Improve D&D validation reliability

assets/dnd/combat-turns/schemas/dnd_combat_turns_llm.v1.json

@@ -16,7 +16,8 @@
             "type": "string"
           },
           "turn_kind": {
-            "type": "string"
+            "type": "string",
+            "enum": ["turn", "reaction", "legendary_action", "lair_action", "other"]
           },
           "source_refs": {
             "type": "array",

assets/dnd/enemy-events/schemas/dnd_enemy_events_llm.v1.json

@@ -13,7 +13,10 @@
         "required": ["name", "kind", "source_refs"],
         "properties": {
           "name": {"type": "string"},
-          "kind": {"type": "string"},
+          "kind": {
+            "type": "string",
+            "enum": ["engaged", "killed", "fled", "captured", "incapacitated"]
+          },
           "source_refs": {
             "type": "array",
             "items": {

assets/dnd/item-occurrences/prompts/instructions.md

@@ -18,12 +18,20 @@ when the transcript explicitly describes it being physically destroyed or
 expended as a non-payment component. Use `transferred` only when possession
 moves between two distinct named party members.
 
-Return both `from` and `to` for every occurrence, using `null` when a holder does not
-apply. For `discovered`, set both holders to `null`. For `acquired`, set `from`
-to `null` and provide `to`; for `lost` and `consumed`, provide `from` and set
-`to` to `null`; and for `transferred`, provide both holders. Use `party` only
-for collective or unresolved party possession, never for either side of a
-transfer. Do not emit a transfer for a gift, sale, or payment outside the party.
+Return both `from` and `to` for every occurrence. Use JSON `null`, not an empty
+string, whenever a holder does not apply. Follow this holder matrix exactly:
+
+| `kind` | required `from` | required `to` |
+| --- | --- | --- |
+| `discovered` | `null` | `null` |
+| `acquired` | `null` | `party` or the named party member gaining possession |
+| `lost` | `party` or the named party member losing possession | `null` |
+| `consumed` | `party` or the named party member consuming the item | `null` |
+| `transferred` | one named party member | a different named party member |
+
+Use `party` only for collective or unresolved party possession, never for
+either side of a transfer. Do not emit a transfer for a gift, sale, or payment
+outside the party.
 
 Ordinary non-depleting use is not an occurrence. Do not infer acquisition from a
 discovery, or discovery from an acquisition: emit both only when each is

assets/dnd/item-occurrences/schemas/dnd_item_occurrences_llm.v1.json

@@ -13,7 +13,10 @@
         "required": ["name", "kind", "quantity", "from", "to", "source_refs"],
         "properties": {
           "name": {"type": "string"},
-          "kind": {"type": "string"},
+          "kind": {
+            "type": "string",
+            "enum": ["discovered", "acquired", "lost", "consumed", "transferred"]
+          },
           "quantity": {"type": ["integer", "null"]},
           "from": {"type": ["string", "null"]},
           "to": {"type": ["string", "null"]},

assets/dnd/npc-occurrences/schemas/dnd_npc_occurrences_llm.v1.json

@@ -16,7 +16,8 @@
             "type": "string"
           },
           "kind": {
-            "type": "string"
+            "type": "string",
+            "enum": ["mentioned", "noncombat_presence", "dialogue", "combat_ally", "combat_opponent", "other"]
           },
           "source_refs": {
             "type": "array",

assets/dnd/scene-descriptions/prompts/instructions.md

@@ -5,9 +5,7 @@ into multiple scenes or use facts that are not supported by it.
 Return one kind, one concise title, and one concise summary. Choose exactly one
 kind:
 
-- combat: active combat materially organizes the scene, including
-  initiative-like exchanges or sustained hostile action. Planning a fight or
-  discussing a completed fight is not combat by itself.
+- combat: a scene classified as combat under the shared combat policy.
 - narrative: current-session in-world play that is not principally active
   combat, a prior-session recap, or sustained out-of-character session
   discussion. This includes exploration, travel, dialogue, investigation,
@@ -20,15 +18,13 @@ kind:
   play.
 
 Narrative is the default for actual current-session gameplay that does not meet
-another definition. When the accepted chunk is mixed:
+another definition. When the accepted chunk has no substantive active combat:
 
-1. use combat when active combat is a substantive central activity, even with
-   brief setup, rules clarification, or immediate aftermath;
-2. otherwise use recap when recounting a previous session is the chunk's
+1. use recap when recounting a previous session is the chunk's
    primary table purpose;
-3. otherwise use meta when sustained out-of-character session discussion is
+2. otherwise use meta when sustained out-of-character session discussion is
    primary and in-world progression is no more than incidental; and
-4. use narrative for all remaining current-session in-world play.
+3. use narrative for all remaining current-session in-world play.
 
 Brief table talk, dice resolution, rules clarification, jokes, or
 administrative comments do not make a gameplay scene meta. A short recollection

assets/dnd/scene-descriptions/prompts/prompt.yaml

@@ -27,6 +27,8 @@ messages:
     content_file: ./sharedassets/common-dnd-transcript-chunk.md
     cache_control:
       type: ephemeral
+  - role: user
+    content_file: ./sharedassets/common-dnd-scene-combat-policy.md
   - role: user
     content_file: ./instructions.md
     cache_control:

assets/dnd/scene-descriptions/schemas/dnd_scene_descriptions_llm.v1.json

@@ -6,7 +6,8 @@
   "required": ["kind", "title", "summary"],
   "properties": {
     "kind": {
-      "type": "string"
+      "type": "string",
+      "enum": ["combat", "narrative", "recap", "meta"]
     },
     "title": {
       "type": "string"

assets/dnd/scene-descriptions/validate/combat-semantics/prompts/instructions.md

@@ -0,0 +1,11 @@
+Review only whether the proposed scene kind correctly represents substantive
+active combat in the supplied transcript chunk under the shared combat policy.
+Do not judge the scene title, summary, non-combat subtype, scene boundary, or
+any other aspect of a scene description.
+
+Return `approved` only when the proposed kind's combat status is supported;
+approval does not endorse any other part of the scene description. Return
+`combat_should_be_added` when a proposed non-combat kind omits substantive
+active combat. Return `combat_should_be_removed` when a proposed `combat` kind
+is unsupported. Give a concise, transcript-grounded explanation for every
+verdict.

assets/dnd/scene-descriptions/validate/combat-semantics/prompts/prompt.yaml

@@ -0,0 +1,30 @@
+id: dnd.scene_descriptions.validate_combat
+version: "v1"
+default_profile: dnd-extraction
+inputs:
+  - name: transcript
+    required: true
+    content_type: application/json
+  - name: proposed_kind
+    required: true
+    content_type: text/plain
+messages:
+  - role: system
+    content_file: ./sharedassets/common-dnd-system.md
+  - role: user
+    content_file: ./sharedassets/common-dnd-scene-combat-policy.md
+  - role: user
+    content_file: ./instructions.md
+  - role: user
+    content_file: ./proposed-kind.md
+    cache_control:
+      type: ephemeral
+  - role: user
+    content_file: ./sharedassets/common-dnd-transcript-chunk.md
+    cache_control:
+      type: ephemeral
+output:
+  format: json
+  validation_mode: json_schema
+  schema_path: dnd_scene_combat_semantics_llm.v1.json
+  repair_attempts: 1

assets/dnd/scene-descriptions/validate/combat-semantics/prompts/proposed-kind.md

@@ -0,0 +1,3 @@
+The proposed scene kind is:
+
+{{ input "proposed_kind" }}

assets/dnd/scene-descriptions/validate/combat-semantics/schemas/dnd_scene_combat_semantics_llm.v1.json

@@ -0,0 +1,18 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$id": "notarius.dnd.scene_descriptions.combat_semantics.llm",
+  "type": "object",
+  "additionalProperties": false,
+  "required": ["verdict", "explanation"],
+  "properties": {
+    "verdict": {
+      "type": "string",
+      "enum": ["approved", "combat_should_be_added", "combat_should_be_removed"]
+    },
+    "explanation": {
+      "type": "string",
+      "minLength": 1,
+      "maxLength": 512
+    }
+  }
+}

assets/dnd/shared/prompts/common-dnd-extraction-evidence.md

@@ -1,6 +1,8 @@
 Transcript units are the only evidence for extracted events and factual claims.
 Every reported factual claim must be supported by cited transcript units. Use
 integer `start_unit_id` and `end_unit_id` values from the transcript.
+Within each range, `start_unit_id` must identify the earlier transcript unit and
+`end_unit_id` the same or a later unit according to transcript order.
 
 When supporting evidence is non-contiguous, use multiple narrow ranges rather
 than a broad range that bridges unrelated conversation.

assets/dnd/shared/prompts/common-dnd-scene-combat-policy.md

@@ -0,0 +1,13 @@
+Use `combat` only when substantive active combat materially organizes the
+scene. Active combat includes initiative or turn exchanges, attacks, combat
+spells, damage, saves, movement, or similarly sustained hostile action.
+
+Do not use `combat` for planning or preparing for a possible fight; threats,
+hostile dialogue, or a tense confrontation; immediate aftermath, looting,
+healing, or discussion of a completed fight; a recap or in-world recollection
+of earlier combat; or out-of-character rules discussion without active
+encounter play.
+
+When a chunk contains substantive active combat alongside brief setup, rules
+clarification, interruption, phase transition, or immediate aftermath, classify
+it as `combat`.

docs/adr/0015-separate-process-warnings-from-quality-diagnostics.md

@@ -0,0 +1,67 @@
+# ADR-0015: Separate process warnings from quality diagnostics
+
+**Status:** Accepted
+**Date:** 2026-08-27
+
+## Context
+
+Notarius currently represents process degradation, incomplete validation,
+extraction-quality doubt, and routine normalization with one flat warning
+record. That makes ordinary successful runs noisy, loses the framework context
+needed to explain a finding, and gives `warning_count` no stable operational
+meaning. It also permits output encoders to add a warning after the durable
+warning file has already been written.
+
+The application needs one bounded diagnostic model that preserves exact
+occurrence counts while retaining only safe, representative samples. Fresh and
+resumed logical runs must present the same groups. The model must not alter
+validation decisions, retry budgets, rejected-output behavior, or process exit
+policy.
+
+## Decision
+
+Warnings are reserved for a completed run that advanced under an allowed
+process-level degradation or incomplete-work policy. Extraction-quality signals
+are advisories, and routine accepted transformations are observations. A
+non-degraded successful run therefore has zero actionable warnings.
+
+Modules and validators own a diagnostic's disposition, category, reason code,
+scope, and safe message. The framework adds pipeline origin, including stage,
+step, lane, module, validator, and chunk context where applicable. It then
+aggregates deterministically by disposition, category, reason code, and full
+origin. Chunk context remains on representative samples so equivalent findings
+across chunks aggregate together.
+
+Diagnostics carry exact occurrence counts, at most three distinct samples, and
+numeric omitted-sample metadata. Producers and validators are bounded to 64
+local groups. Final actionable warning groups are bounded without truncation;
+the non-warning collection may truncate represented groups while preserving an
+exact total occurrence count and explicit truncation metadata.
+
+The public contracts will be versioned: grouped actionable warnings use
+`notarius.warnings.v2`, grouped advisories and observations use
+`notarius.diagnostics.v1`, and the run receipt uses
+`notarius.run-result.v2`. Successful output encoders return logical files or
+an error; they do not add post-encoding warnings.
+
+## Alternatives considered
+
+- Keep one warning list and filter only CLI output. This would leave durable
+  consumers with the same semantically mixed, unbounded contract.
+- Map reason codes to severity in a central framework registry. This would
+  split module-owned meaning between synchronized policy tables and make new
+  diagnostic meaning implicit.
+- Preserve local omission warning records. They inflate visible group counts
+  and lose exact occurrence semantics.
+- Keep output-encoder warnings. A one-pass encoder cannot include those
+  records consistently in files it has already serialized; a two-phase encoder
+  protocol is deferred until a demonstrated need exists.
+
+## Consequences
+
+The framework gains validated diagnostic primitives, local collection,
+origin-aware aggregation, and versioned durable presentation. Existing warning
+transport remains temporarily while producers migrate. Current architecture,
+operator, integration, and internal documentation will describe the behavior
+only as each implementation step lands; this accepted decision does not claim
+that the migration is complete.

docs/adr/0016-feedback-aware-module-requested-retries.md

@@ -0,0 +1,76 @@
+# ADR-0016: Use feedback-aware module-requested retries
+
+**Status:** Accepted
+**Date:** 2026-08-28
+
+## Context
+
+An LLM-backed module can discover that a structurally valid model response is
+unsafe while translating it into a typed candidate, before the ordinary
+validator chain runs. Semantic registry reconciliation is the first such case:
+the shared core can identify invalid duplicate-group proposals, and typed
+application can reject a domain-incompatible group while preserving a safe
+partial result. Repeating the original request without the rejected response or
+corrective explanation gives the model no information with which to improve.
+
+The existing feedback-aware validation mechanism already establishes the safe
+correction protocol, but module-owned retry directives also carry internal
+reason codes, operator messages, and fallback diagnostics. Those values are not
+model instructions. Some module retry conditions, including exhausted
+structured-output repair, also have no valid exact response to append.
+
+## Decision
+
+An LLM-backed normalizer may attach optional, bounded model-facing correction
+guidance to a module-requested retry. Guidance is a separate contract field; the
+framework never derives it from a reason code, operator message, diagnostic, or
+error. A feedback-capable directive must include the exact model candidate that
+controlled the safe fallback and must use `single_response_v1`.
+
+The artifact-neutral producer-attempt state machine constructs the correction
+from that exact latest response and the supplied guidance. The producer rebuilds
+its complete ordinary request and appends the response as one assistant message
+and the guidance as one user message. Earlier attempts do not accumulate, and
+the attempt consumes the same configured stage retry budget as operational,
+structural, validation, and feedback-free module retries.
+
+A module retry without guidance remains valid and starts a fresh attempt. This
+is the required behavior when no valid model candidate exists, including after
+structured-output repair is exhausted. If feedback is supplied without a valid
+supported candidate, the framework fails the module contract rather than
+retrying blindly or inventing guidance.
+
+After retry exhaustion, the normalizer's safe fallback continues through its
+validator chain. Safe semantic groups may remain applied, unsafe groups remain
+separate, and bounded fallback diagnostics may explain the process degradation.
+Exact responses and correction text remain attempt-local and are excluded from
+ordinary errors, warnings, manifests, receipts, caches, and checkpoints.
+
+This decision extends, rather than supersedes,
+[ADR-0014](0014-feedback-aware-validation-retries.md): both correction sources
+use the same transport-neutral payload, replacement-request semantics, outer
+retry budget, and sensitive-data boundary.
+
+## Alternatives considered
+
+- Continue blind module retries. This preserves a smaller contract but wastes
+  the module's deterministic diagnosis and commonly repeats the same defect.
+- Convert module safety checks into validators. Typed reconciliation must apply
+  only safe proposal groups and retain a fallback before validation; moving
+  artifact-owned translation and application policy into validators would blur
+  stage ownership.
+- Copy the retry reason or operator message into the model request. Those values
+  are written for provenance and humans, can contain opaque internal labels,
+  and do not reliably describe a correct replacement.
+- Require feedback for every module retry. Structural failures may have no
+  valid exact candidate, so this would either prevent useful fresh retries or
+  fabricate prior-response material.
+
+## Consequences
+
+The normalize retry contract and generic producer-attempt directive gain an
+optional correction-guidance field and candidate-pairing validation. Modules
+that use it must provide semantically meaningful bounded prose and an exact
+candidate. Registry reconciliation maintains separate operator and model
+renderers, and policy fingerprints change so checkpoints created under blind
+retry behavior are not reused.

docs/cli.md

@@ -103,11 +103,14 @@ names, requiredness, and configured bindings are part of the
 Without **--json**, standard output contains the completed pipeline ID, counts
 of normalized and rejected outputs, and the output directory. A debug-enabled
 run also prints its debug-bundle path to standard output. A successful run with
-warnings reports the warning count to standard error. The published JSON bundle
+actionable process warnings reports their group and occurrence counts to
+standard error. When the selected output module publishes `warnings.json`, the
+summary also reports that durable file's path. Advisory and observation findings
+do not produce a warning line. The published JSON bundle
 is defined by the [JSON output contract](integrations/json-output.md).
 
 With **--json**, successful standard output is exactly one
-`notarius.run-result.v1` JSON document followed by a newline, with no
+`notarius.run-result.v2` JSON document followed by a newline, with no
 human-oriented status or debug-path line. Its fields and compatibility policy
 are defined by the [run-result contract](integrations/run-result.md). A caller
 must check for exit status 0 before decoding this output; a failed write can
@@ -171,7 +174,7 @@ go run ./cmd/notarius pipelines list \
 Successful commands write their primary result to standard output. Warnings and
 errors are written to standard error.
 
-For **run --json**, warnings remain on standard error and standard output is a
+For **run --json**, actionable process warnings remain on standard error and standard output is a
 machine-readable success result only. Syntax and runtime diagnostics remain on
 standard error. Parse the result only after the process exits with status 0.
 

docs/config.md

@@ -504,11 +504,28 @@ Available validator keys are:
 | Item occurrences | **extract/dnd/item-occurrences/shape**, **extract/dnd/item-occurrences/registry**, **extract/dnd/item-occurrences/source_refs**, **extract/dnd/item-occurrences/source_relatedness**, **normalize/dnd/item-occurrences/invariants** |
 | Item registry | **extract/dnd/item-registry/shape**, **extract/dnd/item-registry/source_refs**, **extract/dnd/item-registry/source_relatedness**, **normalize/dnd/item-registry/identity** |
 | NPC occurrences | **extract/dnd/npc-occurrences/shape**, **extract/dnd/npc-occurrences/registry**, **extract/dnd/npc-occurrences/source_refs**, **extract/dnd/npc-occurrences/source_relatedness**, **normalize/dnd/npc-occurrences/invariants** |
-| Scene descriptions | **extract/dnd/scene-descriptions/shape**, **extract/dnd/scene-descriptions/source_refs**, **extract/dnd/scene-descriptions/source_relatedness**, **normalize/dnd/scene-descriptions/invariants** |
+| Scene descriptions | **extract/dnd/scene-descriptions/shape**, **extract/dnd/scene-descriptions/source_refs**, **extract/dnd/scene-descriptions/source_relatedness**, **extract/dnd/scene-descriptions/combat_semantics** (LLM-backed, opt-in), **normalize/dnd/scene-descriptions/invariants** |
 | Enemy events | **extract/dnd/enemy-events/shape**, **extract/dnd/enemy-events/engagements**, **extract/dnd/enemy-events/source_refs**, **extract/dnd/enemy-events/source_relatedness**, **normalize/dnd/enemy-events/invariants** |
 | Location registry | **extract/dnd/location-registry/shape**, **extract/dnd/location-registry/source_refs**, **extract/dnd/location-registry/source_relatedness**, **normalize/dnd/location-registry/identity** |
 | Location occurrences | **extract/dnd/location-occurrences/shape**, **extract/dnd/location-occurrences/registry**, **extract/dnd/location-occurrences/source_refs**, **extract/dnd/location-occurrences/source_relatedness**, **normalize/dnd/location-occurrences/invariants** |
 
+`extract/dnd/scene-descriptions/combat_semantics` is not in a production default chain. To opt in, replace the scene extractor validator chain with the current ordered chain plus the semantic validator last, and set a positive producer retry budget if a rejection should request a corrected scene:
+
+~~~yaml
+extract:
+  module: dnd/scene-descriptions
+  retries: 1
+  validators:
+    - generic/valid_json
+    - extract/dnd/scene-descriptions/shape
+    - extract/dnd/scene-descriptions/source_refs
+    - generic/valid_json_schema
+    - extract/dnd/scene-descriptions/source_relatedness
+    - extract/dnd/scene-descriptions/combat_semantics
+~~~
+
+An override replaces, rather than extends, the default chain. See [Module Bindings And Validators](#module-bindings-and-validators) for binding, profile, repair, retry, and failure-policy rules.
+
 When no override is configured, production D&D bindings use the following
 ordered chains. Each row lists extract then normalize; spell chains are the
 same at both stages.

docs/consumers/dnd-pipeline.md

@@ -92,7 +92,7 @@ stream and exit-status contract.
 ## Discover The Published Bundle
 
 Decode the successful stdout document as a supported run-result schema. For
-the current contract, `schema_version` is `notarius.run-result.v1`. Tolerate
+the current contract, `schema_version` is `notarius.run-result.v2`. Tolerate
 unknown fields allowed by that version, but reject an unsupported schema
 version.
 
@@ -145,7 +145,8 @@ The JSON encoder always publishes these bundle-management files:
 | `index.json` | Discovery document for lane and pipeline-wide artifacts. |
 | `manifest.json` | Run provenance and result summaries. |
 | `rejected.json` | Rejected pipeline outputs. |
-| `warnings.json` | Accepted-output and run warnings. |
+| `warnings.json` | Actionable process-degradation warnings. |
+| `diagnostics.json` | Advisory and observation findings for accepted artifacts. |
 
 The complete configuration also requests two pipeline-wide artifacts:
 

docs/consumers/subprocess.md

@@ -79,7 +79,7 @@ silently treated as fully reviewed by the caller.
 ## Preserve Provenance And Handle Data Carefully
 
 Keep the receipt with the published `manifest.json`, and retain
-`rejected.json` and `warnings.json` when review or later provenance requires
+`rejected.json`, `warnings.json`, and `diagnostics.json` when review or later provenance requires
 them. Treat the input, output bundle, cache, debug bundle, and captured process
 logs as potentially sensitive data. Apply the caller's access controls and
 retention policy, and avoid copying secrets into arguments, logs, or

docs/integrations/json-output.md

@@ -9,7 +9,7 @@ Output configuration, including chunk-map and evidence-context publication, belo
 ## Bundle Layout
 
 All paths below are logical, relative, slash-separated bundle paths. The
-encoder always emits the first four JSON files below and adds lane or
+encoder always emits the first five JSON files below and adds lane or
 pipeline-wide artifact files when their corresponding artifacts are available:
 
 A subprocess caller first obtains the physical bundle root from the
@@ -21,7 +21,8 @@ root for the logical discovery described here.
 | `index.json` | Entry point that names the other published files and lane payloads. |
 | `manifest.json` | Run provenance and result summaries. |
 | `rejected.json` | Rejected pipeline outputs. |
-| `warnings.json` | Accepted-output and run warnings. |
+| `warnings.json` | Actionable process-degradation warnings. |
+| `diagnostics.json` | Accepted-artifact quality advisories and normalization observations. |
 | `lanes/<safe-lane-id>.json` | One normalized artifact payload for each lane. |
 | `chunk-map.json` | Optional accepted chunk map, when its export is enabled and available. |
 | `evidence-context.json` | Optional selected source-unit excerpt, when evidence publication is enabled. |
@@ -39,7 +40,8 @@ normalized lanes has this valid minimal index:
   "manifest_file": "manifest.json",
   "output_files": [],
   "rejected_file": "rejected.json",
-  "warnings_file": "warnings.json"
+  "warnings_file": "warnings.json",
+  "diagnostics_file": "diagnostics.json"
 }
 ```
 
@@ -49,6 +51,7 @@ normalized lanes has this valid minimal index:
 | `output_files` | Yes | Lane descriptors sorted by `lane_id`. |
 | `rejected_file` | Yes | Always `rejected.json`. |
 | `warnings_file` | Yes | Always `warnings.json`. |
+| `diagnostics_file` | Yes | Always `diagnostics.json`. |
 | `chunk_map` | No | Descriptor for the pipeline-wide `chunk-map.json`; never a lane descriptor. |
 | `evidence_context` | No | Descriptor for the pipeline-wide `evidence-context.json`; never a lane descriptor. |
 
@@ -132,7 +135,7 @@ These values describe observed execution; they are not a backend-registration
 interface. Entries that differ by backend or effective reasoning remain
 distinct even when their profile, provider, and model are otherwise equal.
 
-## Rejections And Warnings
+## Rejections, Warnings, And Diagnostics
 
 `rejected.json` is always an object with a `rejected` array. Each entry has
 required `stage` and `message`; `step_id`, `lane_id`, `module_key`, `chunk_id`,
@@ -142,9 +145,47 @@ contain the bounded `validation` summary described above; the existing singular
 validator and reason fields remain the first configured rejection for
 compatibility.
 
-`warnings.json` is always an object with a `warnings` array. Each warning has
-`reason_code` and `message`; `scope` is optional. Both arrays are empty when
-there is nothing to report.
+`warnings.json` is always the `notarius.warnings.v2` envelope:
+
+```json
+{
+  "schema_version": "notarius.warnings.v2",
+  "group_count": 0,
+  "occurrence_count": 0,
+  "groups": []
+}
+```
+
+It contains only process warnings. `group_count` is exact, and
+`occurrence_count` is the exact sum of its group occurrence counts.
+
+`diagnostics.json` is always the `notarius.diagnostics.v1` envelope:
+
+```json
+{
+  "schema_version": "notarius.diagnostics.v1",
+  "group_count": 0,
+  "occurrence_count": 0,
+  "truncated": false,
+  "unrepresented_occurrence_count": 0,
+  "groups": []
+}
+```
+
+It contains only advisory and observation groups. `group_count` counts groups
+represented in `groups`; `occurrence_count` includes both represented and
+unrepresented occurrences. When `truncated` is true,
+`unrepresented_occurrence_count` is the exact number omitted from group
+representation.
+
+Each group has `disposition`, `category`, `reason_code`, framework-owned
+`origin`, exact `occurrence_count`, bounded `samples`, and
+`omitted_sample_count`. Samples carry safe `scope` and `message`, plus a chunk
+ID and zero-based chunk index when applicable. A group retains at most three
+distinct samples. The framework fails rather than truncating actionable
+warnings beyond 128 groups; it represents at most 256 advisory/observation
+groups and records further occurrences through the diagnostic truncation
+fields above.
 
 ## Compatibility
 

docs/integrations/run-result.md

@@ -9,18 +9,22 @@ Command syntax, streams, and exit statuses are defined in the
 
 ## Schema
 
-The current schema version is `notarius.run-result.v1`.
+The current schema version is `notarius.run-result.v2`.
 
 | Field | Required | Meaning |
 | --- | --- | --- |
-| `schema_version` | Yes | Exactly `notarius.run-result.v1`. |
+| `schema_version` | Yes | Exactly `notarius.run-result.v2`. |
 | `run_id` | Yes | The finalized Notarius run identifier. |
 | `pipeline_id` | Yes | The effective pipeline identifier. |
 | `output_directory` | Yes | Absolute path to the published, run-specific output bundle. |
 | `index_file` | For the production JSON output | Logical path `index.json`; omitted for other output modules. |
 | `normalized_output_count` | Yes | Number of final normalized outputs. |
 | `rejected_output_count` | Yes | Number of recorded rejected outputs. |
-| `warning_count` | Yes | Number of final run warnings. |
+| `warning_group_count` | Yes | Exact number of actionable warning groups. |
+| `warning_occurrence_count` | Yes | Exact occurrences represented by actionable warning groups. |
+| `diagnostic_group_count` | Yes | Number of represented advisory and observation groups. |
+| `diagnostic_occurrence_count` | Yes | Advisory and observation occurrences, including unrepresented occurrences. |
+| `diagnostics_truncated` | Yes | Whether advisory/observation group representation was truncated. |
 | `validation_status` | Yes | The final run manifest validation status. |
 | `validation_summaries` | No | Bounded per-producer validation outcomes; present when producer work ran. |
 | `debug_directory` | No | Absolute path to the run-specific debug bundle when requested debug capture completed. |
@@ -34,14 +38,18 @@ means one or more otherwise accepted results advanced under validator-failure
 
 ```json
 {
-  "schema_version": "notarius.run-result.v1",
+  "schema_version": "notarius.run-result.v2",
   "run_id": "run-1770000000000000000-0123456789abcdef0123456789abcdef",
   "pipeline_id": "dnd-session",
   "output_directory": "/work/results/run-1770000000000000000-0123456789abcdef0123456789abcdef",
   "index_file": "index.json",
   "normalized_output_count": 6,
   "rejected_output_count": 2,
-  "warning_count": 1,
+  "warning_group_count": 1,
+  "warning_occurrence_count": 2,
+  "diagnostic_group_count": 3,
+  "diagnostic_occurrence_count": 5,
+  "diagnostics_truncated": false,
   "validation_status": "incomplete",
   "validation_summaries": [
     {

docs/internal/dnd.md

@@ -29,13 +29,17 @@ The D&D registrar registers the family’s artifact codecs, extractors, typed
 append-order mergers, normalizers, validators, prompt assets, fallback LLM
 profile asset, and default validator chains. Each extractor and normalizer has
 a stable module spec, explicit execution class, strict option decoding, and a
-typed builder. Scene chunking, every extractor, and NPC, location, and item-registry
-normalization are registered as `llm_backed`; the remaining current D&D mergers
-and normalizers are `deterministic`. The metadata is available to catalog inspection and
-resolved-pipeline debug data and determines which selected bindings inherit the
-pipeline profile. The registry normalizers use `single_response_v1`, forwarding
-corrections to their reconciliation completion and retaining the accepted raw
-proposal only as an owned model candidate. Configuration remains the canonical owner of the exact keys,
+typed builder. Scene chunking, every extractor, and NPC, location, and
+item-registry normalization are registered as `llm_backed`; the remaining
+current D&D mergers and normalizers are `deterministic`. The metadata is
+available to catalog inspection and resolved-pipeline debug data and determines
+which selected bindings inherit the pipeline profile. The registry normalizers
+use `single_response_v1`, forwarding corrections to their reconciliation
+completion and retaining the accepted raw proposal only as an owned model
+candidate. When deterministic proposal safety or typed application rejects a
+group, they provide separate model-facing prose for a corrective module retry;
+internal issue categories, reason codes, and operator messages remain
+diagnostic-only. Configuration remains the canonical owner of the exact keys,
 profile precedence, and validator order.
 
