git("ls-remote", "--refs", f.remote, "refs/tags/"+version); !strings.HasPrefix(remote, tag+"\t") {
+					t.Fatalf("published tag mismatch: %s", remote)
+				}
+			}
+		})
+	}
+}
+
+func TestReleaseUsesOriginPushEndpoint(t *testing.T) {
+	f := newFixture(t)
+	f.git("remote", "set-url", "--push", "origin", f.remote)
+	f.git("remote", "set-url", "origin", filepath.Join(f.root, "unavailable-fetch.git"))
+	f.publish(true, version)
+}