 Private structured-LLM response schemas are deliberately minimal. They reject
@@ -117,14 +121,27 @@ presentation, and final ephemeral generic transcript windows. These orders and
 cache controls are prompt behavior; change them only through the owning
 manifest and prompt declaration.
 
+NPC, item, and location registry reconciliation translate shared proposal
+safety categories into bounded prose that references only the response-local
+duplicate-group ordinals and candidate handles. Item reconciliation appends its
+typed rule that currency may be consolidated only with aliases of the same
+denomination and never with non-currency items. The next normalize attempt
+receives that prose with the exact defective proposal under the shared
+replacement-request protocol. Structurally invalid output has no valid proposal
+candidate and receives a fresh feedback-free attempt instead. If the stage
+budget is exhausted, safe groups stay applied, unsafe groups stay separate, and
+one fallback warning summarizes the final defect without raw model content.
+
 ## Evidence, Candidates, And Normalization
 
 The current transcript is the only durable evidence source. Extractors assign
-the current source identity, preserve candidate evidence ranges for validators,
-and canonically order or remove exact duplicate ranges without asking the
-model to repair semantic errors. Campaign context and generated artifacts may
-ground names or control routing, but they never establish evidence for a D&D
-result.
+the current source identity and losslessly order any reversed range whose two
+endpoints resolve in that source, using transcript position rather than numeric
+unit-ID order. They then canonically order ranges and remove exact duplicates.
+This routine canonicalization does not request a retry or emit a warning.
+Unresolvable or wrong-source ranges remain unchanged for validators to reject.
+Campaign context and generated artifacts may ground names or control routing,
+but they never establish evidence for a D&D result.
 
 Default chains keep responsibilities separate: structural validators assess the
 candidate, source-reference validators resolve cited ranges against the current
@@ -134,22 +151,113 @@ relatedness validators report advisory evidence concerns. The configured order
 is documented in
 [Configuration](../config.md#production-validator-keys-and-default-chains).
 
-Every D&D rejection describes the correction in transcript-grounded domain
-terms, using contextual names, artifact fields, and source segment ranges when
-useful. The guidance must not ask the model to reproduce durable entity IDs,
-hashes, validator module keys, or reason codes. Those identifiers remain in
-ordinary validation provenance; only the actionable semantic guidance is
-eligible for the correction prompt.
+The optional `extract/dnd/scene-descriptions/combat_semantics` validator is the
+D&D family's LLM-backed review of only combat versus non-combat classification.
+It selects the shared combat-policy prompt fragment, receives the proposed kind
+and current chunk, and maps its verdict deterministically into producer
+guidance. It does not assess titles, summaries, non-combat subtype, or scene
+boundaries; deferred boundary-coherence review remains separate. It is opt-in;
+[Configuration](../config.md) owns selection and retry/failure behavior.
+
+### Combat-semantics provider evaluation
+
+The human-reviewed corpus at
+`internal/modules/dnd/validate/scenedescriptions/combat_semantics/testdata/evaluation_cases.json`
+owns the proposed kind, expected verdict, and reviewer rationale for each
+synthetic case. Its package test validates the fixture contract only. Provider
+evaluation remains an explicit maintainer operation and must not be added to
+the default offline test suite.
+
+Use the following protocol before proposing default-chain inclusion:
+
+1. Record the Notarius commit, prompt and schema fingerprints, provider, model,
+   profile settings, reasoning effort, structural-repair setting, number of
+   repetitions, and evaluation date before collecting results. Do not revise
+   expected verdicts merely to agree with provider output; a substantive corpus
+   correction requires independent human review.
+2. Exercise the production validator construction and prompt assets from an
+   explicitly invoked, disposable evaluation driver or test in the validator
+   package. For each corpus case, construct transcript source units from the
+   listed IDs and text, assign matching per-unit source references, and use
+   `source.MaterializeChunkPlan` with one range spanning those units. Construct
+   exactly one scene whose ID and source range match that chunk and whose kind
+   is the case's `proposed_kind`; title and summary may use fixed placeholders
+   because the validator neither receives nor evaluates them. Invoke the typed
+   validator at extract stage through the production registry and scheduled LLM
+   client. Do not commit provider credentials, generated source material, or an
+   always-on live test.
+3. Compare the resulting approval or reason code with `expected_verdict`.
+   Record an unexpected approval of an expected rejection as a false
+   acceptance, an unexpected rejection of an expected approval as a false
+   rejection, the opposite rejection direction separately, and any validator
+   execution failure separately from semantic accuracy. Retain per-case results
+   so repeated trials and systematic failure modes remain visible.
+4. Evaluate producer correction separately with representative complete
+   scene-description runs configured as shown in
+   [Configuration](../config.md#production-validator-keys-and-default-chains).
+   For every initial semantic rejection, record whether the next producer
+   attempt returns the requested combat status and is approved. Do not count a
+   PromptKit structural repair as a producer-correction attempt.
+5. Run the correction evaluation with debug capture enabled and without reused
+   extraction checkpoints. Record added validator and producer calls, elapsed
+   latency, and cumulative prompt, completion, cached, and total token usage
+   from the debug attempt records. Compare these values with an otherwise
+   identical run whose scene-description chain omits the semantic validator.
+
+The default-chain review must consider false acceptance, false rejection,
+wrong-direction rejection, execution failure, producer-correction success,
+added calls, latency, and token use together. A structurally successful
+provider run alone is not evidence that the validator should become a default.
+
+Every producer-correctable D&D rejection describes all currently detectable
+corrections in transcript-grounded domain terms, using contextual names,
+model-owned artifact fields, and source segment ranges when useful. Validators
+collect independent record defects in one pass so one retry does not merely
+reveal the next issue. Shared D&D diagnostic helpers keep repeated rules and
+record descriptions stable, de-duplicated, and bounded; each artifact family
+continues to own the semantic rule and its prose.
+
+Operator diagnostics and model guidance are separate products of the same
+assessment. Operator messages may use typed paths, reason details, and opaque
+application identities. Correction guidance must not copy those messages or
+ask the model to reproduce durable entity IDs, hashes, validator module keys,
+reason codes, or Go field paths. A registry-normalization rejection instead
+speaks in terms of the duplicate-group proposal response the normalizer can
+actually revise. Normalization-only deterministic invariants retain useful
+operator detail but do not imply that a model controls derived ordering or
+identity. Only bounded actionable semantic guidance is eligible for a
+correction prompt.
+
+Private LLM schemas use simple enums for closed categorical fields when the
+provider-compatible shape can express the rule directly. Deterministic typed
+validators retain the same checks as defense in depth and for non-LLM
+producers. Private schemas keep every property required and avoid optional
+properties, `uniqueItems`, and conditional cross-field logic.
+
+Item-occurrence shape validation groups repeated holder mistakes by occurrence
+kind and gives the producer the required JSON null/non-null relationship. It
+identifies affected records by contextual item name and cited transcript range,
+never by the deterministically attached durable item ID. Holder mistakes remain
+semantic rejections rather than silent rewrites because changing a holder can
+also change the meaning of the occurrence kind.
 
 Enemy-event extraction additionally rejects a second `engaged` observation for
 the same comparison identity within one scene-scoped result. Normalization may
 combine results from distinct scenes, so it intentionally does not apply that
 rule. Configuration owns the exact validator key and chain position.
 
+Extraction source-reference validators share one full-span chunk-containment
+policy. After ordinary reference validity succeeds, the policy resolves both
+endpoints through document order and requires every source unit in the
+inclusive range to be present in the current chunk. It does not assume numeric
+unit-ID ordering, mutate input, or weaken wrong-source and unresolved-reference
+validation. Scene descriptions remain separate because their validator owns an
+exact one-scene range contract rather than general extraction containment.
+
 Normalizers are deterministic for spells, combat turns, item occurrences, NPC
 occurrences, scene descriptions, enemy events, and location occurrences. They
 canonicalize display values and evidence, use source-document order for stable
-output, and issue bounded warnings for changes or collapsed duplicates. NPC,
+output, and emit bounded normalization observations for changes or collapsed duplicates. NPC,
 item, and location registry normalizers are intentional exceptions: each first
 produces a deterministic candidate set, then may use a bounded structured-LLM
 proposal to reconcile identity groups.

docs/internal/llm.md

@@ -247,8 +247,12 @@ redaction boundary.
 Structural repair does not replace pipeline retry behavior: a binding's
 configured retry count reruns its complete stage attempt after an operational
 or structural error, module-requested retry, or actionable semantic rejection.
-The pipeline owns attempt lifecycle, validation chains, and retry diagnostics;
-see [Pipeline Internals](pipeline.md#validation-retries-and-output) and the
+An actionable module-requested retry and a validator rejection both use the
+same correction payload when the producer exposes an exact latest response;
+feedback-free module retries reconstruct the ordinary request without appended
+messages. The pipeline owns attempt lifecycle, validation chains, and retry
+diagnostics; see
+[Pipeline Internals](pipeline.md#validation-retries-and-output) and the
 [binding reference](../config.md#module-bindings-and-validators).
 
 ## Timeout Ownership

docs/internal/modules.md

@@ -102,7 +102,7 @@ returned duplicate groups into a stable non-overlapping plan.
 The normalizer then applies that plan through a typed `ApplicationPolicy`. The
 core preserves ungrouped records, contribution order, and provenance while the
 artifact family owns group guards, field and evidence consolidation, durable
-ID derivation, retry and fallback presentation, warnings, and postconditions.
+ID derivation, retry and fallback presentation, classified diagnostics, and postconditions.
 Request-local handles do not enter the typed value or durable artifact. Fewer
 than two eligible candidates skips model invocation; exceeding a candidate or
 combined-material bound preserves the deterministic result under the family's
@@ -115,6 +115,16 @@ eligible producers. Deterministic skip, limit, and fallback outcomes carry no
 model candidate, so a later rejection applies terminal policy without spending
 an ineffective semantic retry.
 
+When proposal assessment or typed group application rejects a structurally
+valid group, the normalizer may return its safe partial value with a
+module-requested retry. A feedback-capable directive supplies bounded
+model-facing correction guidance separately from operator diagnostics and
+retains the exact proposal response as its candidate. The shared stage retry
+mechanism appends that response and guidance to a fresh complete request. A
+structurally invalid completion has no valid candidate and therefore requests a
+feedback-free fresh attempt. On exhaustion, only the final safe fallback and
+its bounded process diagnostic advance to validation.
+
 The core supplies a conservative generic prompt and the single private
 response schema. A domain prompt may substitute its semantic instructions but
 mounts the core-owned protocol and candidate/transcript presentation assets.

docs/internal/pipeline.md

@@ -12,7 +12,7 @@ own durable output shapes. Concrete production extensions are covered by
 The pipeline framework accepts a resolved composition, registries, shared
 dependencies, input bytes, a supplied prompt session, and state/debug
 collaborators. It returns logical output files, normalized artifacts, recorded
-rejections and warnings, manifest provenance, and checkpoint decisions. The
+rejections, grouped diagnostics, manifest provenance, and checkpoint decisions. The
 CLI owns process arguments, configuration discovery, session resolution,
 physical roots, and placement of returned output files.
 
@@ -150,34 +150,41 @@ oversized aggregate is a framework contract error; guidance is never inferred
 or truncated.
 
 The runner applies the binding's retry policy around a stage operation and its
-complete validation chain. It preserves warnings only from the final accepted
-or rejected attempt, plus one fixed warning per validator whose execution
+complete validation chain. It preserves terminal diagnostics only from the final accepted
+or rejected attempt, plus one fixed validation-incomplete warning per validator whose execution
 budget was exhausted under `warn_continue`. Cancellation stops retries.
 Normalizer-specific retry directives consume this same budget and validate any
-final safe fallback through the normalizer chain.
+final safe fallback through the normalizer chain. A directive may carry bounded
+correction guidance only when it also exposes the exact latest
+`single_response_v1` candidate. The state machine then uses the same replacement
+request shape as validator correction. A directive without guidance clears any
+prior correction and starts a fresh attempt, which preserves structural retry
+behavior when no valid response exists.
 
 The artifact-neutral producer-attempt state machine owns that shared budget,
 attempt provenance, semantic-correction material, and terminal-policy
 selection. It accepts producer and complete-validation closures, so artifact
 materialization, cache handling, checkpoints, and debug output stay at the
 operation boundary. It distinguishes operational, structural, module-requested,
-and semantic retries. A semantic retry is available only for a valid latest
-`single_response_v1` candidate; a deterministic or no-model rejection instead
-settles the semantic policy immediately. Structural-output errors alone use the
-structural policy, and validation failure without rejection settles the
-validator-failure policy without regenerating the producer.
+and validator-semantic retries. Model feedback from either semantic source is
+available only for a valid latest `single_response_v1` candidate. A
+deterministic or no-model validator rejection instead settles the semantic
+policy immediately, while a feedback-free module directive remains an ordinary
+fresh retry. Structural-output errors alone use the structural policy, and
+validation failure without rejection settles the validator-failure policy
+without regenerating the producer.
 
 Chunk planning uses this state machine for generated plans. A rejected or
 validation-incomplete automatic cache hit is not model material and therefore
 falls through to a fresh initial generation at producer attempt one; it neither
 receives a correction, consumes retry budget, promotes cached-candidate
-warnings, nor overwrites the stored record. An incomplete cache validation
+diagnostics, nor overwrites the stored record. An incomplete cache validation
 under `fail_run` terminates instead. Only a newly generated, completely
 validated plan is published to the chunk-plan store. Rejected plans never
 advance, and validation-incomplete plans remain unpublishable.
 
 After terminal lane work, the runner assembles manifest provenance, normalized
-artifacts, rejections, warnings, and an optional accepted chunk map. When an
+artifacts, rejections, final grouped diagnostics, and an optional accepted chunk map. When an
 output policy selected evidence lanes, it decodes accepted serialized normalize
 outputs through their registered codecs and invokes the prepared typed
 projectors. Rejected or absent lanes contribute nothing. This reconstruction is

docs/internal/state.md

@@ -64,12 +64,13 @@ Ordinary resume loads extract, merge, and normalize checkpoints progressively
 and may execute later lane stages after an earlier cache miss. Selective
 recomputation instead asks the loader for the required producer's accepted
 normalize artifact. That lookup reuses the existing normalize files, requires
-workspace schema v3 plus an exact non-empty invocation identity, and deliberately
+workspace schema v4 plus an exact non-empty invocation identity, and deliberately
 does not require extract or merge checkpoint files or dependency fingerprints.
 The runner performs canonical codec and producer-provenance validation before
 cloning the artifact into normal step output. Success restores only stored
-normalize warnings and emits one normalize decision; failure retains the files,
-records the decision, and stops without executing the producer or consumer.
+normalize diagnostics and emits one normalize decision; failure retains the
+files, records the decision, and stops without executing the producer or
+consumer.
 
 The loader assigns a typed category and reason code at each validation site;
 diagnostic prose is not classified after the fact. The runner then applies
@@ -97,7 +98,7 @@ owns the operator workflow and stable reason-code meanings.
 
 `internal/core/debugbundle` allocates an explicitly requested per-run bundle
 with `summary/` and `trace/` roots. `SummaryWriter` persists redacted command,
-resolution, run, warning, and failure artifacts. `internal/framework/debug`
+resolution, run, final grouped diagnostic, and failure artifacts. `internal/framework/debug`
 implements the pipeline-facing trace recorder under the trace root.
 
 The CLI allocates a bundle before pipeline resolution and treats requested

docs/operations.md

@@ -105,8 +105,10 @@ and resolves configuration before module preparation and source parsing. It
 then performs any permitted cache lookup, executes the pipeline, and publishes
 logical output files only after a successful runner result.
 
-On success, the command reports the output bundle path. A warning-bearing run
-still succeeds and reports its warning count on standard error. Errors and
+On success, the command reports the output bundle path. A run with actionable
+process warnings still succeeds and reports warning-group and occurrence counts
+on standard error; advisory and observation findings do not produce a warning
+line. Errors and
 their exit classes are defined in the [CLI reference](cli.md#output-streams-and-exit-statuses).
 
 ## Validation Retries And Terminal Outcomes
@@ -127,13 +129,14 @@ validator execution failure normally uses `warn_continue`, which keeps an
 otherwise accepted result in the current run with `incomplete` validation
 provenance. It emits one bounded warning for every validator whose execution
 budget was exhausted. A corrected result that later passes validation does not
-retain abandoned-attempt warnings.
+retain diagnostics from abandoned attempts.
 
 Treat a successful process exit as a completed run, not as proof that every
 candidate was fully validated. Inspect the receipt's `validation_status`,
-`validation_summaries`, rejection count, and warning count when an orchestrator
-requires complete validation. The durable fields and their meanings are owned
-by the [run-result receipt](integrations/run-result.md) and
+`validation_summaries`, rejection count, and warning group and occurrence
+counts when an orchestrator requires complete validation. The durable fields
+and their meanings are owned by the
+[run-result receipt](integrations/run-result.md) and
 [published JSON output contract](integrations/json-output.md).
 
 ## Output Bundles
@@ -272,7 +275,7 @@ Only a [debug-enabled run](cli.md#run) creates a bundle:
 ~~~
 
 The summary contains redacted invocation and resolution information plus run,
-warning, checkpoint, chunk-plan, and terminal reporting artifacts. Attempt
+final grouped diagnostic, checkpoint, chunk-plan, and terminal reporting artifacts. Attempt
 terminal records contain bounded attempt kinds, validator outcomes, policy,
 decision, PromptKit repair count, and usage; they do not contain assistant
 responses or complete correction messages. The trace contains allowlisted

docs/policy/architecture.md

@@ -146,8 +146,9 @@ lanes, validators, and LLM profile: the canonical source digest selects the
 plan, while the current run still applies its configured chunk validators to
 the materialized chunks.
 
-The framework owns orchestration and handoff provenance. Modules return logical
-results and warnings; they do not own CLI reporting, physical output, cache, or
+The framework owns orchestration, origin enrichment, aggregation, and handoff
+provenance. Modules return logical results and classified diagnostics; they do
+not own CLI reporting, physical output, cache, or
 debug roots, durable file placement, or checkpoint and debug lifecycle.
 
 After pipeline-wide chunking, extraction uses bounded framework concurrency.
@@ -159,17 +160,26 @@ may overlap. The framework must not create unbounded goroutines per lane or
 chunk.
 
 Completion timing does not choose public ordering or errors. The coordinator
-orders accepted artifacts, warnings, rejections, checkpoint events, and
+orders accepted artifacts, grouped diagnostics, rejections, checkpoint events, and
 framework errors by stable pipeline scope. Rejections do not cancel unrelated
 work. A framework error cancels derived work, prevents undispatched work from
 starting, waits for started work, and prevents output encoding.
 
+Warnings are process-only signals: configuration degradation, approved fallback,
+or incomplete configured validation. Quality uncertainty and grounding findings
+are advisories; successful canonicalization and cleanup are observations.
+Modules choose that semantic classification, while the framework attaches
+origin, aggregates groups, enforces bounds, and presents final collections.
+An ordinary successful run therefore has zero warnings. See
+[ADR-0015](../adr/0015-separate-process-warnings-from-quality-diagnostics.md)
+for the decision rationale.
+
 ## Validation
 
 Validation is a framework-managed boundary around outputs from chunk, extract,
 merge, and normalize stages. Validators receive immutable stage output and
-make an explicit whole-output decision: approve, approve with warnings,
-reject, fail, or skip when a runtime prerequisite is unavailable.
+make an explicit whole-output decision: approve, reject, fail, or skip when a
+runtime prerequisite is unavailable.
 
 Typed artifact validators receive the domain value directly. Chunk validators
 receive source-zone chunks, while serialized validators receive immutable
@@ -191,6 +201,13 @@ they are not model instructions. The framework constructs model-facing retry
 text only from the semantic guidance and fails the contract rather than
 inventing or truncating missing guidance.
 
+An LLM-backed module may also request a feedback-aware retry when its own
+deterministic translation or typed safety policy rejects a structurally valid
+model response. It must supply model-facing guidance separately from its
+reason code, operator message, and fallback diagnostics, together with the
+exact `single_response_v1` candidate. A feedback-free module retry remains
+valid when no exact candidate exists.
+
 Default validator chains are production composition policy and are registered
 centrally by stage and module. Configuration may replace a stage-local default,
 including with an explicitly empty chain. Configured validator order is
@@ -212,12 +229,14 @@ the two budgets must remain separate.
 
 An LLM-backed producer can participate in semantic correction only when it
 declares `single_response_v1` and returns the exact one response that directly
-controlled its candidate. On an actionable rejection, the framework rebuilds
-the ordinary request and appends only the latest defective response as an
-`assistant` message plus one aggregated `user` correction message. This is a
-fresh replacement request, not a growing conversation. The retry budgets,
-terminal policy, and sensitive-data rationale are recorded in
-[ADR-0014](../adr/0014-feedback-aware-validation-retries.md).
+controlled its candidate. On an actionable validator rejection or
+feedback-capable module retry, the framework rebuilds the ordinary request and
+appends only the latest defective response as an `assistant` message plus one
+aggregated `user` correction message. This is a fresh replacement request, not
+a growing conversation. The retry budgets, terminal policy, and sensitive-data
+rationale are recorded in
+[ADR-0014](../adr/0014-feedback-aware-validation-retries.md) and
+[ADR-0016](../adr/0016-feedback-aware-module-requested-retries.md).
 
 When a model selects an application entity, callers must supply a contextual
 selection and deterministically attach the opaque application identity whenever

docs/releases/v0.5.0.md

@@ -0,0 +1,85 @@
+# Notarius v0.5.0
+
+This release separates actionable process warnings from extraction-quality
+advisories and routine normalization observations, giving operators a quiet
+warning channel without discarding durable diagnostic detail.
+
+## Summary
+
+Notarius now carries one validated, origin-aware diagnostic contract from
+producers and validators through retries, reusable state, output publication,
+debug summaries, run receipts, and CLI presentation. Warnings are reserved for
+process degradation or incomplete configured work. Data-quality findings are
+advisories, and successful deterministic cleanup is recorded as observations.
+An ordinary successful run therefore reports zero warnings while retaining
+bounded diagnostic provenance for later review.
+
+The framework aggregates findings deterministically by their stable identity
+and complete pipeline origin, preserves exact occurrence counts, and retains
+bounded representative samples. Warning groups fail rather than truncate;
+advisory and observation representation is bounded with explicit truncation
+metadata and exact unrepresented-occurrence counts.
+
+## Compatibility
+
+- `warnings.json` now uses the incompatible grouped
+  `notarius.warnings.v2` envelope and contains process warnings only. Consumers
+  of the former flat warning payload must migrate to the current
+  [JSON output contract](../integrations/json-output.md).
+- The new `diagnostics.json` file uses `notarius.diagnostics.v1` and contains
+  advisory and observation groups. Production `index.json` files always expose
+  both `warnings_file` and `diagnostics_file`.
+- The machine-readable run receipt is now `notarius.run-result.v2`. It replaces
+  `warning_count` with exact warning group and occurrence counts and adds
+  advisory/observation group, occurrence, and truncation fields. See the
+  current [run-result receipt](../integrations/run-result.md).
+- Custom output modules must return their complete logical file set or an
+  error. The former `OutputResult.Warnings` field has been removed; an output
+  module cannot report a warning after serializing its output.
+- Reusable state now uses `notarius.workspace.v4` and chunk-plan records use
+  `notarius.chunk-plan.v3` so they can preserve structured diagnostics. Older
+  pre-release reusable state is not reused under these contracts; start with
+  clean state when deterministic continuity with an older workspace is not
+  required.
+- Validation acceptance, semantic retry budgets, rejection policy, and D&D
+  artifact schema identities are unchanged by this release.
+
+## Upgrade
+
+1. Update subprocess consumers to require `notarius.run-result.v2` and read
+   `warning_group_count`, `warning_occurrence_count`,
+   `diagnostic_group_count`, `diagnostic_occurrence_count`, and
+   `diagnostics_truncated`.
+2. Update output-bundle consumers to decode `notarius.warnings.v2`, discover
+   `diagnostics.json` through `index.json`, and treat diagnostics as review
+   information rather than process warnings.
+3. Update any custom output module for the removal of
+   `OutputResult.Warnings`; return an error when encoding cannot complete.
+4. Clear pre-release reusable state before the first upgraded production run
+   when deterministic continuity with an older workspace is not required.
+5. Run `notarius config validate --config <path> --pipeline <id>` and perform
+   one representative run before promoting the release in an automated
+   pipeline.
+
+## Changes
+
+- Added validated diagnostic dispositions, categories, origins, stable reason
+  codes, exact occurrence counts, and bounded representative samples.
+- Added deterministic run-level aggregation with separate limits for
+  actionable warning groups and advisory/observation groups.
+- Reclassified D&D source-relatedness and unresolved-identity findings as
+  data-quality advisories and routine normalization changes as observations.
+- Preserved structured diagnostics across producer retries, validation,
+  generated-reference handoff, checkpoints, chunk-plan reuse, and debug
+  summaries while discarding superseded-attempt findings.
+- Added grouped `warnings.json`, a new grouped `diagnostics.json`, and the
+  corresponding production index entries.
+- Upgraded the machine-readable run receipt and human CLI summary to report
+  exact warning and diagnostic counts without allowing advisory volume to
+  create warning output.
+- Removed post-encoding output warnings and hardened diagnostic validation,
+  overflow handling, aggregate memory bounds, and warning-file path
+  presentation.
+- Documented diagnostic ownership, classification, operator interpretation,
+  durable contracts, and architectural invariants in ADR-0015 and the
+  canonical CLI, operations, integration, and internal documentation.

docs/roadmap/archive/warning-signal-and-presentation-audit.md

@@ -0,0 +1,648 @@
+# Warning Signal And Presentation Audit
+
+## Executive Assessment
+
+Notarius warning execution is mechanically stronger than its operator-facing
+presentation. Terminal-attempt promotion, stable ordering after concurrent
+work, checkpoint replay, validation summaries, and debug retention are all
+substantially correct. The audit found no general duplicate-append defect in
+the extract, merge, or normalize handoffs and no leakage of abandoned-attempt
+warnings into a successful result.
+
+The warning channel itself is not coherent. One flat `contracts.Warning` type
+currently represents at least four materially different concepts:
+
+- actionable degradation or incomplete validation;
+- heuristic data-quality doubt;
+- successful but potentially reviewable fallback; and
+- routine canonicalization, ordering, and deduplication observations.
+
+That conflation is the primary reason successful runs produce a count that is
+large but operationally weak. The maintained complete D&D example demonstrates
+the problem without a live provider: an approved run with no rejected outputs
+published 12 warning records, all from three advisory relatedness checks. An
+operator separately reported a successful complete D&D run with 10 outputs,
+one rejection, and 85 warnings. The production bundle for that run was not
+available in this environment, so its reason-code distribution could not be
+measured.
+
+The current implementation also has four correctness or robustness gaps:
+
+1. warning records lose stage, step, lane, module, validator, and chunk
+   provenance when promoted, which makes safe aggregation and diagnosis
+   impossible from `warnings.json` alone;
+2. there is no framework-level validation or aggregate bound, and the NPC- and
+   spell-relatedness validators bypass the D&D warning limiter entirely;
+3. warnings returned by an output encoder are added after `warnings.json` has
+   already been encoded, so the receipt, stderr, debug bundle, and published
+   warning file can disagree; and
+4. a skipped validator contributes to `incomplete` validation but does not
+   receive the warning generated for an exhausted validator failure.
+
+The recommended end state is a structured diagnostic contract with explicit
+disposition, category, origin, occurrence count, and bounded samples. Warnings
+are reserved for process-level degradation or incompleteness. LLM-judged or
+deterministically inferred extraction-quality signals are advisories, never
+warnings, and routine normalization observations remain inspectable without
+being reported as top-level warnings. An ordinary successful run in which all
+configured work completes normally should therefore report zero warnings. This
+is an architectural and durable-contract change, not merely revised CLI prose.
+
+## Evidence And Limits
+
+The audit used:
+
+- a complete static search of production `contracts.Warning` constructors,
+  reason-code constants, result fields, and promotion sites under `internal/`;
+- call-path inspection through producer attempts, validators, lane
+  coordination, chunk-plan reuse, checkpoints, output encoding, debug output,
+  CLI presentation, and run-result construction;
+- the maintained complete and minimal D&D examples with offline fake LLMs;
+- focused deterministic tests for warning bounds, semantic-reconciliation
+  fallback, `warn_continue`, semantic retries, terminal rejection, concurrency
+  ordering, and checkpoint reuse; and
+- the operator-provided observation of an 85-warning complete D&D run.
+
+No provider-backed production run was attempted because this environment has
+no API key. Consequently, the audit can establish warning mechanics, possible
+multiplicity, synthetic volume, and obvious heuristic limitations, but cannot
+estimate production frequency or the real false-positive rate of individual
+D&D advisories. Those measurements are not required to choose the recommended
+architecture; they are required before strengthening any heuristic advisory
+into a rejection or setting a numerical production acceptance target.
+
+## Complete Warning-Producer Inventory
+
+### Framework And Generic Boundaries
+
+| Producer | Reason code | Trigger and consequence | Multiplicity and bound | Current surfaces and coverage |
+| --- | --- | --- | --- | --- |
+| Reference materialization in `internal/framework/pipeline/references.go` | `empty_reference` | A bound external reference is a valid, accepted media type but contains zero bytes. The prompt may receive materially incomplete context. | One per empty bound file; finite by configuration but no shared run-level cap. | Enters `RunInput.Warnings`; reference tests protect contextual scope. |
+| Producer-attempt policy in `internal/framework/pipeline/producer_attempts.go` | `validator_execution_incomplete` | An applicable validator exhausted its execution budget and `warn_continue` accepted the otherwise valid candidate. | One per failed validator on each terminal candidate. An extract chain can multiply this by chunks and lanes. There is no global cap. | Durable warning, receipt count, stderr, debug, and validation summary. `TestWarnContinueRecordsOneWarningForEachExhaustedValidator` covers failures. |
+| Chunk, extract, merge, normalize, and output module result contracts | Module-defined | A module may return arbitrary warnings with its successful candidate. | No contract validation, message limit, per-result cap, or global cap. Current production modules are inventoried below. | Terminal-attempt filtering and concurrency ordering are well tested. |
+| Production JSON output encoder | None | The encoder copies incoming warnings into `warnings.json`; it does not currently create warnings. | Same count as its input. | JSON encoder and assembled-pipeline tests compare the incoming run warnings with the published file. |
+| Output encoder result contract | Module-defined | Any output encoder may return warnings discovered during encoding. | Unbounded by contract. No production encoder currently exercises this capability. | Appended to final `RunOutput` only after logical files were encoded; this is the cross-surface defect described in AUD-WARN-004. |
+
+Input adapters and production mergers do not currently have independent
+warning producers. Chunk-plan and checkpoint decisions are structured manifest
+or debug provenance rather than warnings. Cancellation and hard persistence,
+reference, parsing, serialization, and provider failures remain errors.
+
+### D&D Extraction Gates
+
+| Producer | Reason code | Trigger and consequence | Multiplicity and bound |
+| --- | --- | --- | --- |
+| `dnd/combat-turns` extractor | `scene_classification_unavailable` | The chunk has no exact matching scene-description classification. The extractor returns an empty accepted result and skips the LLM, so combat-turn output may be incomplete. | At most one per chunk for this lane. |
+| `dnd/enemy-events` extractor | `scene_classification_unavailable` | The same missing or mismatched scene gate causes accepted empty enemy-event output. | At most one per chunk for this lane. |
+
+An exact non-combat classification produces an intentional empty result without
+a warning. An exact combat classification proceeds normally. The two producers
+share a code and operator consequence but use different messages; their module
+origins are not retained in the final warning record.
+
+### D&D Source-Relatedness Validators
+
+All ten relatedness validators are deterministic advisories: they approve the
+candidate and warn when contextual prose or an entity name is not lexically
+present in cited text. Shape and source-reference failures are deliberately
+left to blocking validators earlier in the chain. The same relatedness
+validator is registered in both the extract and normalize default chain for
+each artifact family in `internal/modules/dnd/register/chains.go`.
+
+| Artifact family | Warning reason | Per-record trigger | Local bound | Omission reason |
+| --- | --- | --- | --- | --- |
+| Combat turns | `combat_turn_not_near_source` | Actor token sequence absent | 20 per validator invocation | `combat_turn_relatedness_warnings_omitted` |
+| Enemy events | `enemy_event_not_near_source` | Subject token sequence absent | 20 | `enemy_event_relatedness_warnings_omitted` |
+| Item occurrences | `item_occurrence_source_unrelated` | Item name token sequence absent | 20 | `item_occurrence_relatedness_warnings_omitted` |
+| Item registry | `item_not_near_source` | Item name token sequence absent | 20 | `item_relatedness_warnings_omitted` |
+| Location occurrences | `location_occurrence_not_near_source` | Location name token sequence absent | 20 | `location_occurrence_relatedness_warnings_omitted` |
+| Location registry | `location_not_near_source` | Location name token sequence absent | 20 | `location_relatedness_warnings_omitted` |
+| NPC occurrences | `npc_occurrence_not_near_source` | NPC name token sequence absent | 20 | `npc_occurrence_relatedness_warnings_omitted` |
+| NPC registry | `npc_not_near_source` | NPC name token sequence absent | **Unbounded** | None |
+| Scene descriptions | `scene_description_not_near_source` | No significant title or summary token appears; up to two findings per scene | 20 | `scene_description_relatedness_warnings_omitted` |
+| Spells | `spell_not_near_source` | Spell-name token sequence absent | **Unbounded** | None |
+
+The eight limiter-generated omission records are presentation artifacts, not
+new source-relatedness conditions. They occupy a warning slot and make list
+length differ from the actual occurrence count.
+
+### D&D Normalizers
+
+Every production D&D normalizer bounds its returned warning slice to 20 through
+`internal/modules/dnd/shared/diagnostics`, including a final omission record
+when needed. Registry semantic retries reserve space for their fallback
+warning. The following table is complete by semantically distinct condition;
+codes listed together are parallel artifact-family variants.
+
+| Condition | Reason codes | Result impact | Current classification assessment |
+| --- | --- | --- | --- |
+| Display or field whitespace/name canonicalization | `npc_fields_normalized`, `item_fields_normalized`, `location_fields_normalized`, `spell_name_canonicalized`, `combat_actor_canonicalized`, `enemy_event_name_canonicalized`, `item_occurrence_name_canonicalized`, `location_occurrence_name_canonicalized`, `scene_description_prose_normalized` | Deterministic successful mutation. The item-occurrence code can also describe `from`/`to` whitespace, not only the item name. | Routine observation. |
+| Durable ID recomputation | `npc_id_recomputed`, `item_id_recomputed`, `location_id_recomputed` | Restores the deterministic name-derived ID. | Routine observation; invalid identity is separately rejected by default chains. |
+| Source-reference sorting or deduplication | `source_references_normalized` | Sorts and removes exact duplicate references while deliberately preserving invalid references for their validators. | Routine observation. Shared code is useful but ambiguous without producer origin. |
+| Canonical record ordering | `combat_turns_reordered`, `enemy_events_reordered`, `item_occurrences_reordered`, `location_occurrences_reordered`, `npc_occurrences_reordered`, `scene_description_order_normalized` | Deterministic order changes only. | Routine observation. |
+| Exact or approved semantic duplicate consolidation | `duplicate_npc_collapsed`, `duplicate_item_collapsed`, `duplicate_location_collapsed`, `duplicate_spell_cast_collapsed`, `duplicate_combat_turn_collapsed`, `duplicate_enemy_event_collapsed`, `duplicate_item_occurrence_collapsed`, `duplicate_location_occurrence_collapsed`, `duplicate_npc_occurrence_collapsed`, `scene_description_duplicate_collapsed` | Removes duplicate records and preserves or combines canonical evidence according to the artifact policy. Registry codes cover both exact and accepted semantic consolidation. | Durable normalization observation; not normally operator-actionable. |
+| Unresolved external membership | `spell_name_unresolved`, `item_occurrence_unknown_item_id`, `location_occurrence_unknown_location_id` | The value is preserved but is not grounded in the effective catalog or registry. Default chains normally reject the same condition before normalization; it remains reachable with validator overrides or defensive direct use. | Actionable data-quality warning. |
+| Unsafe currency consolidation proposal | `item_semantic_proposal_invalid` | The proposed group is rejected and all records are preserved because denominations or currency/non-currency members are incompatible. The same code is also used internally as a retry reason. | Advisory about model proposal quality; no accepted-data loss. The control and diagnostic meanings should be separated. |
+| Semantic reconciliation unavailable or exhausted | `npc_semantic_reconciliation_exhausted`, `item_semantic_reconciliation_exhausted`, `location_semantic_reconciliation_exhausted` | The safe deterministic result is accepted, but possible semantic duplicates remain. | Actionable fallback warning. |
+| Local warning truncation | `npc_normalization_warnings_omitted`, `item_normalization_warnings_omitted`, `location_normalization_warnings_omitted`, `spell_normalization_warnings_omitted`, `combat_turn_normalization_warnings_omitted`, `enemy_event_normalization_warnings_omitted`, `item_occurrence_normalization_warnings_omitted`, `location_occurrence_normalization_warnings_omitted`, `npc_occurrence_normalization_warnings_omitted`, `scene_description_normalization_warnings_omitted` | Reports that individual records were omitted from presentation. | Group metadata, not an independent warning. |
+
+No production D&D normalization warning exposes raw model responses,
+correction guidance, or provider errors. Most dynamic names are quoted and
+truncated by the shared helper. That local discipline is not enforced by the
+generic warning contract, and the spell relatedness message does not use the
+shared truncation helper.
+
+## Warning Propagation And Surface Map
+
+```text
+external-reference warnings -----------------------------+
+                                                           |
+module candidate warnings -> validation chain warnings     |
+             |                      |                       |
+             +---- producer-attempt terminal policy -------+
+                                  |                        |
+                   accepted / terminal rejection only      |
+                                  |                        |
+              chunk or lane result in canonical order      |
+                                  |                        |
+       checkpoint record/replay and ordered step merge     |
+                                  |                        |
+                            RunOutput.Warnings <------------+
+                                  |
+                   OutputRequest -> output encoder
+                                  |             |
+                         warnings.json     OutputResult.Warnings
+                                                |
+                              appended to final RunOutput only
+                                  |
+                 receipt, stderr, final debug warning summary
+```
+
+### Attempts And Validation
+
+- `runProducerAttempts` promotes only the terminal accepted or terminal
+  rejected candidate's module and completed-validator warnings. Operational,
+  structural, semantic, and module-directed attempts that are superseded are
+  retained in attempt debug artifacts but not in the final collection.
+- A module-directed semantic-reconciliation retry adds its fallback warning
+  only when no retry remains. Earlier attempt warnings are discarded.
+- On `warn_continue`, warnings from the otherwise accepted candidate and
+  completed approved or rejected validators are retained. One fixed,
+  non-sensitive `validator_execution_incomplete` warning is added for every
+  failed validator. Skipped validators affect the validation summary and final
+  `incomplete` status but do not receive such a warning.
+- A semantic terminal rejection retains only warnings from that rejected
+  attempt. Structural rejection after producer failure cannot retain a
+  candidate warning because no valid candidate result exists.
+
+These behaviors are protected by the producer-attempt, extract-handoff,
+rejection-warning, normalize-retry, and attempt-debug tests. They are the right
+foundation for the redesign and should not be replaced with early-exit or
+all-attempt accumulation.
+
+### Concurrency And Ordering
+
+Extract jobs are dispatched chunk-first and lane-second. Results are stored by
+chunk index, finalized in ascending chunk order, and lane continuations are
+merged into a slice indexed by configured lane order. Pipeline steps run in
+configured order. The resulting public order is therefore:
+
+1. pre-run reference warnings;
+2. chunk-stage warnings;
+3. step order;
+4. configured lane order within each step;
+5. chunk order within extract;
+6. merge warnings; then
+7. normalize warnings; followed by any output-result warnings.
+
+`TestRunnerBoundsExtractJobsAndStabilizesReverseCompletion` exercises warning
+order under reversed completion. No completion-order leak was found.
+
+### Chunk Plans And Checkpoints
+
+- A reusable chunk plan stores producer warnings only. Current validators run
+  again, and their current warnings are appended. Warnings from a cached plan
+  candidate that fails current validation are discarded before regeneration.
+- Accepted extract, merge, and normalize checkpoints store the terminal
+  warnings for their stage. Reuse loads and appends those warnings once at the
+  same logical handoff. Tests compare fresh and resumed warning collections and
+  preserve their order.
+- Validation-incomplete accepted outputs are not reusable, preventing a later
+  run from silently treating incomplete validation as complete.
+- Required accepted-normalize hydration replays that normalize checkpoint's
+  warnings; checkpoint decisions separately expose that reuse occurred.
+
+The recommended redesign should keep fresh and resumed logical diagnostics
+equivalent. Whether a result was reused belongs in checkpoint provenance, not
+in the diagnostic grouping key; adding a `reused` distinction would fragment
+groups and make equivalent runs present differently.
+
+### Terminal Surfaces
+
+| Surface | Current content | Audience | Audit result |
+| --- | --- | --- | --- |
+| `RunOutput.Warnings` | Flat final slice | Framework and CLI | Canonical in-memory list, but lacks origin and bounds. |
+| Published `warnings.json` | Object containing the warnings passed into the output encoder | Durable consumers | Exact for the production JSON encoder unless the encoder itself returns warnings. |
+| `index.json` | Path to `warnings.json` | Durable consumers | Stable discovery path; no separate diagnostic-detail path. |
+| Run-result v1 | `warning_count = len(final RunOutput.Warnings)` | Subprocess callers | Count only; no group/occurrence distinction. |
+| Human stderr | `run completed with N warning(s)` | Operators | Count only and no direct detail path. Successful exit remains zero. |
+| Manifest | Validation and rejection summaries, no warning collection | Durable provenance | Correctly avoids duplicating the flat list. |
+| Debug summary `warnings.json` | Raw final warning array | Operators/developers | Includes final output-result warnings and can therefore differ from published `warnings.json`. |
+| Debug run report | Final warning count | Operators/developers | Same final slice length as receipt and stderr. |
+| Attempt/stage debug | Candidate-local warning detail and origin in path/envelope | Forensics | Sufficient to diagnose provenance, but debug capture is optional and is not a durable consumer contract. |
+
+## Empirical Measurements
+
+### Offline And Synthetic Runs
+
+| Scenario | Result | What it establishes |
+| --- | --- | --- |
+| Maintained complete D&D config and transcript with the repository's offline fake LLM | Approved, 10 normalized outputs, 0 rejected outputs, 12 warnings; receipt, stderr, and published file all reported 12 | An ordinary structurally successful workflow can be noisy without fallback or incomplete validation. |
+| Same complete run, grouped after publication | Three reason codes, seven exact `(reason, scope, message)` tuples, maximum exact-tuple repetition of three | The flat count materially overstates distinct operator conditions. Scope resets within chunks and does not identify origin. |
+| Maintained focused scene-description workflow | Approved, one normalized output, 0 warnings | The warning channel can be quiet when synthetic model text is lexically grounded. |
+| Generic warning publication contract | One warning reaches successful stderr, durable output, and debug summary | The ordinary pre-output path is consistent. |
+| NPC semantic-reconciliation candidate-limit fallback | No LLM call, all records preserved, one exhaustion warning, total warnings no greater than 20 | Fallback is bounded and materially different from routine normalization. |
+| `warn_continue` with two failed validators and one skipped validator | Validation status contains all three incomplete validators; warning slice contains two execution-incomplete records | Current warning count does not describe all incomplete validation. |
+| Retrying extract candidate | Two producer attempts; only the accepted attempt's one warning is final | Retry does not amplify abandoned warnings. |
+| Terminal semantic rejection | Only the final rejected attempt's operation and validator warnings are final | Rejection diagnostics are retained without retaining superseded warnings. |
+| Fresh versus reused extract checkpoint | Warning collections are deeply equal | Checkpoint replay does not itself amplify warnings. |
+| Spell normalizer with 21 unresolved entries | 20 records: 19 samples plus one omission record saying two additional warnings were omitted | `warning_count` is neither exact occurrence count nor distinct-condition count. |
+
+The focused audit tests passed in `internal/cli`,
+`internal/framework/pipeline`, the NPC-registry and spell normalizers, and all
+D&D packages.
+
+### Bounded Sample Review
+
+The complete offline D&D run produced:
+
+| Reason | Count | Sample | Review |
+| --- | ---: | --- | --- |
+| `location_not_near_source` | 4 | `Moon Gate` was absent from cited text | Correctly identifies deliberately unsupported fake output. Three records shared the same exact tuple because chunk and stage origin were lost. |
+| `location_occurrence_not_near_source` | 4 | A `Moon Gate` visit was absent from cited text | Correctly identifies the same unsupported registry-driven occurrence, but repeats the same operator concern across extraction and normalization. |
+| `scene_description_not_near_source` | 4 | A title or summary had no significant exact token in cited text | Mixed value. Generic `session scene` prose is ungrounded, while `Arrival` versus transcript `arrive` illustrates an expected lexical false positive. |
+
+This fake workflow is an integration fixture, not a model-quality benchmark.
+It nonetheless proves that the checks carry useful evidence while being too
+imprecise and repetitive to serve as one-warning-per-record operator alerts.
+
+### Production Evidence Still Needed
+
+The reported 85-warning run establishes that high volume occurs in practice,
+but the following remain unknown:
+
+- dominant production reason codes and stage/lane sources;
+- unique group count versus repeated occurrence count;
+- false-positive rate for each relatedness family;
+- how much volume comes from normalization observations versus advisories;
+- whether fresh and resumed production runs remain equivalent; and
+- a defensible numerical acceptance target.
+
+If further data is worthwhile, the operator can supply the v1 receipt,
+`warnings.json`, and manifest validation summaries without supplying transcript
+or lane artifacts. An initial privacy-preserving report should group by reason
+code and normalized scope family, count exact repeated tuples, and omit message
+text. Reviewing heuristic precision requires a separately approved bounded
+sample with its cited source context.
+
+## Classification Of Current Conditions
+
+| Target disposition | Current families | Result impact | Operator action | Durable placement |
+| --- | --- | --- | --- | --- |
+| **Warning** | Empty reference; validator failure or skip accepted under `warn_continue`; unavailable required scene classification; exhausted semantic reconciliation | A configured process completed under an allowed degraded or incomplete policy rather than completing normally | Correct reference/configuration, inspect provider/validator, or rerun | Actionable `warnings.json`, receipt summary, stderr summary, debug |
+| **Advisory** | Source-relatedness heuristics; unresolved spell or registry membership; guarded invalid semantic proposal; any future LLM-judged uncertainty or extraction-quality signal | Uncertain data quality or poor model proposal, but accepted data is structurally valid and deterministic guards prevented unsafe mutation | Optional model/source review; no routine action for every record | Durable diagnostic detail and debug; never a top-level warning |
+| **Observation** | Whitespace/name/ID/source-reference canonicalization; canonical ordering; exact and approved semantic duplicate consolidation | Successful intended normalization | None under normal operation | Durable bounded normalization diagnostics or debug; no stderr warning |
+| **Not a diagnostic** | Rejection, invalid structure, cancellation, persistence error, provider failure under fail-run policy | Candidate or run did not complete according to policy | Inspect rejection/error and retry or correct input/configuration | Existing rejection, validation summary, error, and debug contracts |
+
+Exact and semantic duplicate consolidation should remain distinguishable in
+category or reason metadata even though both are observations. Semantic
+reconciliation exhaustion remains a warning because a capability was not
+applied; successful approved consolidation is an observation because it is the
+normalizer's intended work.
+
+## Ranked Findings
+
+### AUD-WARN-001 — The Flat Warning Type Destroys Signal Quality
+
+- **Priority:** High operator impact; high implementation leverage.
+- **Evidence:** `contracts.Warning` has only scope, reason, and message. Routine
+  normalizer changes, heuristic doubt, fallback, and incomplete validation all
+  enter the same slice and the same CLI count. The offline complete run's 12
+  records were all advisories; the operator observed 85 records in a successful
+  real run.
+- **Impact:** Operators cannot tell whether a warning requires a rerun, a
+  configuration repair, optional review, or no action. Repeated routine output
+  trains them to ignore the channel.
+- **Recommendation:** Replace the flat semantic contract with explicit
+  `warning`, `advisory`, and `observation` dispositions plus a small category
+  vocabulary. Do not infer disposition from message text or require every
+  downstream consumer to maintain a reason-code policy table.
+
+### AUD-WARN-002 — Warning Records Lose The Origin Needed For Diagnosis And Aggregation
+
+- **Priority:** High correctness and usability impact.
+- **Evidence:** The runner knows stage, step, lane, module, validator, chunk ID,
+  and chunk index at promotion time, but `terminalWarnings` flattens module and
+  validator records into `[]contracts.Warning`. Per-chunk scopes such as
+  `locations[0]` and `occurrences[0]` then repeat without identifying their
+  chunk or producer. `source_references_normalized` is intentionally shared
+  across families and is therefore especially ambiguous.
+- **Impact:** `warnings.json` cannot answer which stage or module produced a
+  record. Message- or scope-based deduplication would merge unrelated findings
+  or retain accidental duplicates.
+- **Recommendation:** Keep module findings free of framework context, then have
+  the framework add a structured origin envelope before promotion. Validator
+  findings must retain validator identity instead of passing through
+  `validationReport.Warnings()` as a flat slice.
+
+### AUD-WARN-003 — Warning Volume Is Not End-To-End Bounded Or Validated
+
+- **Priority:** High robustness impact; medium immediate likelihood.
+- **Evidence:** D&D's `LimitWarnings` caps most individual producers at 20, but
+  NPC- and spell-relatedness return one warning per record without the helper.
+  Every extract validator is invoked per chunk, all ten relatedness checks run
+  again after normalization, and there is no run-level collector. The generic
+  contract validates neither disposition nor reason/message size, UTF-8,
+  blankness, or total records.
+- **Impact:** Warning memory and output grow with chunks, lanes, configured
+  validators, and record counts. Local omission records lose exact occurrence
+  semantics while still incrementing `warning_count`.
+- **Recommendation:** Add a generic bounded diagnostic collector that preserves
+  exact occurrence counts and bounded samples. Validate all diagnostic fields
+  at the module/framework boundary. Immediately bring NPC and spell
+  relatedness under the existing cap if the full redesign is staged.
+
+### AUD-WARN-004 — Output Encoder Warnings Make Durable Surfaces Disagree
+
+- **Priority:** Medium current impact; high contract correctness risk.
+- **Evidence:** `Runner.Run` passes existing warnings to `encoder.Encode`, then
+  the production encoder serializes `warnings.json`. Only after encoding does
+  the runner append `OutputResult.Warnings`. The receipt, stderr, debug summary,
+  and debug run report see the final slice; the already-created published file
+  cannot. No production encoder currently returns a warning, so ordinary JSON
+  runs do not trigger the defect.
+- **Impact:** A valid output-module implementation can violate the documented
+  claim that `warning_count` describes the published warning collection.
+- **Recommendation:** Remove successful output warnings from the output-module
+  contract unless a demonstrated use case requires them; encoding failures
+  should be errors and optional encoder observations should be debug data. A
+  two-phase finalize API is the viable but more complex alternative.
+
+### AUD-WARN-005 — Validation Skips Are Incomplete But Not Warned
+
+- **Priority:** Medium operator/correctness impact.
+- **Evidence:** `firstIncompleteValidation` treats failed and skipped validators
+  alike, and validation summaries include both. `incompleteValidationWarnings`
+  emits records only for `validationFailed`. The focused test demonstrates
+  three incomplete validators but two warnings.
+- **Impact:** A successful run can have `validation_status: incomplete` while
+  its warning count understates or even omits the affected validators. A caller
+  that checks only warnings receives a weaker signal than the manifest and
+  receipt status.
+- **Recommendation:** Produce one aggregated incomplete-validation warning
+  group whose occurrences cover both failure and skip, while retaining typed
+  outcome and safe reason metadata in the validation summary. Do not expose
+  provider errors or arbitrary skip prose in model or operator messages.
+
+### AUD-WARN-006 — Relatedness Checks Are Useful But Repetitive And Lexically Weak
+
+- **Priority:** Medium operator impact; low acceptance-policy urgency.
+- **Evidence:** Every family runs the advisory in both extract and normalize
+  chains. The complete fixture contains exact repeated tuples, and the checks
+  rely on exact normalized token sequences or a minimal significant-token
+  overlap. Reason naming drifts between `*_not_near_source` and
+  `*_source_unrelated`.
+- **Impact:** The checks can catch unsupported entities, but aliases, pronouns,
+  inflection, and generic scene prose create predictable false positives. Flat
+  per-record presentation magnifies them.
+- **Recommendation:** Retain the validators and their stage-local execution,
+  but classify and aggregate them as advisories. Normalize reason-code naming
+  when the diagnostic contract changes. Do not strengthen them into rejection
+  rules without a human-reviewed production evaluation.
+
+### AUD-WARN-007 — `warning_count` Has No Stable Operational Meaning
+
+- **Priority:** High downstream-contract impact.
+- **Evidence:** The receipt and CLI use `len(output.Warnings)`. One list element
+  can be an omission summary representing several hidden occurrences; repeated
+  records can represent the same condition; and skipped validators can be
+  absent. A 21-occurrence spell test produces a list length of 20.
+- **Impact:** The value is neither an exact occurrence count nor a distinct
+  warning-group count. Consumers cannot set policy or present a trustworthy
+  summary from it.
+- **Recommendation:** Introduce explicit warning-group and warning-occurrence
+  counts in a versioned receipt. Do not silently redefine the v1 field.
+
+## Recommended Target Contract And Presentation Model
+
+### Diagnostic Model
+
+Use one validated internal diagnostic model with these concepts:
+
+- **disposition:** `warning`, `advisory`, or `observation`;
+- **category:** a small enum such as `configuration`, `degradation`,
+  `validation_incomplete`, `data_quality`, `fallback`, or `normalization`;
+- **reason code:** stable semantic identity owned by the producer;
+- **origin:** framework-added phase/stage, step ID, lane ID, module key,
+  validator name, chunk ID, and chunk index when applicable;
+- **occurrence count:** exact number of matching findings;
+- **samples:** a small deterministic list of bounded scope/message pairs; and
+- **omitted sample count:** `occurrence_count - len(samples)`, represented as
+  metadata rather than another diagnostic record.
+
+Errors and rejected outputs must not become diagnostic dispositions. A warning
+means that the run completed under policy despite a process-level degradation
+or incomplete configured operation. Advisory and observation dispositions can
+describe accepted artifact quality and transformation provenance, but no
+LLM-judged extraction-quality signal may be promoted to a warning. Validation
+status remains authoritative for approval, rejection, and incomplete
+validation.
+
+### Aggregation
+
+The framework runner should own aggregation after it enriches findings with
+origin and before public output construction. Modules and validators retain
+semantic ownership of disposition, category, reason, scope, and message; they
+must not own CLI or file presentation.
+
+The default stable key should be:
+
+```text
+disposition + category + reason_code
++ phase/stage + step_id + lane_id + module_key + validator_name
+```
+
+Chunk, record scope, and message text belong in samples and must not be part of
+the group key. This groups repeated per-chunk findings without merging the same
+code across distinct producers or pipeline locations. Group order should be
+the first occurrence in the runner's existing canonical order; sample order
+should follow the same order. A final canonical sort by the complete origin key
+is also viable, but completion timing must never choose either order.
+
+Aggregation must be incremental and bounded. Producers should use a shared
+collector that counts every occurrence while retaining only bounded samples;
+the framework then merges producer groups without reconstructing counts from
+omission prose. A global maximum group count is also required, with overflow
+represented by structured aggregate metadata and with actionable groups given
+priority over lower dispositions.
+
+### Durable Files
+
+Keep one canonical home for each class:
+
+- `warnings.json` should contain versioned, grouped actionable warnings only;
+- a new `diagnostics.json` should contain versioned advisory and observation
+  groups only, avoiding duplication of warning groups;
+- `index.json` should link both files;
+- `rejected.json` and manifest validation summaries should retain their current
+  separate responsibilities; and
+- debug bundles should retain candidate-attempt detail plus the final grouped
+  projections.
+
+This is preferable to keeping all detail in `warnings.json` and filtering only
+the CLI: downstream consumers would otherwise continue to receive a semantically
+mixed warning contract, and routine observations would still dominate the
+durable file.
+
+### CLI And Receipt
+
+For a successful human run with actionable warnings, print a concise summary
+such as:
+
+```text
+notarius: run completed with 2 warning groups (7 occurrences); details=/.../warnings.json
+```
+
+Advisories and observations should not produce the warning line. Their durable
+path remains discoverable through `index.json`; a concise non-warning count can
+be added to the ordinary success line only if operator testing shows value. An
+ordinary successful run with no process degradation should write nothing to
+the warning stream even when it publishes quality advisories or normalization
+observations.
+
+Create `notarius.run-result.v2` rather than redefining v1. It should expose at
+least:
+
+- `warning_group_count`;
+- `warning_occurrence_count`; and
+- `diagnostic_group_count` for non-warning durable groups.
+
+The receipt should continue to expose validation status, validation summaries,
+and rejected-output count independently. Process exit behavior should not
+change as part of warning presentation reform.
+
+### Checkpoint Semantics
+
+Store the structured terminal diagnostic groups with accepted checkpoints and
+replay them exactly once at their logical stage. Fresh and reused runs should
+produce the same public groups and counts. Checkpoint events and debug records,
+not diagnostic identity, should disclose whether computation was reused.
+
+### Output Encoder Boundary
+
+Prefer removing `OutputResult.Warnings`. A successful output encoder should
+either return the complete logical files or fail. If future encoders genuinely
+need to produce durable post-encoding warnings, introduce an explicit
+two-phase prepare/finalize contract so those warnings can be included in the
+same published collection. Do not retain the current self-inconsistent
+one-pass capability.
+
+## Resolution Of Required Design Questions
+
+| Question | Recommendation | Viable alternative and tradeoff |
+| --- | --- | --- |
+| Explicit severity/disposition or external reason mapping? | Put validated disposition and category in the contract. | A central reason-code registry avoids payload fields but makes new modules depend on a second synchronized policy table and leaves downstream meaning implicit. |
+| Keep all detail in `warnings.json` or separate it? | Separate grouped actionable warnings from grouped advisories/observations in `diagnostics.json`. | Keep the flat durable list and aggregate only CLI output; simpler migration, but it preserves the noisy downstream contract and ambiguous count. |
+| Who owns aggregation? | Framework runner/coordinator after origin enrichment. | Output module aggregation keeps framework types smaller but duplicates policy across encoders and cannot repair missing validator origin. |
+| Stable aggregation key? | Disposition, category, reason code, and full producer origin; exclude chunk/scope/message. | Explicit producer-supplied grouping keys offer flexibility but add another identity that can drift from reason codes. Message-template grouping is brittle and unsafe. |
+| Samples and omissions? | Exact occurrence count plus deterministic bounded samples and numeric omitted-sample count. | Omission warning records preserve the current representation but inflate group counts and require prose parsing. |
+| `warning_count` semantics? | Version receipt and replace ambiguity with group and occurrence counts. | Keep v1 count as published record length and add optional fields; compatible, but two competing warning counts remain easy to misuse. |
+| Which normalization changes remain warnings? | Only exhausted process fallback. Unresolved membership is a data-quality advisory; successful canonicalization, reordering, ID repair, source-ref dedupe, and duplicate consolidation are observations. | Treat unresolved membership or semantic duplicate consolidation as warnings because they affect grounding or cardinality; more conservative, but it violates the process-only warning rule and reports accepted artifact quality as an operational failure. |
+| Source-relatedness disposition? | Grouped advisory by default; preserve current approve behavior. | Retain warning disposition or make rejection configurable. Rejection requires production precision evidence; current lexical rules are not strong enough. |
+| Checkpoint-loaded warnings? | Present the same logical groups as fresh execution and use checkpoint events for reuse provenance. | Mark groups as replayed; aids forensics but fragments aggregation and makes semantically equivalent runs differ. |
+| ADR and schema versions? | Add an ADR and version the run receipt and diagnostic files. | Treat the work as CLI-only presentation and avoid an ADR; insufficient because module contracts, output files, checkpoint payloads, and downstream fields change. |
+
+## Compatibility, Documentation, And ADR Implications
+
+The target alters public and internal contracts enough to require a new ADR.
+It should record:
+
+- the distinction among warnings, advisories, observations, rejections, and
+  errors;
+- the invariant that warnings are process-level signals, LLM-judged extraction
+  quality is never a warning, and ordinary non-degraded success has zero
+  warnings;
+- module semantic ownership versus framework origin/aggregation ownership;
+- bounded group and sample semantics;
+- fresh/checkpoint equivalence; and
+- the output-encoder decision.
+
+Implementation should introduce `notarius.run-result.v2`. The grouped warning
+and diagnostic envelopes should each carry their own schema version. Because
+the content of `warnings.json` changes incompatibly from a flat array wrapper
+to groups, release notes and the published JSON integration contract must call
+out the migration. `index.json` gains the diagnostic file path.
+
+Canonical documentation updates belong in:
+
+- `docs/cli.md` for stderr presentation only;
+- `docs/operations.md` for operator review and debug workflow;
+- `docs/integrations/json-output.md` for warning and diagnostic file schemas;
+- `docs/integrations/run-result.md` for v2 fields and compatibility;
+- `docs/consumers/subprocess.md` and `docs/consumers/dnd-pipeline.md` for
+  downstream policy checks;
+- `docs/internal/pipeline.md` for promotion, aggregation, retry, and checkpoint
+  mechanics;
+- `docs/internal/modules.md` and `docs/internal/dnd.md` for producer rules and
+  the D&D classification matrix; and
+- `docs/policy/architecture.md` for the durable ownership invariant after the
+  ADR is accepted and implemented.
+
+No configuration knob is required for the first implementation. A fixed,
+well-documented taxonomy is easier to reason about than per-reason display
+overrides. Configurable escalation or suppression can be considered only after
+production review demonstrates a concrete operator need.
+
+## Test Coverage Assessment
+
+Existing coverage worth preserving includes:
+
+- accepted-attempt and terminal-rejection warning promotion;
+- validator failure retry exhaustion and `warn_continue`;
+- module semantic retry fallback;
+- deterministic warning order under concurrent lane completion;
+- chunk-plan invalidation and discarded-cache warning behavior;
+- fresh/checkpoint warning equivalence;
+- local D&D warning caps and safe dynamic-message quoting;
+- JSON warning-file publication; and
+- CLI stderr, debug, and receipt counts.
+
+Material gaps are:
+
+- no bound test for NPC- or spell-relatedness warnings;
+- no generic warning-field or result-size validation;
+- no test for output encoder warnings versus published `warnings.json`;
+- no operator-level aggregation or bounded-sample tests;
+- no fresh/resume test for grouped counts because groups do not yet exist; and
+- no production evaluation of advisory precision.
+
+Tests should protect the semantic relationships: exact occurrence counts,
+bounded samples, deterministic group order, actionable-only warning
+presentation, and cross-surface equality. They should not assert one exact
+warning count for every complete D&D run or treat message wording as a public
+API unless the wording itself enforces a security boundary.
+
+## Audit Conclusion
+
+The application is in a good position for warning reform. Its retry,
+validation, checkpoint, and concurrency mechanics provide reliable points at
+which to attach structured diagnostics. The most valuable change is not to
+suppress individual reason codes; it is to replace the semantically flat,
+origin-free collection with bounded typed groups and to reserve the word
+“warning” for conditions that merit operator attention.
+
+Provider-backed runs would improve prioritization and help tune the D&D
+advisories, but they are not necessary to conclude that routine normalization
+and heuristic doubt should not dominate stderr or the durable warning
+contract. They should be gathered before changing heuristic acceptance policy
+or adopting a numerical production warning-volume target.

docs/roadmap/dnd-validation-reliability.md

@@ -0,0 +1,240 @@
+# D&D Validation Reliability Hardening
+
+## Purpose
+
+Improve the correctness and efficiency of D&D validation retries by giving the
+producer complete, contextual, semantically useful correction guidance; moving
+simple closed-value checks into compatible LLM-facing schemas; and removing
+duplicated source-range containment logic. The work should make one retry more
+likely to repair every detectable defect without weakening deterministic
+validation or exposing internal identifiers to a model.
+
+## Current State And Findings
+
+The validation architecture is coherent and the major safety boundaries are in
+place. Validators run through the framework-managed retry protocol, model-facing
+guidance is distinct from reason codes and operator diagnostics, schemas use
+provider-compatible required fields, and D&D extraction adapters canonically
+order safely resolvable source-reference endpoints before validation. The
+review nevertheless found four opportunities to make that architecture more
+consistent and effective.
+
+### Generic correction guidance
+
+Most rejecting D&D validators produce detailed operator-facing messages but
+pair them with a fixed, general correction sentence. The framework correctly
+constructs retry text from `CorrectionGuidance` alone, so the model does not see
+the internal diagnostic—and therefore often does not learn which records were
+wrong or how each record must change. Item-occurrence shape validation and the
+combat-semantics validator already demonstrate the intended contextual pattern.
+
+Passing the operator message through unchanged is not an acceptable fix.
+Operator messages may contain Go paths, array indexes that do not reliably map
+back to canonicalized model output, reason codes, validator terminology, or
+opaque application identities. Those details are useful for diagnosis but are
+not appropriate model context.
+
+### Missing closed-value schema constraints
+
+Four private LLM response schemas describe categorical fields as unconstrained
+strings even though their prompts, durable artifact schemas, and deterministic
+validators define closed value sets:
+
+- `combat-turns`: `turn_kind`;
+- `enemy-events`: `kind`;
+- `npc-occurrences`: `kind`; and
+- `scene-descriptions`: `kind`.
+
+This defers an inexpensive structural check until the outer semantic-validation
+loop. Item and location occurrences already use the preferable private-schema
+enum pattern.
+
+### First-defect validation
+
+Spell shape validation returns after the first invalid cast field, and enemy
+engagement validation returns after the first duplicated engagement subject.
+When multiple defects exist, each retry can therefore reveal only one of them.
+That needlessly consumes the bounded producer-attempt budget and makes
+correction less reliable for smaller models.
+
+### Duplicated source-range containment
+
+Nine extraction source-reference validators independently determine whether a
+citation belongs to the current chunk. Five validate every source unit spanned
+by a reference, while four check only the endpoints. Those behaviors are
+equivalent for today's contiguous materialized chunks, but the duplicated
+implementations and different apparent contracts create a drift risk.
+Scene-description validation has a separate exact-one-scene range contract and
+should remain specialized.
+
+## Target End State
+
+- Every producer-correctable D&D rejection supplies bounded contextual guidance
+  that identifies all affected records using transcript-grounded names, source
+  ranges, and response fields as appropriate, explains each semantic defect,
+  and requests one complete corrected replacement.
+- Model-facing guidance never contains reason codes, validator keys, Go field
+  paths, opaque or hash-derived entity IDs, or raw internal error text.
+  Operator diagnostics remain detailed and separately available.
+- Every detectable issue in a candidate is collected in one validator pass,
+  subject to the repository's established diagnostic bounds. A retry is not
+  spent merely to reveal the next defect.
+- Closed categorical fields are constrained by compatible enums in the private
+  LLM schemas. Deterministic typed validators retain the same rules as
+  defense-in-depth and as protection for non-LLM producers and later stages.
+- One D&D-owned helper implements full-span chunk containment for extraction
+  evidence. All applicable source-reference validators use it and no
+  module-local endpoint-only or coverage implementation remains.
+- Validator results remain immutable, deterministic, bounded, and suitable for
+  the existing feedback-aware replacement-request protocol. Retry budgets,
+  warning policy, durable artifact schemas, and accepted output semantics do
+  not change.
+
+## Required Work
+
+### Contextual correction guidance
+
+Audit each rejecting D&D validator according to the stage that can actually
+correct its result:
+
+- Extraction validators should build model-facing issue descriptions from the
+  candidate and its transcript context. Shape, registry-membership, identity,
+  catalog, engagement, and source-reference validators should name the
+  contextual artifact and cited range where useful, state the invalid value or
+  relationship in plain language, and state the required replacement shape.
+- LLM-backed NPC, item, and location registry reconciliation should continue to
+  translate its typed proposal issues through the shared category-to-prose
+  renderer, including any domain-specific supplement.
+- Normalization-only deterministic invariant validators should retain useful
+  operator diagnostics. They should not claim that a producer can directly
+  repair deterministically derived ordering, identity, or normalization state.
+  If such a rejection can reach a feedback-capable producer, guidance must be
+  expressed only in terms of the source candidate that producer controls.
+- LLM-backed semantic validators may continue using their verdict explanation
+  when it is bounded and semantically meaningful.
+
+Introduce a small D&D-shared diagnostic utility only for demonstrated common
+mechanics such as stable grouping, de-duplication, contextual source-range
+rendering, and bounded correction aggregation. Domain validators must continue
+to own the meaning and prose of their rules. The generic pipeline must not gain
+D&D knowledge, and operator `Message` values must never be mechanically copied
+into `CorrectionGuidance`.
+
+Update validator policy fingerprints wherever correction behavior changes so
+checkpoints created under generic feedback are not reused as though the policy
+were identical.
+
+### Private-schema enum guardrails
+
+Add the existing supported value sets to the four private LLM schemas:
+
+- `assets/dnd/combat-turns/schemas/dnd_combat_turns_llm.v1.json`;
+- `assets/dnd/enemy-events/schemas/dnd_enemy_events_llm.v1.json`;
+- `assets/dnd/npc-occurrences/schemas/dnd_npc_occurrences_llm.v1.json`; and
+- `assets/dnd/scene-descriptions/schemas/dnd_scene_descriptions_llm.v1.json`.
+
+Use the exact values owned by the corresponding durable contract and typed
+domain constants. Keep all object properties required, retain current nullable
+types where present, and do not introduce `uniqueItems`, optional properties,
+or conditional schema logic. Revise schema tests that currently accept unknown
+values, and rely on computed asset fingerprints to invalidate incompatible
+LLM-output checkpoints. Keep deterministic enum validation in place.
+
+### Complete per-attempt issue collection
+
+Refactor spell shape and enemy engagement validation to inspect the complete
+candidate and collect every detectable violation before returning. Build the
+operator diagnostic and contextual correction request from the same evaluated
+issue set while preserving their different audiences. De-duplicate repeated
+semantic instructions, retain enough contextual identification for every
+affected record, and use the established bounded diagnostic behavior rather
+than an unbounded error string.
+
+Review the surrounding D&D validators while applying the contextual-guidance
+change. Remove any additional accidental early exits that prevent independent
+candidate defects from being reported together, but retain immediate returns
+for request-level prerequisites whose absence makes further inspection unsafe
+or meaningless.
+
+### Shared full-span chunk containment
+
+Add one helper under `internal/modules/dnd/shared` that determines whether a
+source reference's complete document-ordered span is contained by a chunk. The
+helper must:
+
+- validate source identity;
+- resolve endpoints through `source.DocumentIndex` rather than numeric-ID
+  assumptions;
+- require every unit in the inclusive span to be present in the chunk;
+- handle nil or unresolved inputs without panic;
+- avoid mutating the source, chunk, or reference; and
+- leave source-reference validity and error wording to the consuming validator.
+
+Use the helper from the spell, NPC-registry, NPC-occurrence, item-registry,
+item-occurrence, location-registry, location-occurrence, combat-turn, and
+enemy-event source-reference validators. Keep the scene-description exact-range
+validator separate because it enforces a materially different contract.
+
+## Testing And Documentation
+
+Add lean offline behavioral coverage at the narrowest stable boundary:
+
+- correction tests should prove that multiple contextual defects produce one
+  actionable, bounded request and that internal identifiers and diagnostic
+  syntax are absent; they should not snapshot exact prose or message length;
+- schema tests should prove rejection of representative unsupported categorical
+  values and acceptance of the supported sets without duplicating every
+  provider behavior;
+- spell-shape and engagement tests should prove that independent defects are
+  reported together;
+- the shared containment helper should own the full case matrix, including
+  partial spans, non-monotonic unit IDs, wrong sources, unresolved endpoints,
+  and nil inputs; consuming validators need only enough coverage to prove they
+  use the common policy; and
+- existing validator tests should be simplified when the shared helper makes
+  module-local cases redundant.
+
+Update `docs/internal/dnd.md` to document the durable conventions for
+contextual correction guidance, complete issue collection, private-schema enum
+guardrails, and shared full-span extraction containment. Update
+`docs/internal/pipeline.md` only if framework behavior changes; the intended
+work applies its existing contract and should normally require only a link or
+no change. No new ADR is required because ADR-0014 and the architecture policy
+already decide the separation between semantic guidance and internal
+diagnostics. Create an ADR only if implementation requires changing that
+framework-level decision.
+
+## Non-Goals
+
+- Passing operator diagnostics or internal errors directly to an LLM.
+- Exposing durable IDs, hashes, reason codes, validator keys, or raw provider
+  responses in correction text.
+- Changing stage retry counts, PromptKit structural-repair budgets, validator
+  failure policy, or warning classification.
+- Silently repairing domain-semantic defects or weakening deterministic
+  validators after adding schema guardrails.
+- Adding provider-sensitive schema constructs beyond simple enums.
+- Moving D&D-specific behavior into the generic pipeline framework.
+- Generalizing the scene-description exact-range contract into the shared
+  extraction containment helper.
+
+## Acceptance Criteria
+
+- Every producer-correctable D&D rejection reviewed in this work gives the next
+  attempt all currently detectable, actionable corrections in contextual prose.
+- No model-facing correction request contains an opaque entity ID, hash,
+  validator key, reason code, Go-style field path, or unfiltered operator error.
+- Spell shape and enemy engagement validation aggregate independent defects in
+  one pass and keep outputs immutable.
+- The four private LLM schemas reject unsupported categorical values, retain
+  provider-compatible required-only shapes, and leave durable v1 contracts
+  unchanged.
+- All nine applicable extraction source-reference validators use one D&D-shared
+  full-span containment policy; scene descriptions retain their specialized
+  exact-range check.
+- Validator and asset fingerprints change wherever their effective policy
+  changes, preventing reuse of stale checkpoints.
+- Canonical internal documentation records the conventions future D&D
+  validators must follow.
+- Focused tests, `go test ./...`, `go vet ./...`, and
+  `go build ./cmd/notarius` pass in a supported development environment.

docs/roadmap/future.md

@@ -10,77 +10,12 @@ not as committed release dates.
 PromptKit now owns structural output repair within one completion. Notarius
 owns stage candidates, validator chains, semantic rejection policy, bounded
 feedback-aware stage retries, validation provenance, and reusable-state
-eligibility. The remaining near-term work applies those completed foundations
-to domain review and operator-facing diagnostics.
+eligibility, and the separation of actionable process warnings from quality
+diagnostics. The remaining near-term work applies those completed foundations
+to domain review and empirical evaluation.
 
-### D&D Combat Scene Semantic Validation
-
-- Add an optional production LLM-backed D&D validator that determines whether
-  proposed scene boundaries and classifications represent substantive active
-  combat correctly. Its central quality goal is that active combat is kept in
-  coherent scenes classified as `combat`, rather than split incorrectly or
-  hidden inside scenes classified as `narrative`, `recap`, or `meta`.
-- Resolve the validator's exact target before implementation. The current
-  `dnd/scenes` chunker owns only complete, gap-free source ranges, while the
-  per-chunk `dnd/scene-descriptions` extractor owns the `combat`, `narrative`,
-  `recap`, and `meta` classification. The preferred initial placement is
-  therefore an extract-stage validator for `dnd/scene-descriptions`, where it
-  can compare one proposed kind with the corresponding transcript chunk.
-- Consider a chunk-stage LLM validator only for a distinct boundary-coherence
-  question that can be answered from the complete transcript and proposed
-  range map, such as whether one continuous combat was fragmented across
-  inappropriate scene boundaries. Do not duplicate the same classification
-  judgment at both stages. Moving classification into chunk-plan annotations
-  would change the deliberately minimal, annotation-free chunk contract and
-  requires an explicit architecture review before it is selected.
-- Validate both false negatives and false positives: a non-combat kind must not
-  omit substantive active combat, and a combat kind must be supported by such
-  combat. Keep the existing deterministic downstream rule that combat-turn
-  extraction runs only for an exact `combat` scene classification; semantic
-  review improves the upstream classification but does not replace that gate.
-- Run the semantic validator through PromptKit, use a minimal required-field
-  structured response schema, and let PromptKit repair structural validator
-  output within its bounded budget. A contract-invalid final validator response
-  is a validator execution failure, not a semantic rejection and not a reason
-  to recursively validate the validator.
-- Evaluate the prompt and decision policy against a small human-reviewed set
-  containing combat setup, active turns, interruptions, multi-phase encounters,
-  brief rules discussion, aftermath, recalled combat, and false-positive
-  hostile dialogue. Measure false acceptance, false rejection, retry success,
-  added calls, latency, and token cost before placing it in the production
-  default chain.
-- An ADR is not required if classification remains owned by
-  `dnd/scene-descriptions` and the validator follows the generic validation ADR.
-  Create or supersede an ADR if the work transfers scene classification into
-  the chunker or otherwise changes stage ownership or the durable chunk-plan
-  contract.
-
-### Warning Signal And Presentation Reform
-
-- Audit every warning producer and representative successful runs. Ordinary
-  success producing dozens of warnings is a failed operator experience: the
-  volume obscures actionable problems and trains operators to ignore the
-  warning channel.
-- Define a small warning taxonomy that distinguishes actionable degradation,
-  incomplete validation, lossy fallback, and data-quality risk from routine
-  normalization observations or informational diagnostics. Preserve detailed
-  traceability in debug or manifest data without promoting every observation
-  to a top-level CLI warning.
-- Consider stable deduplication and aggregation by scope and reason code,
-  bounded samples plus omitted counts, and a concise CLI summary with a path to
-  detailed diagnostics. Do not suppress genuine validator execution failures
-  merely to reduce the count.
-- Decide which warnings affect process status, rejection summaries, durable run
-  receipts, or only debug output. Ensure warning ordering and aggregation are
-  deterministic across concurrent execution.
-- Establish a representative warning-volume acceptance target and human review
-  workflow before changing individual producers piecemeal. The intended result
-  is not zero warnings; it is a small set in which every surfaced warning merits
-  operator attention.
-- This work does not require an ADR unless it changes validation acceptance,
-  failure, or durable contract semantics. CLI presentation and diagnostic
-  taxonomy otherwise belong in a feature roadmap followed by updates to their
-  canonical configuration, operations, integration, and internal documents.
+The active D&D reliability work is defined by
+[Feedback-Aware Semantic-Reconciliation Retries](semantic-reconciliation-retry-feedback.md).
 
 ## Near-Term D&D Pipeline
 

docs/roadmap/item-occurrence-holder-reliability.md

@@ -0,0 +1,95 @@
+# D&D Item-Occurrence Holder Reliability
+
+## Purpose
+
+Improve item-occurrence extraction reliability for smaller models by making
+holder rules easier to follow and semantic retry feedback specific enough to
+correct a rejected response.
+
+## Problem
+
+The item-occurrence prompt correctly defines the permitted `from` and `to`
+values for every occurrence kind, but presents the rules in dense prose. The
+private schema requires both nullable fields but cannot express their
+cross-field relationship. The deterministic shape validator correctly rejects
+incompatible combinations, yet its model-facing correction guidance only asks
+for generally valid holder combinations. Notarius therefore resends the
+defective response without telling the model which contextual records failed or
+what their corrected holder shape must be.
+
+## Target State
+
+- The module instructions present one compact, unambiguous holder matrix:
+  `discovered` uses two nulls; `acquired` uses a null `from` and non-null `to`;
+  `lost` and `consumed` use a non-null `from` and null `to`; and `transferred`
+  uses two distinct named party members.
+- The private LLM schema continues requiring `quantity`, `from`, and `to` with
+  nullable types, and constrains `kind` to the five supported values. The
+  durable v1 artifact contract remains unchanged.
+- A holder-combination rejection supplies bounded, semantically meaningful
+  correction guidance that identifies affected occurrences by contextual item
+  name and transcript evidence where useful, states the required JSON
+  null/non-null shape, and requests a complete corrected replacement.
+- Correction guidance never exposes durable item IDs, hashes, validator keys,
+  reason codes, or other opaque implementation identifiers. It does not rely
+  solely on mapped array indexes, because canonical extraction ordering may
+  differ from the raw response order appended to the retry prompt.
+- Operator-facing validation messages remain specific and include enough
+  information to distinguish a missing holder from an extra or misplaced
+  holder.
+- The existing validator remains strict. Holder combinations are not silently
+  rewritten: unlike reversed evidence endpoints, changing a holder can conceal
+  a misclassified discovery, acquisition, loss, consumption, or transfer.
+
+## Required Work
+
+1. Rewrite the holder paragraph in
+   `assets/dnd/item-occurrences/prompts/instructions.md` as a compact matrix,
+   retaining the existing classification and currency rules without adding
+   contradictory repetition.
+2. Add the five supported `kind` values as an enum in the private item-occurrence
+   response schema. Keep every listed field required and keep `quantity`,
+   `from`, and `to` explicitly nullable.
+3. Refactor the item-occurrence shape validator to build detailed bounded
+   operator issues and separate actionable correction guidance from the same
+   evaluated candidate. Group repeated holder failures where practical while
+   retaining contextual record identification and every distinct correction
+   rule needed by the model.
+4. Preserve the generic validation retry protocol: the complete original
+   prompt, exact defective assistant response, and one user correction message.
+   Do not add a module-local retry loop.
+5. Ensure the changed prompt and schema produce new computed fingerprints, and
+   bump the validator policy fingerprint so stale checkpoints cannot be reused
+   across the behavior change. Update internal documentation where needed; the
+   external item-occurrence contract requires no semantic change.
+6. Add lean offline regression coverage for the holder matrix, schema enum,
+   contextual correction guidance, bounded aggregation, absence of opaque IDs,
+   and a rejected-then-corrected extraction attempt. Avoid exact prose snapshots
+   and live-provider tests.
+
+## Non-Goals
+
+- Automatically deleting, moving, or inferring holder values.
+- Changing the durable item-occurrence schema or event taxonomy.
+- Encoding the complete semantic holder matrix through provider-sensitive
+  conditional JSON Schema constructs.
+- Adding warnings for ordinary semantic rejection or successful correction.
+- Tuning unrelated D&D extraction prompts or validators.
+
+## Acceptance Criteria
+
+- The maintained item-occurrence prompt shows the clear holder matrix,
+  and the private schema rejects unsupported kind strings while remaining
+  compatible with strict structured-output providers.
+- A rejected `acquired` occurrence receives guidance that explicitly requires
+  `from: null` and a non-null `to`; equivalent exact guidance exists for the
+  other four kinds.
+- Multiple invalid occurrences yield one bounded correction request containing
+  all distinct required fixes and contextual identifiers that can be matched to
+  the appended raw response.
+- No correction message contains an item ID, hash, validator key, or reason
+  code, and the raw model candidate remains byte-faithful.
+- Valid candidates and the durable output contract are unchanged; invalid
+  holder combinations still fail after configured retries are exhausted.
+- Focused tests, `go test ./...`, `go vet ./...`, and `go build ./cmd/notarius`
+  pass.

docs/roadmap/semantic-reconciliation-retry-feedback.md

@@ -0,0 +1,117 @@
+# Feedback-Aware Semantic-Reconciliation Retries
+
+## Purpose
+
+Improve semantic registry reconciliation for smaller models by turning the
+existing module-requested retry into a bounded corrective exchange. Preserve
+the current deterministic safety checks, shared retry budget, and safe fallback
+when correction is unsuccessful.
+
+## Problem
+
+NPC, item, and location registry normalizers can reject unsafe duplicate-group
+proposals and retain a safe partial result. They currently ask the pipeline for
+another attempt, but the next attempt receives the original prompt without the
+defective response or an explanation of the defect. The normalizers also build
+their retry messages from internal issue-category strings, which are suitable
+for diagnostics but not useful model instructions. If all attempts are
+exhausted, the warning reports the omitted-group count without explaining the
+final defect or confirming that affected candidates were preserved separately.
+
+## Target State
+
+- A normalizer retry directive may carry model-facing correction guidance
+  separately from its stable reason code, operator message, and fallback
+  diagnostics.
+- When the directive includes correction guidance and a valid
+  `single_response_v1` model candidate, the shared producer-attempt state
+  machine reconstructs the complete original prompt and appends the exact
+  defective response as an assistant message followed by one user correction
+  message. Earlier failed attempts do not accumulate.
+- This correction consumes the existing normalize-stage retry budget. It does
+  not add a module-local loop or alter PromptKit's independent structured-output
+  repair budget.
+- A retry caused by structurally invalid output remains a fresh stage attempt
+  without semantic feedback when no valid model candidate is available.
+- Internal issue categories remain available for operator and debug
+  diagnostics, but are never sent to a model. A shared renderer translates
+  every known category into concise prose that explains the defect and the
+  required valid shape.
+- Item reconciliation appends its domain rule when a proposal attempts to
+  combine different currency denominations or currency with a non-currency
+  item. NPC and location reconciliation use only the shared protocol guidance.
+- On final exhaustion, safe proposal groups remain applied, unsafe groups
+  remain separate, and one process warning gives a bounded summary of the final
+  defect and explicitly states that the affected candidates were preserved
+  separately.
+
+## Required Work
+
+1. Extend `contracts.NormalizeRetry` and its clone, validation, and debug paths
+   with an optional, bounded model-facing correction-guidance field. Keep
+   reason codes and operator messages out of model input. Require guidance to
+   be paired with a valid `single_response_v1` model candidate; preserve the
+   existing feedback-free directive for structural retry cases.
+2. Extend the generic producer-attempt state machine so a feedback-capable
+   module retry constructs `SemanticCorrection` from the latest exact response
+   and the supplied guidance. Preserve attempt provenance, cancellation,
+   terminal policy, diagnostic handling, byte-faithful response provenance,
+   and the rule that only the latest response is included.
+3. Add a model-facing renderer to `internal/framework/semanticreconcile`.
+   Implement it with a typed lookup table keyed by `IssueCategory` and an
+   authoritative list of all categories, rather than a positional array.
+   Render response-local duplicate-group numbers in human-friendly form,
+   de-duplicate repeated guidance, enforce a size bound, and return an error for
+   any unmapped category instead of exposing its raw value. Retain
+   `IssueDetails` for operator/debug diagnostics.
+4. Update the NPC, item, and location registry normalizers to populate the new
+   correction guidance for semantically invalid proposals. Add the item-only
+   currency guidance after the shared prose when typed application rejects a
+   group. Do not add guidance to the invalid-structured-output path when no
+   exact valid candidate exists.
+5. Improve exhausted-fallback diagnostics across all three registries. Describe
+   the final proposal failure accurately, state that affected candidates were
+   preserved separately, and keep warning samples bounded and free of raw model
+   responses.
+6. Update the shared semantic-reconciliation and affected normalizer policy
+   fingerprints so checkpoints produced under blind retry behavior are not
+   reused. Record the durable module-retry correction decision in a new ADR,
+   and update the architecture, pipeline, LLM-runtime, and D&D internal docs in
+   their canonical scopes when the behavior is implemented.
+7. Add lean offline behavioral coverage for corrective module retries, exact
+   latest-response use, feedback-free structural retries, contract validation,
+   complete category-to-prose coverage, domain-specific item guidance, bounded
+   messages, safe final fallback, and warning content. Avoid exact prose
+   snapshots, prompt-length assertions, live providers, and redundant tests of
+   PromptKit internals.
+
+## Non-Goals
+
+- Changing normalize retry counts, PromptKit repair settings, or pipeline
+  terminal policies.
+- Changing the reconciliation prompt protocol, response schema, registry
+  artifact schemas, or request-local candidate IDs.
+- Reintroducing provider-sensitive schema constraints such as `uniqueItems`.
+- Automatically applying an unsafe duplicate proposal or treating a fallback
+  warning as an extraction-quality warning.
+- Persisting raw model responses or correction text in ordinary warnings,
+  receipts, manifests, caches, or checkpoints.
+
+## Acceptance Criteria
+
+- A semantic proposal rejected by an NPC, item, or location registry normalizer
+  receives the exact defective response and actionable prose on the next
+  normalize attempt, using only the configured stage retry budget.
+- Every `IssueCategory` has non-empty model-facing guidance, no raw category or
+  reason-code string reaches the model, and an unmapped future category fails
+  closed.
+- An invalid item consolidation additionally tells the model to keep currency
+  denominations distinct and not mix currency with non-currency items.
+- Structural-output failures without a valid model candidate retain the current
+  fresh-retry behavior.
+- After retry exhaustion, safe consolidations remain applied, affected
+  candidates remain separate, and the single fallback warning explains both
+  the final cause and preservation behavior without including raw model output.
+- Checkpoint identities change where policy changed, canonical documentation
+  and a new ADR describe the implemented behavior, and focused tests plus
+  `go test ./...`, `go vet ./...`, and `go build ./cmd/notarius` pass.

docs/roadmap/source-reference-canonicalization.md

@@ -0,0 +1,96 @@
+# D&D Source-Reference Endpoint Canonicalization
+
+## Purpose
+
+Make model-supplied D&D evidence ranges resilient to reversed endpoints without
+spending producer retries on a losslessly repairable representation error.
+
+## Problem
+
+The shared extraction-evidence prompt requires `start_unit_id` and
+`end_unit_id`, but does not state that start means the earlier unit in
+transcript order. The private LLM schemas cannot express that cross-field
+relationship. Current extraction adapters sort and de-duplicate ranges but
+preserve reversed endpoints, so deterministic source-reference validators
+reject otherwise useful candidates and may exhaust the producer retry budget.
+
+## Target State
+
+- The shared D&D extraction-evidence prompt states that `start_unit_id` is the
+  earlier endpoint and `end_unit_id` is the same or a later endpoint according
+  to transcript order.
+- The existing LLM-facing and durable `start_unit_id`/`end_unit_id` shapes stay
+  unchanged. No unordered-pair schema or compatibility migration is introduced.
+- One helper owned by `internal/modules/dnd/shared` orders the endpoints of a
+  source reference using `source.DocumentIndex` positions. It never assumes
+  that numerically smaller unit IDs occur earlier.
+- The helper swaps endpoints only when the reference belongs to the current
+  source and both endpoint IDs resolve in that source. Forward and single-unit
+  ranges remain unchanged. Wrong-source, missing, non-positive, or otherwise
+  unresolved endpoints remain unchanged for deterministic validators to reject.
+- Every D&D extraction adapter that consumes model-supplied evidence ranges
+  applies this endpoint operation before its existing range sorting,
+  de-duplication, earliest-evidence calculation, and validation. The covered
+  artifact families are spells, NPC registry, NPC occurrences, item registry,
+  item occurrences, location registry, location occurrences, combat turns, and
+  enemy events.
+- Successful endpoint swapping is ordinary deterministic canonicalization. It
+  does not emit a warning, quality diagnostic, or retry request.
+- The exact raw model response remains available through existing model-candidate
+  and debug provenance. Only the typed candidate presented to validators and
+  later stages receives the canonical endpoint order.
+- Source-reference validators remain strict and continue rejecting reversed
+  ranges that reach them from another producer or indicate an application bug.
+  Normalizers do not become a fallback repair boundary for invalid extraction
+  candidates.
+
+## Required Work
+
+1. Clarify endpoint ordering in
+   `assets/dnd/shared/prompts/common-dnd-extraction-evidence.md`, preserving one
+   byte-identical shared instruction for every consuming prompt.
+2. Extend the shared source-reference ordering utility with a mutation-safe
+   endpoint-ordering operation that uses document position and preserves
+   unresolvable references.
+3. Integrate the shared operation into all nine model-response-to-candidate
+   mappings listed above. Do not duplicate endpoint comparison logic in the
+   individual artifact packages.
+4. Update `docs/internal/dnd.md` to distinguish lossless endpoint ordering from
+   malformed-reference validation. External artifact documentation and schemas
+   require no change because their contract already requires ordered ranges.
+5. Add lean regression coverage for forward, reversed, single-unit,
+   non-monotonic-ID, wrong-source, and unresolved endpoint cases. Add enough
+   adapter-level coverage to prove the shared operation is wired into every
+   affected extraction family without duplicating the helper's full case
+   matrix in each package.
+6. Update prompt-asset and fingerprint expectations affected by the shared
+   prompt revision, while avoiding exact token-, byte-length-, or prose-snapshot
+   change-detector tests.
+
+## Non-Goals
+
+- Changing durable source-reference schemas or replacing named endpoints with
+  an unordered LLM response shape.
+- Weakening source-reference validation or silently repairing missing,
+  out-of-source, or out-of-chunk evidence.
+- Applying this policy to scene chunk-plan ranges, which are not extraction
+  evidence references and have a separate ownership and validation boundary.
+- Adding warnings or durable normalization observations for routine endpoint
+  ordering.
+
+## Acceptance Criteria
+
+- A D&D extraction candidate whose two resolvable evidence endpoints are
+  reversed reaches validation as the equivalent forward range without a
+  producer retry.
+- Ordering follows document position even when unit IDs are not numerically
+  increasing.
+- Every affected D&D extractor uses the shared implementation; no module-local
+  endpoint-swap implementation remains.
+- References that cannot be safely ordered remain unchanged and are rejected by
+  the existing deterministic validators where applicable.
+- Raw model-candidate provenance remains byte-faithful to the provider response.
+- LLM-facing and durable schemas, validator strictness, module keys, default
+  chains, and output contracts remain unchanged.
+- Focused package tests, repository-wide tests, `go vet ./...`, and
+  `go build ./cmd/notarius` pass.

internal/cli/assembled_spell_pipeline_contract_test.go

@@ -71,28 +71,39 @@ func TestAssembledSpellPipelineNormalizesMergedCasts(t *testing.T) {
 		t.Fatalf("distinct cast = %#v, want separate evidence event", distinct)
 	}
 
-	wantWarningReasons := []string{
+	wantDiagnosticReasons := []string{
 		spellnormalize.ReasonCodeSpellNameCanonicalized,
 		spellnormalize.ReasonCodeSourceReferencesNormalized,
 		spellnormalize.ReasonCodeDuplicateSpellCastCollapsed,
 		"spell_not_near_source",
 	}
-	gotWarningReasons := make([]string, len(output.Warnings))
-	for index, warning := range output.Warnings {
-		gotWarningReasons[index] = warning.ReasonCode
+	gotDiagnosticReasons := make([]string, len(output.Diagnostics.Groups))
+	for index, group := range output.Diagnostics.Groups {
+		gotDiagnosticReasons[index] = group.ReasonCode
 	}
-	if !reflect.DeepEqual(gotWarningReasons, wantWarningReasons) {
-		t.Fatalf("warnings = %#v, want deterministic normalize and validation warnings", output.Warnings)
+	if !reflect.DeepEqual(gotDiagnosticReasons, wantDiagnosticReasons) {
+		t.Fatalf("diagnostics = %#v, want deterministic normalize and validation diagnostics", output.Diagnostics)
 	}
-	if output.Warnings[2].Scope != "spell_casts[0]" || !strings.Contains(output.Warnings[2].Message, "retained input index 0") || !strings.Contains(output.Warnings[2].Message, "removed input indices [1]") {
-		t.Fatalf("duplicate warning = %#v, want retained and removed merged indices", output.Warnings[2])
+	if output.Diagnostics.Groups[2].Samples[0].Scope != "spell_casts[0]" || !strings.Contains(output.Diagnostics.Groups[2].Samples[0].Message, "retained input index 0") || !strings.Contains(output.Diagnostics.Groups[2].Samples[0].Message, "removed input indices [1]") {
+		t.Fatalf("duplicate diagnostic = %#v, want retained and removed merged indices", output.Diagnostics.Groups[2])
 	}
 
 	warningsFile := decodeAssembledOutput[struct {
-		Warnings []contracts.Warning `json:"warnings"`
+		Groups []contracts.DiagnosticGroup `json:"groups"`
 	}](t, output.OutputFiles, "warnings.json")
-	if !reflect.DeepEqual(warningsFile.Warnings, output.Warnings) {
-		t.Fatalf("warnings file = %#v, run warnings = %#v, want manifest output path to preserve warnings", warningsFile.Warnings, output.Warnings)
+	if len(warningsFile.Groups) != 0 {
+		t.Fatalf("warnings file = %#v, want no process warnings for advisory-only diagnostics", warningsFile.Groups)
+	}
+	diagnosticsFile := decodeAssembledOutput[struct {
+		SchemaVersion                string                      `json:"schema_version"`
+		GroupCount                   int                         `json:"group_count"`
+		OccurrenceCount              int                         `json:"occurrence_count"`
+		Truncated                    bool                        `json:"truncated"`
+		UnrepresentedOccurrenceCount int                         `json:"unrepresented_occurrence_count"`
+		Groups                       []contracts.DiagnosticGroup `json:"groups"`
+	}](t, output.OutputFiles, "diagnostics.json")
+	if diagnosticsFile.SchemaVersion != "notarius.diagnostics.v1" || diagnosticsFile.GroupCount != len(output.Diagnostics.Groups) || !reflect.DeepEqual(diagnosticsFile.Groups, output.Diagnostics.Groups) || diagnosticsFile.OccurrenceCount != diagnosticGroupOccurrences(output.Diagnostics.Groups)+output.Diagnostics.UnrepresentedOccurrenceCount || diagnosticsFile.Truncated != output.Diagnostics.Truncated || diagnosticsFile.UnrepresentedOccurrenceCount != output.Diagnostics.UnrepresentedOccurrenceCount {
+		t.Fatalf("diagnostics file = %#v, run diagnostics = %#v", diagnosticsFile, output.Diagnostics)
 	}
 	manifest := decodeAssembledOutput[artifacts.RunManifest](t, output.OutputFiles, "manifest.json")
 	if len(manifest.ArtifactLanes) != 1 || manifest.ArtifactLanes[0].Normalizer != spellnormalize.Key {
@@ -155,14 +166,14 @@ func TestAssembledSpellPipelineHonorsNormalizeValidatorOverride(t *testing.T) {
 	if err != nil || output.Manifest.ValidationStatus != "approved" || len(output.Rejected) != 0 || len(output.NormalizeOutputs) != 1 {
 		t.Fatalf("Run() error = %v output = %#v, want approved override run", err, output)
 	}
-	for _, warning := range output.Warnings {
-		if warning.ReasonCode == "spell_not_near_source" {
-			t.Fatalf("warnings = %#v, want explicit validator override to replace default relatedness chain", output.Warnings)
+	for _, group := range output.Diagnostics.Groups {
+		if group.ReasonCode == "spell_not_near_source" {
+			t.Fatalf("diagnostics = %#v, want explicit validator override to replace default relatedness chain", output.Diagnostics)
 		}
 	}
 }
 
-func TestAssembledSpellPipelinePromotesTerminalUnknownSpellWarning(t *testing.T) {
+func TestAssembledSpellPipelinePromotesTerminalUnknownSpellDiagnostics(t *testing.T) {
 	registries, resolved, _ := assembledSpellPipeline(t, assembledSpellPipelineOptions{unknownSpell: true})
 	resolved.Steps[0].ArtifactLanes[0].NormalizeValidationPolicy.SemanticRejection = pipeline.SemanticRejectionRejectOutput
 	prepared, err := pipeline.Prepare(resolved, registries, pipeline.ModuleDependencies{})
@@ -190,12 +201,12 @@ func TestAssembledSpellPipelinePromotesTerminalUnknownSpellWarning(t *testing.T)
 	if !reflect.DeepEqual(rejectedFile.Rejected, output.Rejected) {
 		t.Fatalf("rejected file = %#v, run rejections = %#v, want durable rejection diagnostic", rejectedFile.Rejected, output.Rejected)
 	}
-	if len(output.Warnings) != 2 || output.Warnings[0].ReasonCode != spellnormalize.ReasonCodeSpellNameUnresolved || output.Warnings[0].Scope != "spell_casts[0]" || output.Warnings[1].ReasonCode != "spell_not_near_source" {
-		t.Fatalf("warnings = %#v, want complete terminal normalize validation warnings", output.Warnings)
+	if len(output.Diagnostics.Groups) != 2 || output.Diagnostics.Groups[0].ReasonCode != spellnormalize.ReasonCodeSpellNameUnresolved || output.Diagnostics.Groups[0].Samples[0].Scope != "spell_casts[0]" || output.Diagnostics.Groups[1].ReasonCode != "spell_not_near_source" {
+		t.Fatalf("diagnostics = %#v, want complete terminal normalize validation diagnostics", output.Diagnostics)
 	}
 }
 
-func TestAssembledSpellPipelinePromotesUnknownSpellWarningWhenOverrideAccepts(t *testing.T) {
+func TestAssembledSpellPipelinePromotesUnknownSpellAdvisoryWhenOverrideAccepts(t *testing.T) {
 	registries, resolved, _ := assembledSpellPipeline(t, assembledSpellPipelineOptions{normalizeValidatorOverride: true, unknownSpell: true})
 	prepared, err := pipeline.Prepare(resolved, registries, pipeline.ModuleDependencies{})
 	if err != nil {
@@ -219,14 +230,20 @@ func TestAssembledSpellPipelinePromotesUnknownSpellWarningWhenOverrideAccepts(t
 	if len(normalized.SpellCasts) != 1 || normalized.SpellCasts[0].Spell != "Mysterious Burst" {
 		t.Fatalf("normalized casts = %#v, want unresolved name preserved", normalized.SpellCasts)
 	}
-	if len(output.Warnings) != 1 || output.Warnings[0].ReasonCode != spellnormalize.ReasonCodeSpellNameUnresolved || output.Warnings[0].Scope != "spell_casts[0]" {
-		t.Fatalf("warnings = %#v, want promoted scoped unresolved-name warning", output.Warnings)
+	if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].ReasonCode != spellnormalize.ReasonCodeSpellNameUnresolved || output.Diagnostics.Groups[0].Samples[0].Scope != "spell_casts[0]" {
+		t.Fatalf("diagnostics = %#v, want promoted scoped unresolved-name diagnostic", output.Diagnostics)
 	}
 	warningsFile := decodeAssembledOutput[struct {
-		Warnings []contracts.Warning `json:"warnings"`
+		Groups []contracts.DiagnosticGroup `json:"groups"`
 	}](t, output.OutputFiles, "warnings.json")
-	if !reflect.DeepEqual(warningsFile.Warnings, output.Warnings) {
-		t.Fatalf("warnings file = %#v, run warnings = %#v, want durable unresolved-name warning", warningsFile.Warnings, output.Warnings)
+	if len(warningsFile.Groups) != 0 {
+		t.Fatalf("warnings file = %#v, want no process warnings for an advisory diagnostic", warningsFile.Groups)
+	}
+	diagnosticsFile := decodeAssembledOutput[struct {
+		Groups []contracts.DiagnosticGroup `json:"groups"`
+	}](t, output.OutputFiles, "diagnostics.json")
+	if !reflect.DeepEqual(diagnosticsFile.Groups, output.Diagnostics.Groups) {
+		t.Fatalf("diagnostics file = %#v, run diagnostics = %#v", diagnosticsFile.Groups, output.Diagnostics)
 	}
 }
 
@@ -429,6 +446,14 @@ func (e *assembledSpellExtractor) chunkIndexesSnapshot() []int {
 	return append([]int(nil), e.chunkIndexes...)
 }
 
+func diagnosticGroupOccurrences(groups []contracts.DiagnosticGroup) int {
+	count := 0
+	for _, group := range groups {
+		count += group.OccurrenceCount
+	}
+	return count
+}
+
 func decodeAssembledOutput[T any](t *testing.T, files []contracts.OutputFile, name string) T {
 	t.Helper()
 	for _, file := range files {

internal/cli/dnd_scene_descriptions_contract_test.go

@@ -91,8 +91,8 @@ func TestProductionSceneDescriptionWorkflow(t *testing.T) {
 	if !reflect.DeepEqual(durable, want) {
 		t.Fatalf("durable output payload = %#v, want %#v", durable, want)
 	}
-	if len(output.Warnings) != 0 {
-		t.Fatalf("warnings = %#v, want grounded descriptions without warnings", output.Warnings)
+	if len(output.Diagnostics.Groups) != 0 {
+		t.Fatalf("diagnostics = %#v, want grounded descriptions without diagnostics", output.Diagnostics)
 	}
 }
 

internal/cli/example_contract_test.go

@@ -239,10 +239,10 @@ func TestMaintainedMinimalInvocationProducesJSONBundle(t *testing.T) {
 		t.Fatalf("rejected = %#v, want empty rejection list", rejected.Rejected)
 	}
 	warnings := readProductionJSON[struct {
-		Warnings []json.RawMessage `json:"warnings"`
+		Groups []json.RawMessage `json:"groups"`
 	}](t, filepath.Join(runRoot, "warnings.json"))
-	if len(warnings.Warnings) != 0 {
-		t.Fatalf("warnings = %#v, want empty warning list", warnings.Warnings)
+	if len(warnings.Groups) != 0 {
+		t.Fatalf("warnings = %#v, want empty warning list", warnings.Groups)
 	}
 }
 

internal/cli/production_contract_test.go

@@ -852,10 +852,10 @@ func TestProductionSceneRunRecordsAnnotationFreeChunkPlanAndProvenance(t *testin
 		t.Fatalf("chunk map range annotations = %#v, want none", chunkMap.Chunks[0].Annotations)
 	}
 	warnings := readProductionJSON[struct {
-		Warnings []contracts.Warning `json:"warnings"`
+		Groups []contracts.DiagnosticGroup `json:"groups"`
 	}](t, filepath.Join(outputRoot, productionRunID, "warnings.json"))
-	if len(warnings.Warnings) != 0 {
-		t.Fatalf("warnings = %#v, want none", warnings.Warnings)
+	if len(warnings.Groups) != 0 {
+		t.Fatalf("warnings = %#v, want none", warnings.Groups)
 	}
 	if len(fake.requestsFor(scenes.PromptID)) != 1 || len(fake.requestsFor(spells.PromptID)) != 1 || len(fake.requestsFor(itemoccurrenceextract.PromptID)) != 1 {
 		t.Fatalf("fake prompt requests = %#v, want one scene, spell, and item-occurrence request", fake.requestPrompts())

internal/cli/run.go

@@ -375,7 +375,7 @@ func runPipelineCommand(args []string, stdout, stderr io.Writer, opts Options) i
 	if err != nil {
 		return failPipelineCommand(stderr, commandState, terminalWriter, fmt.Errorf("resolve working directory: %w", err))
 	}
-	materialized, referenceWarnings, err := pipeline.MaterializeReferences(effective.ResolvedPipeline, catalog, pipeline.ReferenceMaterializationOptions{
+	materialized, referenceDiagnostics, err := pipeline.MaterializeReferences(effective.ResolvedPipeline, catalog, pipeline.ReferenceMaterializationOptions{
 		ConfigPath: loadedConfigPath,
 		WorkingDir: workingDir,
 	})
@@ -462,7 +462,7 @@ func runPipelineCommand(args []string, stdout, stderr io.Writer, opts Options) i
 		StartedAt:        startedAt,
 		LLMProfiles:      llmProfiles,
 		Metadata:         runMetadata(effective.Config.Output.Directory, debugPath),
-		Warnings:         referenceWarnings,
+		Diagnostics:      referenceDiagnostics,
 		ChunkCacheMode:   effective.Config.Cache.ChunkPlans.Mode,
 		ChunkPlans:       chunkPlans,
 		Checkpoints:      checkpointRecorder,
@@ -471,7 +471,10 @@ func runPipelineCommand(args []string, stdout, stderr io.Writer, opts Options) i
 		Debug:            debugRecorder,
 		ExtractWorkers:   cfg.Concurrency.StageWorkers["extract"],
 	})
-	commandState.observeOutput(output)
+	diagnosticProjection, diagnosticErr := contracts.ProjectDiagnosticCollection(output.Diagnostics)
+	if diagnosticErr == nil {
+		commandState.observeOutput(output, diagnosticProjection)
+	}
 	if err != nil {
 		primaryErr := fmt.Errorf("run pipeline %q: %w", pipelineID, err)
 		if output.Manifest.PipelineID != "" {
@@ -479,15 +482,21 @@ func runPipelineCommand(args []string, stdout, stderr io.Writer, opts Options) i
 				return failPipelineCommand(stderr, commandState, terminalWriter, primaryErr, fmt.Errorf("write debug summary: %w", summaryErr))
 			}
 		}
+		if diagnosticErr != nil {
+			return failPipelineCommand(stderr, commandState, terminalWriter, primaryErr, fmt.Errorf("summarize run diagnostics: %w", diagnosticErr))
+		}
 		return failPipelineCommand(stderr, commandState, terminalWriter, primaryErr)
 	}
+	if diagnosticErr != nil {
+		return failPipelineCommand(stderr, commandState, terminalWriter, fmt.Errorf("summarize run diagnostics: %w", diagnosticErr))
+	}
 
 	if err := writePartialSummary(summary, output); err != nil {
 		return failPipelineCommand(stderr, commandState, terminalWriter, fmt.Errorf("write debug summary: %w", err))
 	}
 	var encodedResult []byte
 	if *machineOutput {
-		result, err := newRunResult(effective.ResolvedPipeline, output, runOutputDir, debugPath)
+		result, err := newRunResultWithDiagnostics(effective.ResolvedPipeline, output, runOutputDir, debugPath, diagnosticProjection)
 		if err != nil {
 			return failPipelineCommand(stderr, commandState, terminalWriter, err)
 		}
@@ -513,12 +522,25 @@ func runPipelineCommand(args []string, stdout, stderr io.Writer, opts Options) i
 			fmt.Fprintf(stdout, "debug=%s\n", debugPath)
 		}
 	}
-	if len(output.Warnings) > 0 {
-		fmt.Fprintf(stderr, "notarius: run completed with %d warning(s)\n", len(output.Warnings))
+	if warningGroups := len(diagnosticProjection.Warnings); warningGroups > 0 {
+		fmt.Fprintf(stderr, "notarius: run completed with %d warning group(s), %d occurrence(s)", warningGroups, diagnosticProjection.WarningOccurrenceCount)
+		if warningFile, ok := logicalOutputFile(output.OutputFiles, "warnings.json"); ok {
+			fmt.Fprintf(stderr, "; details=%s", filepath.Join(runOutputDir, warningFile))
+		}
+		fmt.Fprintln(stderr)
 	}
 	return 0
 }
 
+func logicalOutputFile(files []contracts.OutputFile, name string) (string, bool) {
+	for _, file := range files {
+		if file.Name == name {
+			return file.Name, true
+		}
+	}
+	return "", false
+}
+
 func writeSummary(summary *debugbundle.SummaryWriter, write func() error) error {
 	if summary == nil {
 		return nil
@@ -538,7 +560,7 @@ func writePartialSummary(summary *debugbundle.SummaryWriter, output pipeline.Run
 			return err
 		}
 	}
-	if err := summary.WriteWarnings(output.Warnings); err != nil {
+	if err := summary.WriteDiagnostics(output.Diagnostics); err != nil {
 		return err
 	}
 	return summary.WriteCheckpointEvents(output.CheckpointEvents)

internal/cli/run_contract_test.go

@@ -590,10 +590,11 @@ func TestRunWarningsRemainSuccessfulAndReachDurableSurfaces(t *testing.T) {
 	roots := newStateTestRoots(t)
 	harness := newStateTestHarness()
 	harness.includeWarnings = true
-	harness.chunkWarnings = []contracts.Warning{{Scope: "chunk", ReasonCode: "contract-warning", Message: "warning retained"}}
+	harness.includeWarningFile = true
+	harness.chunkDiagnostics = []contracts.ProducerDiagnostic{stateTestDiagnostic("chunk", "contract-warning", "warning retained")}
 	var stdout, stderr bytes.Buffer
 	code := RunWithOptions([]string{"run", "sample", "--config", roots.config, "--input", roots.input, "--chunk_cache", "bypass", "--debug"}, &stdout, &stderr, harness.options())
-	if code != 0 || !strings.Contains(stdout.String(), "outputs=1") || !strings.Contains(stderr.String(), "1 warning(s)") {
+	if code != 0 || !strings.Contains(stdout.String(), "outputs=1") || !strings.Contains(stderr.String(), "1 warning group(s), 1 occurrence(s)") || !strings.Contains(stderr.String(), "warnings.json") {
 		t.Fatalf("code=%d stdout=%q stderr=%q", code, stdout.String(), stderr.String())
 	}
 	outputPath := filepath.Join(onlyChildDir(t, roots.output), "result.json")
@@ -601,11 +602,12 @@ func TestRunWarningsRemainSuccessfulAndReachDurableSurfaces(t *testing.T) {
 	if err != nil || !strings.Contains(string(output), "contract-warning") {
 		t.Fatalf("durable output = %q, %v", output, err)
 	}
+	assertFile(t, filepath.Join(filepath.Dir(outputPath), "warnings.json"))
 	bundle := onlyChildDir(t, roots.debug)
-	var warnings []contracts.Warning
-	readStateTestSummaryJSON(t, bundle, "warnings.json", &warnings)
-	if len(warnings) != 1 || warnings[0].ReasonCode != "contract-warning" {
-		t.Fatalf("debug warnings = %#v", warnings)
+	var diagnostics contracts.DiagnosticCollection
+	readStateTestSummaryJSON(t, bundle, "final-diagnostics.json", &diagnostics)
+	if len(diagnostics.Groups) != 1 || diagnostics.Groups[0].ReasonCode != "contract-warning" {
+		t.Fatalf("debug diagnostics = %#v", diagnostics)
 	}
 }
 

internal/cli/run_result.go

@@ -8,26 +8,39 @@ import (
 	"strings"
 
 	"gitea.maximumdirect.net/eric/notarius/internal/core/artifacts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
 )
 
-const runResultSchemaVersion = "notarius.run-result.v1"
+const runResultSchemaVersion = "notarius.run-result.v2"
 
 type runResult struct {
-	SchemaVersion         string                        `json:"schema_version"`
-	RunID                 string                        `json:"run_id"`
-	PipelineID            string                        `json:"pipeline_id"`
-	OutputDirectory       string                        `json:"output_directory"`
-	IndexFile             string                        `json:"index_file,omitempty"`
-	NormalizedOutputCount int                           `json:"normalized_output_count"`
-	RejectedOutputCount   int                           `json:"rejected_output_count"`
-	WarningCount          int                           `json:"warning_count"`
-	ValidationStatus      string                        `json:"validation_status"`
-	ValidationSummaries   []artifacts.ValidationSummary `json:"validation_summaries,omitempty"`
-	DebugDirectory        string                        `json:"debug_directory,omitempty"`
+	SchemaVersion             string                        `json:"schema_version"`
+	RunID                     string                        `json:"run_id"`
+	PipelineID                string                        `json:"pipeline_id"`
+	OutputDirectory           string                        `json:"output_directory"`
+	IndexFile                 string                        `json:"index_file,omitempty"`
+	NormalizedOutputCount     int                           `json:"normalized_output_count"`
+	RejectedOutputCount       int                           `json:"rejected_output_count"`
+	WarningGroupCount         int                           `json:"warning_group_count"`
+	WarningOccurrenceCount    int                           `json:"warning_occurrence_count"`
+	DiagnosticGroupCount      int                           `json:"diagnostic_group_count"`
+	DiagnosticOccurrenceCount int                           `json:"diagnostic_occurrence_count"`
+	DiagnosticsTruncated      bool                          `json:"diagnostics_truncated"`
+	ValidationStatus          string                        `json:"validation_status"`
+	ValidationSummaries       []artifacts.ValidationSummary `json:"validation_summaries,omitempty"`
+	DebugDirectory            string                        `json:"debug_directory,omitempty"`
 }
 
 func newRunResult(resolved pipeline.ResolvedPipeline, output pipeline.RunOutput, outputDirectory, debugDirectory string) (runResult, error) {
+	diagnosticProjection, err := contracts.ProjectDiagnosticCollection(output.Diagnostics)
+	if err != nil {
+		return runResult{}, fmt.Errorf("summarize run diagnostics: %w", err)
+	}
+	return newRunResultWithDiagnostics(resolved, output, outputDirectory, debugDirectory, diagnosticProjection)
+}
+
+func newRunResultWithDiagnostics(resolved pipeline.ResolvedPipeline, output pipeline.RunOutput, outputDirectory, debugDirectory string, diagnosticProjection contracts.DiagnosticProjection) (runResult, error) {
 	if strings.TrimSpace(output.Manifest.RunID) == "" {
 		return runResult{}, fmt.Errorf("run result requires a run ID")
 	}
@@ -53,15 +66,19 @@ func newRunResult(resolved pipeline.ResolvedPipeline, output pipeline.RunOutput,
 	}
 
 	result := runResult{
-		SchemaVersion:         runResultSchemaVersion,
-		RunID:                 output.Manifest.RunID,
-		PipelineID:            output.Manifest.PipelineID,
-		OutputDirectory:       absOutputDirectory,
-		NormalizedOutputCount: len(output.NormalizeOutputs),
-		RejectedOutputCount:   len(output.Rejected),
-		WarningCount:          len(output.Warnings),
-		ValidationStatus:      output.Manifest.ValidationStatus,
-		ValidationSummaries:   cloneValidationSummaries(output.Manifest.ValidationSummaries),
+		SchemaVersion:             runResultSchemaVersion,
+		RunID:                     output.Manifest.RunID,
+		PipelineID:                output.Manifest.PipelineID,
+		OutputDirectory:           absOutputDirectory,
+		NormalizedOutputCount:     len(output.NormalizeOutputs),
+		RejectedOutputCount:       len(output.Rejected),
+		WarningGroupCount:         len(diagnosticProjection.Warnings),
+		WarningOccurrenceCount:    diagnosticProjection.WarningOccurrenceCount,
+		DiagnosticGroupCount:      len(diagnosticProjection.Diagnostics),
+		DiagnosticOccurrenceCount: diagnosticProjection.DiagnosticOccurrenceCount,
+		DiagnosticsTruncated:      output.Diagnostics.Truncated,
+		ValidationStatus:          output.Manifest.ValidationStatus,
+		ValidationSummaries:       cloneValidationSummaries(output.Manifest.ValidationSummaries),
 	}
 
 	if strings.TrimSpace(debugDirectory) != "" {

internal/cli/run_result_command_test.go

@@ -27,7 +27,7 @@ func TestMaintainedMinimalInvocationEmitsRunResult(t *testing.T) {
 	}
 
 	receipt := decodeRunResultDocument(t, stdout.String())
-	if got := receipt["schema_version"]; got != "notarius.run-result.v1" {
+	if got := receipt["schema_version"]; got != "notarius.run-result.v2" {
 		t.Fatalf("schema_version = %q", got)
 	}
 	if got := receipt["run_id"]; got != productionRunID {
@@ -45,8 +45,8 @@ func TestMaintainedMinimalInvocationEmitsRunResult(t *testing.T) {
 	if got := receipt["rejected_output_count"]; got != float64(0) {
 		t.Fatalf("rejected_output_count = %v", got)
 	}
-	if got := receipt["warning_count"]; got != float64(0) {
-		t.Fatalf("warning_count = %v", got)
+	if got := receipt["warning_group_count"]; got != float64(0) || receipt["warning_occurrence_count"] != float64(0) || receipt["diagnostic_group_count"] != float64(0) || receipt["diagnostic_occurrence_count"] != float64(0) || receipt["diagnostics_truncated"] != false {
+		t.Fatalf("diagnostic counts = %#v", receipt)
 	}
 	if got := receipt["validation_status"]; got != "approved" {
 		t.Fatalf("validation_status = %q", got)
@@ -63,19 +63,19 @@ func TestMaintainedMinimalInvocationEmitsRunResult(t *testing.T) {
 func TestRunResultReportsWarningsAndDebugBundle(t *testing.T) {
 	roots := newStateTestRoots(t)
 	harness := newStateTestHarness()
-	harness.chunkWarnings = []contracts.Warning{{Scope: "chunk", ReasonCode: "contract-warning", Message: "warning retained"}}
+	harness.chunkDiagnostics = []contracts.ProducerDiagnostic{stateTestDiagnostic("chunk", "contract-warning", "warning retained")}
 	var stdout, stderr bytes.Buffer
 	code := RunWithOptions([]string{
 		"run", "sample", "--config", roots.config, "--input", roots.input,
 		"--chunk_cache", "bypass", "--debug", "--json",
 	}, &stdout, &stderr, harness.options())
-	if code != 0 || !strings.Contains(stderr.String(), "1 warning(s)") {
+	if code != 0 || !strings.Contains(stderr.String(), "1 warning group(s), 1 occurrence(s)") || strings.Contains(stderr.String(), "warnings.json") {
 		t.Fatalf("code=%d stdout=%q stderr=%q", code, stdout.String(), stderr.String())
 	}
 
 	receipt := decodeRunResultDocument(t, stdout.String())
-	if got := receipt["warning_count"]; got != float64(1) {
-		t.Fatalf("warning_count = %v", got)
+	if got := receipt["warning_group_count"]; got != float64(1) || receipt["warning_occurrence_count"] != float64(1) || receipt["diagnostic_group_count"] != float64(0) || receipt["diagnostic_occurrence_count"] != float64(0) || receipt["diagnostics_truncated"] != false {
+		t.Fatalf("diagnostic counts = %#v", receipt)
 	}
 	debugDirectory, ok := receipt["debug_directory"].(string)
 	if !ok || !filepath.IsAbs(debugDirectory) || debugDirectory != onlyChildDir(t, roots.debug) {

internal/cli/run_result_test.go

@@ -52,8 +52,8 @@ func TestRunResultEncodesRequiredFieldsAndCounts(t *testing.T) {
 	if got := decoded["rejected_output_count"]; got != float64(1) {
 		t.Fatalf("rejected_output_count = %v", got)
 	}
-	if got := decoded["warning_count"]; got != float64(1) {
-		t.Fatalf("warning_count = %v", got)
+	if got := decoded["warning_group_count"]; got != float64(1) || decoded["warning_occurrence_count"] != float64(1) || decoded["diagnostic_group_count"] != float64(0) || decoded["diagnostic_occurrence_count"] != float64(0) || decoded["diagnostics_truncated"] != false {
+		t.Fatalf("diagnostic counts = %#v", decoded)
 	}
 	if got := decoded["validation_summaries"]; got != nil {
 		t.Fatalf("validation_summaries = %#v, want omitted when empty", got)
@@ -182,8 +182,15 @@ func testRunOutput() pipeline.RunOutput {
 		Manifest:         artifacts.RunManifest{RunID: "run-123", PipelineID: "sample", ValidationStatus: "rejected"},
 		NormalizeOutputs: []contracts.SerializedOutput{{}, {}},
 		Rejected:         []contracts.RejectedOutput{{}},
-		Warnings:         []contracts.Warning{{}},
-		OutputFiles:      []contracts.OutputFile{{Name: "index.json"}},
+		Diagnostics: contracts.DiagnosticCollection{Groups: []contracts.DiagnosticGroup{{
+			Disposition:     contracts.DiagnosticDispositionWarning,
+			Category:        contracts.DiagnosticCategoryFallback,
+			ReasonCode:      "fallback",
+			Origin:          contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageNormalize, StepID: "step", LaneID: "lane", ModuleKey: "module"},
+			OccurrenceCount: 1,
+			Samples:         []contracts.DiagnosticSample{{Scope: "scope", Message: "message"}},
+		}}},
+		OutputFiles: []contracts.OutputFile{{Name: "index.json"}},
 	}
 }
 

internal/cli/run_terminal.go

@@ -6,6 +6,7 @@ import (
 	"io"
 
 	"gitea.maximumdirect.net/eric/notarius/internal/core/debugbundle"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
 )
 
@@ -33,13 +34,17 @@ func (s *pipelineCommandState) setDebugPath(debugPath string) {
 	}
 }
 
-func (s *pipelineCommandState) observeOutput(output pipeline.RunOutput) {
+func (s *pipelineCommandState) observeOutput(output pipeline.RunOutput, diagnostics contracts.DiagnosticProjection) {
 	if s == nil {
 		return
 	}
 	s.report.OutputCount = len(output.NormalizeOutputs)
 	s.report.RejectedCount = len(output.Rejected)
-	s.report.WarningCount = len(output.Warnings)
+	s.report.WarningGroupCount = len(diagnostics.Warnings)
+	s.report.WarningOccurrenceCount = diagnostics.WarningOccurrenceCount
+	s.report.DiagnosticGroupCount = len(diagnostics.Diagnostics)
+	s.report.DiagnosticOccurrenceCount = diagnostics.DiagnosticOccurrenceCount
+	s.report.DiagnosticsTruncated = output.Diagnostics.Truncated
 	s.report.ValidationStatus = output.Manifest.ValidationStatus
 }
 

internal/cli/spell_catalog_identity_contract_test.go

@@ -238,7 +238,7 @@ func TestChangedSemanticSpellCatalogFingerprintCannotResumeRecordedCheckpoint(t
 			Kind: dnd.SpellListKind, Schema: normalizeSchema, MediaType: "application/json", Content: []byte(`{"spell_casts":[]}`),
 		},
 	}
-	if err := recorder.NormalizeSucceeded("spells", spellnormalize.Key, normalizeDependencies, normalizeArtifact, nil); err != nil {
+	if err := recorder.NormalizeSucceeded("spells", spellnormalize.Key, normalizeDependencies, normalizeArtifact); err != nil {
 		t.Fatal(err)
 	}
 
@@ -264,7 +264,7 @@ func TestChangedSemanticSpellCatalogFingerprintCannotResumeRecordedCheckpoint(t
 	if _, decision := changedLoader.Normalize("spells", spellnormalize.Key, normalizeDependencies); decision.Reused {
 		t.Fatalf("changed normalize fingerprint decision = %#v, want normalize checkpoint cold miss", decision)
 	}
-	changedMapping := replaceCheckpointFingerprintValue(t, fingerprints, extractSpellMappingFingerprintName(), "dnd.spells.extract_mapping.v3")
+	changedMapping := replaceCheckpointFingerprintValue(t, fingerprints, extractSpellMappingFingerprintName(), "dnd.spells.extract_mapping.changed")
 	assertOnlyCheckpointFingerprintChanged(t, fingerprints, changedMapping, extractSpellMappingFingerprintName())
 	_, mappingLoader, err := checkpointHandlersForRun(settings, Options{}, materialized, changedMapping, llmFingerprints, []byte("same input"), nil, nil, "", "", LLMRuntimeOverrides{}, true)
 	if err != nil {

internal/cli/spell_catalog_retry_contract_test.go

@@ -19,12 +19,12 @@ import (
 func TestProductionSpellCatalogValidationRetries(t *testing.T) {
 	const retries = 2
 	tests := []struct {
-		name            string
-		responses       []string
-		wantCalls       int
-		wantRejected    bool
-		wantSpell       string
-		wantWarningCode string
+		name             string
+		responses        []string
+		wantCalls        int
+		wantRejected     bool
+		wantSpell        string
+		wantAdvisoryCode string
 	}{
 		{
 			name: "unknown spell remains rejected after exhaustion",
@@ -42,9 +42,9 @@ func TestProductionSpellCatalogValidationRetries(t *testing.T) {
 				productionSpellResponse("Unknown Spell"),
 				productionSpellResponse("Aegis of Emberfall"),
 			},
-			wantCalls:       2,
-			wantSpell:       "Aegis of Emberfall",
-			wantWarningCode: "spell_not_near_source",
+			wantCalls:        2,
+			wantSpell:        "Aegis of Emberfall",
+			wantAdvisoryCode: "spell_not_near_source",
 		},
 	}
 
@@ -92,8 +92,8 @@ func TestProductionSpellCatalogValidationRetries(t *testing.T) {
 				if rejection.ReasonCode != "unknown_spell" || rejection.AttemptCount != retries+1 {
 					t.Fatalf("rejection = %#v, want exhausted unknown-spell rejection", rejection)
 				}
-				if len(output.Warnings) != 1 || output.Warnings[0].ReasonCode != "spell_not_near_source" {
-					t.Fatalf("warnings = %#v, want complete terminal validation warnings", output.Warnings)
+				if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].ReasonCode != "spell_not_near_source" {
+					t.Fatalf("diagnostics = %#v, want complete terminal validation diagnostic", output.Diagnostics)
 				}
 				return
 			}
@@ -108,8 +108,16 @@ func TestProductionSpellCatalogValidationRetries(t *testing.T) {
 			if len(value.SpellCasts) != 1 || value.SpellCasts[0].Spell != tt.wantSpell {
 				t.Fatalf("normalized spell list = %#v, want accepted overlay spell", value)
 			}
-			if len(output.Warnings) != 2 || output.Warnings[0].ReasonCode != tt.wantWarningCode || output.Warnings[1].ReasonCode != tt.wantWarningCode {
-				t.Fatalf("warnings = %#v, want accepted-attempt warnings from extract and normalize validation", output.Warnings)
+			if len(output.Diagnostics.Groups) != 2 || output.Diagnostics.Groups[0].ReasonCode != tt.wantAdvisoryCode {
+				t.Fatalf("diagnostics = %#v, want terminal extract and normalize diagnostics", output.Diagnostics)
+			}
+			if len(output.Diagnostics.Groups) != 2 {
+				t.Fatalf("diagnostics = %#v, want extract and normalize data-quality advisories", output.Diagnostics)
+			}
+			for _, diagnostic := range output.Diagnostics.Groups {
+				if diagnostic.Disposition != contracts.DiagnosticDispositionAdvisory || diagnostic.ReasonCode != tt.wantAdvisoryCode {
+					t.Fatalf("diagnostics = %#v, want only data-quality advisories", output.Diagnostics)
+				}
 			}
 		})
 	}

internal/cli/state_hardening_test.go

@@ -349,7 +349,7 @@ func TestRunUsesOneInjectedIdentityForDebugOutputAndManifest(t *testing.T) {
 		t.Fatalf("debug invocation session = %q, want %q", invocation.SessionID, wantSessionID)
 	}
 	report := readStateTestRunReport(t, debugPath)
-	if !report.Succeeded || report.RunID != runID || report.PipelineID != "sample" || report.OutputPath != outputPath || report.DebugPath != debugPath || report.OutputCount != 1 || report.RejectedCount != 0 || report.WarningCount != 0 || report.ValidationStatus != "approved" {
+	if !report.Succeeded || report.RunID != runID || report.PipelineID != "sample" || report.OutputPath != outputPath || report.DebugPath != debugPath || report.OutputCount != 1 || report.RejectedCount != 0 || report.WarningGroupCount != 0 || report.WarningOccurrenceCount != 0 || report.DiagnosticGroupCount != 0 || report.DiagnosticOccurrenceCount != 0 || report.DiagnosticsTruncated || report.ValidationStatus != "approved" {
 		t.Fatalf("success report = %#v", report)
 	}
 	if !strings.Contains(result.stdout, "outputs=1 rejected=0") {
@@ -431,7 +431,7 @@ func TestRunWritesTerminalArtifactsForResolutionPipelineAndOutputFailures(t *tes
 			bundlePath := onlyChildDir(t, roots.debug)
 			runID := filepath.Base(bundlePath)
 			report := readStateTestRunReport(t, bundlePath)
-			if report.Succeeded || report.RunID != runID || report.PipelineID != tc.pipelineID || report.OutputPath != filepath.Join(roots.output, runID) || report.DebugPath != bundlePath || report.OutputCount != tc.wantOutputs || report.RejectedCount != 0 || report.WarningCount != 0 || report.ValidationStatus != tc.wantValidation {
+			if report.Succeeded || report.RunID != runID || report.PipelineID != tc.pipelineID || report.OutputPath != filepath.Join(roots.output, runID) || report.DebugPath != bundlePath || report.OutputCount != tc.wantOutputs || report.RejectedCount != 0 || report.WarningGroupCount != 0 || report.WarningOccurrenceCount != 0 || report.DiagnosticGroupCount != 0 || report.DiagnosticOccurrenceCount != 0 || report.DiagnosticsTruncated || report.ValidationStatus != tc.wantValidation {
 				t.Fatalf("failure report = %#v", report)
 			}
 			errorLog, err := os.ReadFile(filepath.Join(bundlePath, "summary", "error.log"))
@@ -445,7 +445,7 @@ func TestRunWritesTerminalArtifactsForResolutionPipelineAndOutputFailures(t *tes
 func TestRunRetainsPartialPipelineOutcomeInFailureSummary(t *testing.T) {
 	roots := newStateTestRoots(t)
 	harness := newStateTestHarness()
-	harness.chunkWarnings = []contracts.Warning{{Scope: "chunk", ReasonCode: "partial-warning", Message: "warning retained before failure"}}
+	harness.chunkDiagnostics = []contracts.ProducerDiagnostic{stateTestDiagnostic("chunk", "partial-warning", "warning retained before failure")}
 	harness.extractErr = errors.New("synthetic partial pipeline failure")
 
 	result := runStateTest(t, roots, harness.options(), true, true, "bypass")
@@ -454,7 +454,7 @@ func TestRunRetainsPartialPipelineOutcomeInFailureSummary(t *testing.T) {
 	}
 	bundlePath := onlyChildDir(t, roots.debug)
 	report := readStateTestRunReport(t, bundlePath)
-	if report.Succeeded || report.OutputCount != 0 || report.RejectedCount != 0 || report.WarningCount != 1 || report.ValidationStatus != "failed" {
+	if report.Succeeded || report.OutputCount != 0 || report.RejectedCount != 0 || report.WarningGroupCount != 1 || report.WarningOccurrenceCount != 1 || report.DiagnosticGroupCount != 0 || report.DiagnosticOccurrenceCount != 0 || report.DiagnosticsTruncated || report.ValidationStatus != "failed" {
 		t.Fatalf("partial failure report = %#v", report)
 	}
 
@@ -463,10 +463,10 @@ func TestRunRetainsPartialPipelineOutcomeInFailureSummary(t *testing.T) {
 	if manifest.RunID != report.RunID || manifest.PipelineID != "sample" || manifest.ValidationStatus != "failed" {
 		t.Fatalf("partial manifest = %#v", manifest)
 	}
-	var warnings []contracts.Warning
-	readStateTestSummaryJSON(t, bundlePath, "warnings.json", &warnings)
-	if len(warnings) != 1 || warnings[0].ReasonCode != "partial-warning" {
-		t.Fatalf("partial warnings = %#v", warnings)
+	var diagnostics contracts.DiagnosticCollection
+	readStateTestSummaryJSON(t, bundlePath, "final-diagnostics.json", &diagnostics)
+	if len(diagnostics.Groups) != 1 || diagnostics.Groups[0].ReasonCode != "partial-warning" {
+		t.Fatalf("partial diagnostics = %#v", diagnostics)
 	}
 	var events []pipeline.CheckpointEvent
 	readStateTestSummaryJSON(t, bundlePath, "checkpoint-events.json", &events)
@@ -863,11 +863,12 @@ type stateTestHarness struct {
 	chunkCalls, extractCalls int
 	runIDCalls               uint64
 	extractErr               error
-	chunkWarnings            []contracts.Warning
+	chunkDiagnostics         []contracts.ProducerDiagnostic
 	moduleProfiles           []string
 	sessionIDs               []string
-	outputWarnings           []contracts.Warning
+	outputDiagnostics        contracts.DiagnosticCollection
 	includeWarnings          bool
+	includeWarningFile       bool
 }
 
 func newStateTestHarness() *stateTestHarness { return &stateTestHarness{} }
@@ -892,7 +893,7 @@ func (h *stateTestHarness) options() Options {
 		panic(err)
 	}
 	if err := registries.Outputs.RegisterWithSpec(pipeline.ModuleSpec{Key: "test/output", Stage: pipeline.StageOutput, ExecutionClass: contracts.ExecutionClassDeterministic, Requires: []string{"normalized"}, Provides: []string{"output"}}, func() (contracts.OutputEncoder, error) {
-		return stateTestOutput{harness: h, includeWarnings: h.includeWarnings}, nil
+		return stateTestOutput{harness: h, includeDiagnostics: h.includeWarnings}, nil
 	}); err != nil {
 		panic(err)
 	}
@@ -931,7 +932,7 @@ func (c stateTestChunker) Plan(_ context.Context, req contracts.ChunkRequest) (c
 	c.harness.mu.Lock()
 	c.harness.chunkCalls++
 	c.harness.mu.Unlock()
-	return contracts.ChunkPlanResult{Plan: source.ChunkPlan{SourceDigest: req.Source.Digest, Ranges: []source.ChunkRange{{StartUnitID: 1, EndUnitID: 1}}}, Warnings: append([]contracts.Warning(nil), c.harness.chunkWarnings...)}, nil
+	return contracts.ChunkPlanResult{Plan: source.ChunkPlan{SourceDigest: req.Source.Digest, Ranges: []source.ChunkRange{{StartUnitID: 1, EndUnitID: 1}}}, Diagnostics: contracts.CloneProducerDiagnostics(c.harness.chunkDiagnostics)}, nil
 }
 
 const stateTestArtifactKind contracts.ArtifactKind = "test/artifact"
@@ -999,20 +1000,28 @@ func (n stateTestNormalizer) Normalize(_ context.Context, req contracts.TypedNor
 }
 
 type stateTestOutput struct {
-	harness         *stateTestHarness
-	includeWarnings bool
+	harness            *stateTestHarness
+	includeDiagnostics bool
 }
 
 func (o stateTestOutput) Key() string { return "test/output" }
 func (o stateTestOutput) Encode(_ context.Context, req contracts.OutputRequest) (contracts.OutputResult, error) {
 	o.harness.mu.Lock()
-	o.harness.outputWarnings = append([]contracts.Warning(nil), req.Warnings...)
+	o.harness.outputDiagnostics = contracts.CloneDiagnosticCollection(req.Diagnostics)
 	o.harness.mu.Unlock()
 	data := []byte("{\"ok\":true}\n")
-	if o.includeWarnings && len(req.Warnings) > 0 {
-		data = []byte(fmt.Sprintf("{\"ok\":true,\"warnings\":%q}\n", req.Warnings[0].ReasonCode))
+	if o.includeDiagnostics && len(req.Diagnostics.Groups) > 0 {
+		data = []byte(fmt.Sprintf("{\"ok\":true,\"diagnostics\":%q}\n", req.Diagnostics.Groups[0].ReasonCode))
 	}
-	return contracts.OutputResult{Files: []contracts.OutputFile{{Name: "result.json", Bytes: data}}}, nil
+	files := []contracts.OutputFile{{Name: "result.json", Bytes: data}}
+	if o.harness.includeWarningFile {
+		files = append(files, contracts.OutputFile{Name: "warnings.json", Bytes: []byte("{\"warnings\":true}\n")})
+	}
+	return contracts.OutputResult{Files: files}, nil
+}
+
+func stateTestDiagnostic(scope, reasonCode, message string) contracts.ProducerDiagnostic {
+	return contracts.ProducerDiagnostic{Disposition: contracts.DiagnosticDispositionWarning, Category: contracts.DiagnosticCategoryDegradation, ReasonCode: reasonCode, OccurrenceCount: 1, Samples: []contracts.DiagnosticSample{{Scope: scope, Message: message}}}
 }
 
 type failingDebugRecorder struct{}

internal/core/debugbundle/bundle_test.go

@@ -110,7 +110,7 @@ func TestSummaryWriterWritesEverySummaryArtifact(t *testing.T) {
 	if err := summary.WriteRunReport(RunReport{RunID: bundle.RunID(), PipelineID: "test"}); err != nil {
 		t.Fatal(err)
 	}
-	if err := summary.WriteWarnings([]contracts.Warning{{ReasonCode: "test"}}); err != nil {
+	if err := summary.WriteDiagnostics(contracts.DiagnosticCollection{}); err != nil {
 		t.Fatal(err)
 	}
 	if err := summary.WriteError("failed"); err != nil {
@@ -126,7 +126,7 @@ func TestSummaryWriterWritesEverySummaryArtifact(t *testing.T) {
 		ArtifactRunManifest,
 		ArtifactChunkPlan,
 		ArtifactRunReport,
-		ArtifactWarnings,
+		ArtifactDiagnostics,
 		ArtifactErrorLog,
 	} {
 		info, err := os.Stat(filepath.Join(bundle.SummaryRoot(), name))

internal/core/debugbundle/summary.go

@@ -20,7 +20,7 @@ const (
 	ArtifactRunManifest        = "run-manifest.json"
 	ArtifactChunkPlan          = "chunk-plan.json"
 	ArtifactRunReport          = "run-report.json"
-	ArtifactWarnings           = "warnings.json"
+	ArtifactDiagnostics        = "final-diagnostics.json"
 	ArtifactErrorLog           = "error.log"
 )
 
@@ -45,15 +45,19 @@ type Invocation struct {
 	StartedAt               time.Time `json:"started_at"`
 }
 type RunReport struct {
-	RunID            string `json:"run_id"`
-	PipelineID       string `json:"pipeline_id"`
-	OutputPath       string `json:"output_path,omitempty"`
-	DebugPath        string `json:"debug_path,omitempty"`
-	Succeeded        bool   `json:"succeeded"`
-	OutputCount      int    `json:"output_count"`
-	RejectedCount    int    `json:"rejected_count"`
-	WarningCount     int    `json:"warning_count"`
-	ValidationStatus string `json:"validation_status,omitempty"`
+	RunID                     string `json:"run_id"`
+	PipelineID                string `json:"pipeline_id"`
+	OutputPath                string `json:"output_path,omitempty"`
+	DebugPath                 string `json:"debug_path,omitempty"`
+	Succeeded                 bool   `json:"succeeded"`
+	OutputCount               int    `json:"output_count"`
+	RejectedCount             int    `json:"rejected_count"`
+	WarningGroupCount         int    `json:"warning_group_count"`
+	WarningOccurrenceCount    int    `json:"warning_occurrence_count"`
+	DiagnosticGroupCount      int    `json:"diagnostic_group_count"`
+	DiagnosticOccurrenceCount int    `json:"diagnostic_occurrence_count"`
+	DiagnosticsTruncated      bool   `json:"diagnostics_truncated"`
+	ValidationStatus          string `json:"validation_status,omitempty"`
 }
 type SummaryWriter struct {
 	root, runID string
@@ -101,8 +105,8 @@ func (w *SummaryWriter) WriteChunkPlan(v artifacts.ChunkPlanSummary) error {
 	return w.writeJSON(ArtifactChunkPlan, v)
 }
 func (w *SummaryWriter) WriteRunReport(v RunReport) error { return w.writeJSON(ArtifactRunReport, v) }
-func (w *SummaryWriter) WriteWarnings(v []contracts.Warning) error {
-	return w.writeJSON(ArtifactWarnings, v)
+func (w *SummaryWriter) WriteDiagnostics(v contracts.DiagnosticCollection) error {
+	return w.writeJSON(ArtifactDiagnostics, contracts.CloneDiagnosticCollection(v))
 }
 func (w *SummaryWriter) WriteError(message string) error {
 	return w.writeBytes(ArtifactErrorLog, []byte(message+"\n"))

internal/framework/checkpoint/filesystem_test.go

@@ -24,7 +24,6 @@ type filesystemCheckpointFixture struct {
 	merge        pipeline.CheckpointArtifact
 	normalize    pipeline.CheckpointArtifact
 	dependencies []pipeline.CheckpointFingerprint
-	warnings     []contracts.Warning
 	rejected     []contracts.RejectedOutput
 }
 
@@ -36,9 +35,9 @@ func TestFilesystemCheckpointRoundTripsAllStages(t *testing.T) {
 	fixture.doc.Metadata["owner"] = "caller mutation"
 	fixture.extract.Artifact.Content[0] = 'x'
 	fixture.extract.Artifact.Metadata["content"] = "caller mutation"
+	fixture.extract.Diagnostics[0].Diagnostic.Samples[0].Message = "caller mutation"
 	fixture.merge.Artifact.Content[0] = 'x'
 	fixture.normalize.Artifact.Content[0] = 'x'
-	fixture.warnings[0].Message = "caller mutation"
 	fixture.rejected[0].Message = "caller mutation"
 
 	t.Run("source", func(t *testing.T) {
@@ -60,11 +59,11 @@ func TestFilesystemCheckpointRoundTripsAllStages(t *testing.T) {
 
 	t.Run("extract", func(t *testing.T) {
 		got, decision := fixture.loader.Extract("lane-a", "extract-module", fixture.dependencies)
-		if !decision.Reused || len(got.Outputs) != 1 || len(got.Rejected) != 1 || len(got.Warnings) != 1 {
+		if !decision.Reused || len(got.Outputs) != 1 || len(got.Rejected) != 1 || len(got.Outputs[0].Diagnostics) != 1 {
 			t.Fatalf("extract result=%#v decision=%#v", got, decision)
 		}
 		output := got.Outputs[0]
-		if !bytes.Equal(output.Artifact.Content, []byte(`{"spell":"fire"}`)) || output.Artifact.Kind != "spell" || output.Artifact.Schema.ID != "spell-schema" || output.Artifact.Schema.Version != "1" || output.Artifact.MediaType != "application/json" || output.Artifact.Metadata["chunk"] != "chunk-a" || output.ChunkRef.StartUnitID != 1 || got.Warnings[0].ReasonCode != "partial" || got.Rejected[0].ReasonCode != "invalid_source" {
+		if !bytes.Equal(output.Artifact.Content, []byte(`{"spell":"fire"}`)) || output.Artifact.Kind != "spell" || output.Artifact.Schema.ID != "spell-schema" || output.Artifact.Schema.Version != "1" || output.Artifact.MediaType != "application/json" || output.Artifact.Metadata["chunk"] != "chunk-a" || output.ChunkRef.StartUnitID != 1 || output.Diagnostics[0].Diagnostic.ReasonCode != "normalized_record" || got.Rejected[0].ReasonCode != "invalid_source" {
 			t.Fatalf("extract values were not restored: %#v", got)
 		}
 		manifest := readManifest[ExtractLaneManifest](t, filepath.Join(fixture.root, mustRelativePath(t, fixture.identity), "extract", "lane-a", "manifest.json"))
@@ -73,37 +72,36 @@ func TestFilesystemCheckpointRoundTripsAllStages(t *testing.T) {
 		}
 
 		got.Outputs[0].Artifact.Content[0] = 'y'
-		got.Warnings[0].Message = "loaded mutation"
 		reloaded, decision := fixture.loader.Extract("lane-a", "extract-module", fixture.dependencies)
-		if !decision.Reused || !bytes.Equal(reloaded.Outputs[0].Artifact.Content, []byte(`{"spell":"fire"}`)) || reloaded.Warnings[0].Message != "partial output" {
+		if !decision.Reused || !bytes.Equal(reloaded.Outputs[0].Artifact.Content, []byte(`{"spell":"fire"}`)) {
 			t.Fatalf("extract reload changed after loaded mutation: %#v decision=%#v", reloaded, decision)
 		}
 	})
 
 	for _, tt := range []struct {
 		name string
-		load func() (pipeline.CheckpointArtifact, []contracts.Warning, pipeline.CheckpointDecision)
+		load func() (pipeline.CheckpointArtifact, pipeline.CheckpointDecision)
 		want []byte
 	}{
-		{name: "merge", load: func() (pipeline.CheckpointArtifact, []contracts.Warning, pipeline.CheckpointDecision) {
+		{name: "merge", load: func() (pipeline.CheckpointArtifact, pipeline.CheckpointDecision) {
 			got, decision := fixture.loader.Merge("lane-a", "merge-module", fixture.dependencies)
-			return got.Output, got.Warnings, decision
+			return got.Output, decision
 		}, want: []byte(`{"spells":["fire"]}`)},
-		{name: "normalize", load: func() (pipeline.CheckpointArtifact, []contracts.Warning, pipeline.CheckpointDecision) {
+		{name: "normalize", load: func() (pipeline.CheckpointArtifact, pipeline.CheckpointDecision) {
 			got, decision := fixture.loader.Normalize("lane-a", "normalize-module", fixture.dependencies)
-			return got.Output, got.Warnings, decision
+			return got.Output, decision
 		}, want: []byte(`{"spells":["fire"],"normalized":true}`)},
 	} {
 		t.Run(tt.name, func(t *testing.T) {
-			got, warnings, decision := tt.load()
-			if !decision.Reused || !bytes.Equal(got.Artifact.Content, tt.want) || got.Artifact.Kind != "spell" || got.Artifact.Schema.ID != "spell-schema" || got.Artifact.Schema.Version != "1" || got.Artifact.Metadata["lane"] != "lane-a" || len(warnings) != 1 || warnings[0].ReasonCode != "review" {
-				t.Fatalf("%s result=%#v warnings=%#v decision=%#v", tt.name, got, warnings, decision)
+			got, decision := tt.load()
+			if !decision.Reused || !bytes.Equal(got.Artifact.Content, tt.want) || got.Artifact.Kind != "spell" || got.Artifact.Schema.ID != "spell-schema" || got.Artifact.Schema.Version != "1" || got.Artifact.Metadata["lane"] != "lane-a" || len(got.Diagnostics) != 1 || got.Diagnostics[0].Diagnostic.ReasonCode != "normalized_record" {
+				t.Fatalf("%s result=%#v decision=%#v", tt.name, got, decision)
 			}
 
 			got.Artifact.Content[0] = 'z'
-			reloaded, warnings, decision := tt.load()
-			if !decision.Reused || !bytes.Equal(reloaded.Artifact.Content, tt.want) || warnings[0].Message != "review manually" {
-				t.Fatalf("%s reload changed after loaded mutation: %#v warnings=%#v decision=%#v", tt.name, reloaded, warnings, decision)
+			reloaded, decision := tt.load()
+			if !decision.Reused || !bytes.Equal(reloaded.Artifact.Content, tt.want) {
+				t.Fatalf("%s reload changed after loaded mutation: %#v decision=%#v", tt.name, reloaded, decision)
 			}
 		})
 	}
@@ -159,6 +157,7 @@ func TestFilesystemCheckpointRejectsIncompatibleManifests(t *testing.T) {
 	}{
 		{"v1 schema", func(m map[string]any) { m["workspace_schema_version"] = WorkspaceSchemaVersionV1 }, pipeline.CheckpointReasonWorkspaceSchemaIncompatible},
 		{"v2 schema", func(m map[string]any) { m["workspace_schema_version"] = WorkspaceSchemaVersionV2 }, pipeline.CheckpointReasonWorkspaceSchemaIncompatible},
+		{"v3 schema", func(m map[string]any) { m["workspace_schema_version"] = WorkspaceSchemaVersionV3 }, pipeline.CheckpointReasonWorkspaceSchemaIncompatible},
 		{"unknown schema", func(m map[string]any) { m["workspace_schema_version"] = "notarius.workspace.future" }, pipeline.CheckpointReasonWorkspaceSchemaIncompatible},
 		{"identity", func(m map[string]any) { m["metadata"].(map[string]any)["checkpoint_identity_digest"] = "sha256:other" }, pipeline.CheckpointReasonIdentityMismatch},
 		{"stage", func(m map[string]any) { m["stage"] = string(StageMerge) }, pipeline.CheckpointReasonStageMismatch},
@@ -204,6 +203,11 @@ func TestFilesystemCheckpointRejectsIncompleteArtifactsAndContent(t *testing.T)
 		{"content digest", func(m map[string]any) {
 			m["outputs"].([]any)[0].(map[string]any)["content"].(map[string]any)["content_digest"] = "sha256:other"
 		}, pipeline.CheckpointReasonArtifactDigestMismatch},
+		{"diagnostics", func(m map[string]any) {
+			m["outputs"].([]any)[0].(map[string]any)["diagnostics"] = []any{map[string]any{
+				"diagnostic": map[string]any{"disposition": "warning", "category": "configuration", "reason_code": "invalid", "occurrence_count": float64(0)},
+			}}
+		}, pipeline.CheckpointReasonArtifactPayloadInvalid},
 	} {
 		t.Run(tt.name, func(t *testing.T) {
 			fixture := seedFilesystemCheckpoints(t)
@@ -337,9 +341,6 @@ func TestFilesystemLoaderReadsAcceptedNormalizeWithoutStageDependencies(t *testi
 	if checkpoint.Output.Artifact.Content == nil || string(checkpoint.Output.Artifact.Content) != string(fixture.normalize.Artifact.Content) {
 		t.Fatalf("accepted normalize output = %#v, want recorded artifact", checkpoint.Output)
 	}
-	if len(checkpoint.Warnings) != 1 || checkpoint.Warnings[0].ReasonCode != "normalized" {
-		t.Fatalf("accepted normalize warnings = %#v", checkpoint.Warnings)
-	}
 }
 
 func TestFilesystemLoaderRejectsInvalidAcceptedNormalize(t *testing.T) {
@@ -402,8 +403,7 @@ func seedAcceptedNormalizeCheckpoint(t *testing.T) filesystemCheckpointFixture {
 		t.Fatal(err)
 	}
 	stepRecorder := recorder.(pipeline.StepCheckpointRecorder)
-	warnings := []contracts.Warning{{Scope: "normalize", ReasonCode: "normalized", Message: "normalized warning"}}
-	if err := stepRecorder.NormalizeSucceededForStep("step-1", "lane-a", "normalize-module", []pipeline.CheckpointFingerprint{{Name: "merge", Value: "sha256:unavailable"}}, artifact, warnings); err != nil {
+	if err := stepRecorder.NormalizeSucceededForStep("step-1", "lane-a", "normalize-module", []pipeline.CheckpointFingerprint{{Name: "merge", Value: "sha256:unavailable"}}, artifact); err != nil {
 		t.Fatal(err)
 	}
 	loader, err := NewFilesystemLoader(root, identity)
@@ -493,20 +493,18 @@ func seedFilesystemCheckpoints(t *testing.T) filesystemCheckpointFixture {
 		merge:        checkpointArtifact("merge", `{"spells":["fire"]}`),
 		normalize:    checkpointArtifact("normalize", `{"spells":["fire"],"normalized":true}`),
 		dependencies: []pipeline.CheckpointFingerprint{{Name: "source", Value: "sha256:source"}, {Name: "chunk-plan", Value: "sha256:plan"}},
-		warnings:     []contracts.Warning{{Scope: "extract", ReasonCode: "partial", Message: "partial output"}},
 		rejected:     []contracts.RejectedOutput{{Stage: "extract", LaneID: "lane-a", ModuleKey: "extract-module", ChunkID: "chunk-a", ValidatorName: "source_refs", ReasonCode: "invalid_source", Message: "source reference is invalid", AttemptCount: 1}},
 	}
 	if err := recorder.SourceSucceeded("source-module", &fixture.doc); err != nil {
 		t.Fatal(err)
 	}
-	if err := recorder.ExtractSucceeded("lane-a", "extract-module", fixture.dependencies, []pipeline.CheckpointArtifact{fixture.extract}, fixture.rejected, fixture.warnings); err != nil {
+	if err := recorder.ExtractSucceeded("lane-a", "extract-module", fixture.dependencies, []pipeline.CheckpointArtifact{fixture.extract}, fixture.rejected); err != nil {
 		t.Fatal(err)
 	}
-	mergeWarnings := []contracts.Warning{{Scope: "merge", ReasonCode: "review", Message: "review manually"}}
-	if err := recorder.MergeSucceeded("lane-a", "merge-module", fixture.dependencies, fixture.merge, mergeWarnings); err != nil {
+	if err := recorder.MergeSucceeded("lane-a", "merge-module", fixture.dependencies, fixture.merge); err != nil {
 		t.Fatal(err)
 	}
-	if err := recorder.NormalizeSucceeded("lane-a", "normalize-module", fixture.dependencies, fixture.normalize, mergeWarnings); err != nil {
+	if err := recorder.NormalizeSucceeded("lane-a", "normalize-module", fixture.dependencies, fixture.normalize); err != nil {
 		t.Fatal(err)
 	}
 	fixture.loader, err = NewFilesystemLoader(root, identity)
@@ -527,7 +525,8 @@ func checkpointArtifact(module, content string) pipeline.CheckpointArtifact {
 	return pipeline.CheckpointArtifact{
 		LaneID: "lane-a", ModuleKey: module, SourceID: "document-1", ChunkID: "chunk-a", ChunkIndex: 0,
 		ChunkRef: source.SourceRef{SourceID: "document-1", StartUnitID: 1, EndUnitID: 1}, SchemaDigest: "sha256:schema",
-		Artifact: contracts.SerializedArtifact{Kind: "spell", Schema: contracts.ArtifactSchema{ID: "spell-schema", Name: "Spell", Version: "1", JSONSchema: []byte(`{"type":"object"}`)}, MediaType: "application/json", Content: []byte(content), Metadata: map[string]any{"chunk": "chunk-a", "lane": "lane-a"}},
+		Artifact:    contracts.SerializedArtifact{Kind: "spell", Schema: contracts.ArtifactSchema{ID: "spell-schema", Name: "Spell", Version: "1", JSONSchema: []byte(`{"type":"object"}`)}, MediaType: "application/json", Content: []byte(content), Metadata: map[string]any{"chunk": "chunk-a", "lane": "lane-a"}},
+		Diagnostics: []pipeline.CheckpointDiagnostic{{Diagnostic: contracts.ProducerDiagnostic{Disposition: contracts.DiagnosticDispositionObservation, Category: contracts.DiagnosticCategoryNormalization, ReasonCode: "normalized_record", OccurrenceCount: 1, Samples: []contracts.DiagnosticSample{{Scope: "fixture", Message: "record normalized"}}}}},
 	}
 }
 

internal/framework/checkpoint/loader.go

@@ -86,7 +86,7 @@ func (l *FilesystemLoader) ExtractForStep(stepID, laneID, moduleKey string, depe
 	if !fingerprintsEqual(checkpointToPipelineFingerprints(manifest.OutputDigests), artifactOutputDigests(outputs)) {
 		return pipeline.ExtractCheckpoint{}, decision(pipeline.CheckpointDecisionExecuted, pipeline.CheckpointReasonArtifactDigestMismatch)
 	}
-	return pipeline.ExtractCheckpoint{Outputs: outputs, Rejected: cloneRejectedOutputs(payload.Rejected), Warnings: cloneWarnings(payload.Warnings)}, reusedDecision()
+	return pipeline.ExtractCheckpoint{Outputs: outputs, Rejected: cloneRejectedOutputs(payload.Rejected)}, reusedDecision()
 }
 
 func (l *FilesystemLoader) Merge(laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint) (pipeline.MergeCheckpoint, pipeline.CheckpointDecision) {
@@ -112,7 +112,7 @@ func (l *FilesystemLoader) MergeForStep(stepID, laneID, moduleKey string, depend
 	if !fingerprintsEqual(checkpointToPipelineFingerprints(manifest.OutputDigests), artifactOutputDigests(values)) {
 		return pipeline.MergeCheckpoint{}, decision(pipeline.CheckpointDecisionExecuted, pipeline.CheckpointReasonArtifactDigestMismatch)
 	}
-	return pipeline.MergeCheckpoint{Output: values[0], Warnings: cloneWarnings(payload.Warnings)}, reusedDecision()
+	return pipeline.MergeCheckpoint{Output: values[0]}, reusedDecision()
 }
 
 func (l *FilesystemLoader) Normalize(laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint) (pipeline.NormalizeCheckpoint, pipeline.CheckpointDecision) {
@@ -138,7 +138,7 @@ func (l *FilesystemLoader) NormalizeForStep(stepID, laneID, moduleKey string, de
 	if !fingerprintsEqual(checkpointToPipelineFingerprints(manifest.OutputDigests), artifactOutputDigests(values)) {
 		return pipeline.NormalizeCheckpoint{}, decision(pipeline.CheckpointDecisionExecuted, pipeline.CheckpointReasonArtifactDigestMismatch)
 	}
-	return pipeline.NormalizeCheckpoint{Output: values[0], Warnings: cloneWarnings(payload.Warnings)}, reusedDecision()
+	return pipeline.NormalizeCheckpoint{Output: values[0]}, reusedDecision()
 }
 
 func (l *FilesystemLoader) AcceptedNormalize(stepID, laneID, moduleKey string) (pipeline.NormalizeCheckpoint, pipeline.CheckpointDecision) {
@@ -163,7 +163,7 @@ func (l *FilesystemLoader) AcceptedNormalize(stepID, laneID, moduleKey string) (
 	if !fingerprintsEqual(checkpointToPipelineFingerprints(manifest.OutputDigests), artifactOutputDigests(values)) {
 		return pipeline.NormalizeCheckpoint{}, decision(pipeline.CheckpointDecisionExecuted, pipeline.CheckpointReasonArtifactDigestMismatch)
 	}
-	return pipeline.NormalizeCheckpoint{Output: values[0], Warnings: cloneWarnings(payload.Warnings)}, decision(pipeline.CheckpointDecisionReused, pipeline.CheckpointReasonAcceptedArtifactReused)
+	return pipeline.NormalizeCheckpoint{Output: values[0]}, decision(pipeline.CheckpointDecisionReused, pipeline.CheckpointReasonAcceptedArtifactReused)
 }
 
 func (l *FilesystemLoader) validateAcceptedNormalizeManifest(manifest StageManifest, stepID, laneID, moduleKey string) pipeline.CheckpointDecision {
@@ -205,11 +205,33 @@ func artifactCheckpointOutputs(values []artifactCheckpointEnvelope) ([]pipeline.
 		if strings.TrimSpace(string(v.Kind)) == "" || strings.TrimSpace(v.Schema.ID) == "" || strings.TrimSpace(v.Schema.Version) == "" || strings.TrimSpace(v.SchemaDigest) == "" {
 			return nil, &artifactPayloadError{code: pipeline.CheckpointReasonArtifactCodecIncompatible, err: fmt.Errorf("artifact codec identity is incomplete")}
 		}
-		out = append(out, pipeline.CheckpointArtifact{LaneID: v.LaneID, ModuleKey: v.ModuleKey, SourceID: v.SourceID, ChunkID: v.ChunkID, ChunkIndex: v.ChunkIndex, ChunkRef: v.ChunkRef, SchemaDigest: v.SchemaDigest, Artifact: contracts.SerializedArtifact{Kind: v.Kind, Schema: v.Schema, MediaType: v.Content.MediaType, Content: content, Metadata: cloneMetadata(v.Content.Metadata)}})
+		diagnostics, err := cloneAndValidateCheckpointDiagnostics(v.Diagnostics)
+		if err != nil {
+			return nil, err
+		}
+		out = append(out, pipeline.CheckpointArtifact{LaneID: v.LaneID, ModuleKey: v.ModuleKey, SourceID: v.SourceID, ChunkID: v.ChunkID, ChunkIndex: v.ChunkIndex, ChunkRef: v.ChunkRef, SchemaDigest: v.SchemaDigest, Artifact: contracts.SerializedArtifact{Kind: v.Kind, Schema: v.Schema, MediaType: v.Content.MediaType, Content: content, Metadata: cloneMetadata(v.Content.Metadata)}, Diagnostics: diagnostics})
 	}
 	return out, nil
 }
 
+func cloneAndValidateCheckpointDiagnostics(values []pipeline.CheckpointDiagnostic) ([]pipeline.CheckpointDiagnostic, error) {
+	if len(values) == 0 {
+		return nil, nil
+	}
+	diagnostics := make([]contracts.ProducerDiagnostic, len(values))
+	for index, value := range values {
+		diagnostics[index] = value.Diagnostic
+	}
+	if err := contracts.ValidateProducerDiagnostics(diagnostics); err != nil {
+		return nil, &artifactPayloadError{code: pipeline.CheckpointReasonArtifactPayloadInvalid, err: fmt.Errorf("checkpoint diagnostics: %w", err)}
+	}
+	cloned := make([]pipeline.CheckpointDiagnostic, len(values))
+	for index, value := range values {
+		cloned[index] = pipeline.CheckpointDiagnostic{Diagnostic: contracts.CloneProducerDiagnostics([]contracts.ProducerDiagnostic{value.Diagnostic})[0], ValidatorKey: value.ValidatorKey}
+	}
+	return cloned, nil
+}
+
 func (l *FilesystemLoader) readJSON(name string, out any) pipeline.CheckpointDecision {
 	if !l.Enabled() {
 		return decision(pipeline.CheckpointDecisionExecuted, pipeline.CheckpointReasonLoadingDisabled)

internal/framework/checkpoint/manifest.go

@@ -3,7 +3,8 @@ package checkpoint
 import "time"
 
 const (
-	WorkspaceSchemaVersion   = "notarius.workspace.v3"
+	WorkspaceSchemaVersion   = "notarius.workspace.v4"
+	WorkspaceSchemaVersionV3 = "notarius.workspace.v3"
 	WorkspaceSchemaVersionV2 = "notarius.workspace.v2"
 	WorkspaceSchemaVersionV1 = "notarius.workspace.v1"
 )

internal/framework/checkpoint/recorder.go

@@ -81,18 +81,18 @@ func (r *FilesystemRecorder) ExtractRunningForStep(stepID, laneID string, module
 	return r.writeManifest(laneManifestPath("extract", stepID, laneID), ExtractLaneManifest{StageManifest: manifest})
 }
 
-func (r *FilesystemRecorder) ExtractSucceeded(laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, outputs []pipeline.CheckpointArtifact, rejected []contracts.RejectedOutput, warnings []contracts.Warning) error {
-	return r.ExtractSucceededForStep("", laneID, moduleKey, dependencies, outputs, rejected, warnings)
+func (r *FilesystemRecorder) ExtractSucceeded(laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, outputs []pipeline.CheckpointArtifact, rejected []contracts.RejectedOutput) error {
+	return r.ExtractSucceededForStep("", laneID, moduleKey, dependencies, outputs, rejected)
 }
 
-func (r *FilesystemRecorder) ExtractSucceededForStep(stepID, laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, outputs []pipeline.CheckpointArtifact, rejected []contracts.RejectedOutput, warnings []contracts.Warning) error {
-	payload := artifactExtractEnvelope{Outputs: artifactCheckpointEnvelopes(outputs), Rejected: cloneRejectedOutputs(rejected), Warnings: cloneWarnings(warnings)}
+func (r *FilesystemRecorder) ExtractSucceededForStep(stepID, laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, outputs []pipeline.CheckpointArtifact, rejected []contracts.RejectedOutput) error {
+	payload := artifactExtractEnvelope{Outputs: artifactCheckpointEnvelopes(outputs), Rejected: cloneRejectedOutputs(rejected)}
 	if err := r.writePayload(lanePayloadPath("extract", stepID, laneID, "outputs.json"), payload); err != nil {
 		return err
 	}
 	manifest := r.laneManifest(StageExtract, statusForRejected(rejected), stepID, laneID, moduleKey, dependencies)
 	manifest.OutputDigests = checkpointFingerprints(artifactOutputDigests(outputs))
-	manifest.ValidationStatus = validationStatusString(warnings, rejected)
+	manifest.ValidationStatus = validationStatusString(rejected)
 	manifest.Rejections = rejectionSummaries(rejected)
 	manifest.CompletedAt = timePtr(r.timestamp())
 	return r.writeManifest(laneManifestPath("extract", stepID, laneID), ExtractLaneManifest{StageManifest: manifest, ChunkCount: len(outputs) + len(rejected), OutputCount: len(outputs)})
@@ -119,17 +119,17 @@ func (r *FilesystemRecorder) MergeRunningForStep(stepID, laneID string, moduleKe
 	return r.writeManifest(laneManifestPath("merge", stepID, laneID), MergeLaneManifest{StageManifest: manifest})
 }
 
-func (r *FilesystemRecorder) MergeSucceeded(laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, output pipeline.CheckpointArtifact, warnings []contracts.Warning) error {
-	return r.MergeSucceededForStep("", laneID, moduleKey, dependencies, output, warnings)
+func (r *FilesystemRecorder) MergeSucceeded(laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, output pipeline.CheckpointArtifact) error {
+	return r.MergeSucceededForStep("", laneID, moduleKey, dependencies, output)
 }
 
-func (r *FilesystemRecorder) MergeSucceededForStep(stepID, laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, output pipeline.CheckpointArtifact, warnings []contracts.Warning) error {
-	if err := r.writePayload(lanePayloadPath("merge", stepID, laneID, "output.json"), artifactSingleEnvelope{Output: artifactCheckpointEnvelopeFromOutput(output), Warnings: cloneWarnings(warnings)}); err != nil {
+func (r *FilesystemRecorder) MergeSucceededForStep(stepID, laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, output pipeline.CheckpointArtifact) error {
+	if err := r.writePayload(lanePayloadPath("merge", stepID, laneID, "output.json"), artifactSingleEnvelope{Output: artifactCheckpointEnvelopeFromOutput(output)}); err != nil {
 		return err
 	}
 	manifest := r.laneManifest(StageMerge, StatusSucceeded, stepID, laneID, moduleKey, dependencies)
 	manifest.OutputDigests = checkpointFingerprints(artifactOutputDigests([]pipeline.CheckpointArtifact{output}))
-	manifest.ValidationStatus = validationStatusString(warnings, nil)
+	manifest.ValidationStatus = validationStatusString(nil)
 	manifest.CompletedAt = timePtr(r.timestamp())
 	return r.writeManifest(laneManifestPath("merge", stepID, laneID), MergeLaneManifest{StageManifest: manifest, InputCount: len(dependencies)})
 }
@@ -167,17 +167,17 @@ func (r *FilesystemRecorder) NormalizeRunningForStep(stepID, laneID string, modu
 	return r.writeManifest(laneManifestPath("normalize", stepID, laneID), NormalizeLaneManifest{StageManifest: manifest})
 }
 
-func (r *FilesystemRecorder) NormalizeSucceeded(laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, output pipeline.CheckpointArtifact, warnings []contracts.Warning) error {
-	return r.NormalizeSucceededForStep("", laneID, moduleKey, dependencies, output, warnings)
+func (r *FilesystemRecorder) NormalizeSucceeded(laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, output pipeline.CheckpointArtifact) error {
+	return r.NormalizeSucceededForStep("", laneID, moduleKey, dependencies, output)
 }
 
-func (r *FilesystemRecorder) NormalizeSucceededForStep(stepID, laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, output pipeline.CheckpointArtifact, warnings []contracts.Warning) error {
-	if err := r.writePayload(lanePayloadPath("normalize", stepID, laneID, "output.json"), artifactSingleEnvelope{Output: artifactCheckpointEnvelopeFromOutput(output), Warnings: cloneWarnings(warnings)}); err != nil {
+func (r *FilesystemRecorder) NormalizeSucceededForStep(stepID, laneID, moduleKey string, dependencies []pipeline.CheckpointFingerprint, output pipeline.CheckpointArtifact) error {
+	if err := r.writePayload(lanePayloadPath("normalize", stepID, laneID, "output.json"), artifactSingleEnvelope{Output: artifactCheckpointEnvelopeFromOutput(output)}); err != nil {
 		return err
 	}
 	manifest := r.laneManifest(StageNormalize, StatusSucceeded, stepID, laneID, moduleKey, dependencies)
 	manifest.OutputDigests = checkpointFingerprints(artifactOutputDigests([]pipeline.CheckpointArtifact{output}))
-	manifest.ValidationStatus = validationStatusString(warnings, nil)
+	manifest.ValidationStatus = validationStatusString(nil)
 	manifest.CompletedAt = timePtr(r.timestamp())
 	return r.writeManifest(laneManifestPath("normalize", stepID, laneID), NormalizeLaneManifest{StageManifest: manifest, InputCount: len(dependencies)})
 }
@@ -249,39 +249,37 @@ type sourceDocumentEnvelope struct {
 }
 
 type binaryEnvelope struct {
-	ContentBase64 string              `json:"content_base64,omitempty"`
-	ContentDigest string              `json:"content_digest,omitempty"`
-	MediaType     string              `json:"media_type,omitempty"`
-	Metadata      map[string]any      `json:"metadata,omitempty"`
-	Warnings      []contracts.Warning `json:"warnings,omitempty"`
+	ContentBase64 string         `json:"content_base64,omitempty"`
+	ContentDigest string         `json:"content_digest,omitempty"`
+	MediaType     string         `json:"media_type,omitempty"`
+	Metadata      map[string]any `json:"metadata,omitempty"`
 }
 
 type artifactCheckpointEnvelope struct {
-	LaneID       string                   `json:"lane_id"`
-	ModuleKey    string                   `json:"module_key"`
-	SourceID     string                   `json:"source_id,omitempty"`
-	ChunkID      string                   `json:"chunk_id,omitempty"`
-	ChunkIndex   int                      `json:"chunk_index,omitempty"`
-	ChunkRef     source.SourceRef         `json:"chunk_ref,omitempty"`
-	Kind         contracts.ArtifactKind   `json:"artifact_kind"`
-	Schema       contracts.ArtifactSchema `json:"schema"`
-	SchemaDigest string                   `json:"schema_digest"`
-	Content      binaryEnvelope           `json:"content"`
+	LaneID       string                          `json:"lane_id"`
+	ModuleKey    string                          `json:"module_key"`
+	SourceID     string                          `json:"source_id,omitempty"`
+	ChunkID      string                          `json:"chunk_id,omitempty"`
+	ChunkIndex   int                             `json:"chunk_index,omitempty"`
+	ChunkRef     source.SourceRef                `json:"chunk_ref,omitempty"`
+	Kind         contracts.ArtifactKind          `json:"artifact_kind"`
+	Schema       contracts.ArtifactSchema        `json:"schema"`
+	SchemaDigest string                          `json:"schema_digest"`
+	Content      binaryEnvelope                  `json:"content"`
+	Diagnostics  []pipeline.CheckpointDiagnostic `json:"diagnostics,omitempty"`
 }
 type artifactExtractEnvelope struct {
 	Outputs  []artifactCheckpointEnvelope `json:"outputs"`
 	Rejected []contracts.RejectedOutput   `json:"rejected,omitempty"`
-	Warnings []contracts.Warning          `json:"warnings,omitempty"`
 }
 type artifactSingleEnvelope struct {
-	Output   artifactCheckpointEnvelope `json:"output"`
-	Warnings []contracts.Warning        `json:"warnings,omitempty"`
+	Output artifactCheckpointEnvelope `json:"output"`
 }
 
 func artifactCheckpointEnvelopeFromOutput(output pipeline.CheckpointArtifact) artifactCheckpointEnvelope {
 	schema := contracts.CloneArtifactSchema(output.Artifact.Schema)
 	schema.JSONSchema = nil
-	return artifactCheckpointEnvelope{LaneID: output.LaneID, ModuleKey: output.ModuleKey, SourceID: output.SourceID, ChunkID: output.ChunkID, ChunkIndex: output.ChunkIndex, ChunkRef: output.ChunkRef, Kind: output.Artifact.Kind, Schema: schema, SchemaDigest: output.SchemaDigest, Content: binaryEnvelopeFromContent(output.Artifact.Content, output.Artifact.MediaType, output.Artifact.Metadata, nil)}
+	return artifactCheckpointEnvelope{LaneID: output.LaneID, ModuleKey: output.ModuleKey, SourceID: output.SourceID, ChunkID: output.ChunkID, ChunkIndex: output.ChunkIndex, ChunkRef: output.ChunkRef, Kind: output.Artifact.Kind, Schema: schema, SchemaDigest: output.SchemaDigest, Content: binaryEnvelopeFromContent(output.Artifact.Content, output.Artifact.MediaType, output.Artifact.Metadata), Diagnostics: cloneCheckpointDiagnostics(output.Diagnostics)}
 }
 func artifactCheckpointEnvelopes(outputs []pipeline.CheckpointArtifact) []artifactCheckpointEnvelope {
 	if len(outputs) == 0 {
@@ -293,6 +291,20 @@ func artifactCheckpointEnvelopes(outputs []pipeline.CheckpointArtifact) []artifa
 	}
 	return out
 }
+
+func cloneCheckpointDiagnostics(values []pipeline.CheckpointDiagnostic) []pipeline.CheckpointDiagnostic {
+	if len(values) == 0 {
+		return nil
+	}
+	cloned := make([]pipeline.CheckpointDiagnostic, len(values))
+	for index, value := range values {
+		cloned[index] = pipeline.CheckpointDiagnostic{
+			Diagnostic:   contracts.CloneProducerDiagnostics([]contracts.ProducerDiagnostic{value.Diagnostic})[0],
+			ValidatorKey: value.ValidatorKey,
+		}
+	}
+	return cloned
+}
 func artifactOutputDigests(outputs []pipeline.CheckpointArtifact) []pipeline.CheckpointFingerprint {
 	values := make([]pipeline.CheckpointFingerprint, 0, len(outputs))
 	for i, v := range outputs {
@@ -301,13 +313,12 @@ func artifactOutputDigests(outputs []pipeline.CheckpointArtifact) []pipeline.Che
 	return normalizeFingerprints(values)
 }
 
-func binaryEnvelopeFromContent(content []byte, mediaType string, metadata map[string]any, warnings []contracts.Warning) binaryEnvelope {
+func binaryEnvelopeFromContent(content []byte, mediaType string, metadata map[string]any) binaryEnvelope {
 	return binaryEnvelope{
 		ContentBase64: base64.StdEncoding.EncodeToString(content),
 		ContentDigest: contentDigest(content),
 		MediaType:     mediaType,
 		Metadata:      cloneMetadata(metadata),
-		Warnings:      cloneWarnings(warnings),
 	}
 }
 
@@ -334,13 +345,6 @@ func cloneSourceUnits(units []source.SourceUnit) []source.SourceUnit {
 	return out
 }
 
-func cloneWarnings(warnings []contracts.Warning) []contracts.Warning {
-	if len(warnings) == 0 {
-		return nil
-	}
-	return append([]contracts.Warning(nil), warnings...)
-}
-
 func cloneRejectedOutputs(rejected []contracts.RejectedOutput) []contracts.RejectedOutput {
 	if len(rejected) == 0 {
 		return nil
@@ -461,13 +465,10 @@ func statusForRejected(rejected []contracts.RejectedOutput) StageStatus {
 	return StatusSucceeded
 }
 
-func validationStatusString(warnings []contracts.Warning, rejected []contracts.RejectedOutput) string {
+func validationStatusString(rejected []contracts.RejectedOutput) string {
 	if len(rejected) > 0 {
 		return "rejected"
 	}
-	if len(warnings) > 0 {
-		return "approved_with_warnings"
-	}
 	return "approved"
 }
 

internal/framework/checkpoint/recorder_test.go

@@ -17,7 +17,7 @@ func TestRootBasedRecorderOutputIsReusable(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	if err := recorder.ExtractSucceeded("lane", "module", nil, nil, nil, nil); err != nil {
+	if err := recorder.ExtractSucceeded("lane", "module", nil, nil, nil); err != nil {
 		t.Fatal(err)
 	}
 	loader, err := NewFilesystemLoader(root, identity)
@@ -44,7 +44,7 @@ func TestStepAwareRecorderAndLoaderIsolateLaneState(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	if err := recorder.(pipeline.StepCheckpointRecorder).ExtractSucceededForStep("step-a", "lane", "module", nil, nil, nil, nil); err != nil {
+	if err := recorder.(pipeline.StepCheckpointRecorder).ExtractSucceededForStep("step-a", "lane", "module", nil, nil, nil); err != nil {
 		t.Fatal(err)
 	}
 	loader, err := NewFilesystemLoader(root, identity)
@@ -87,7 +87,7 @@ func TestStepAwareCheckpointPreservesDistinctDotIdentities(t *testing.T) {
 			LaneID: "lane", ModuleKey: "normalize-module", SourceID: "source", ChunkID: "chunk", ChunkRef: source.SourceRef{SourceID: "source", StartUnitID: 1, EndUnitID: 1}, SchemaDigest: "sha256:schema",
 			Artifact: contracts.SerializedArtifact{Kind: "kind", Schema: contracts.ArtifactSchema{ID: "schema", Name: "Schema", Version: "1"}, MediaType: "application/json", Content: []byte(test.content)},
 		}
-		if err := stepRecorder.NormalizeSucceededForStep(test.stepID, "lane", "normalize-module", nil, artifact, nil); err != nil {
+		if err := stepRecorder.NormalizeSucceededForStep(test.stepID, "lane", "normalize-module", nil, artifact); err != nil {
 			t.Fatalf("record %q: %v", test.stepID, err)
 		}
 	}
@@ -119,7 +119,7 @@ func TestStepAwareCheckpointPreservesDistinctDotIdentities(t *testing.T) {
 }
 
 func TestCheckpointSchemaCompatibilityIdentifiers(t *testing.T) {
-	if WorkspaceSchemaVersion != "notarius.workspace.v3" || WorkspaceSchemaVersionV2 != "notarius.workspace.v2" || WorkspaceSchemaVersionV1 != "notarius.workspace.v1" {
+	if WorkspaceSchemaVersion != "notarius.workspace.v4" || WorkspaceSchemaVersionV3 != "notarius.workspace.v3" || WorkspaceSchemaVersionV2 != "notarius.workspace.v2" || WorkspaceSchemaVersionV1 != "notarius.workspace.v1" {
 		t.Fatal("checkpoint schema identifiers are incorrect")
 	}
 }

internal/framework/chunkplan/store.go

@@ -14,6 +14,7 @@ import (
 	"syscall"
 
 	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
 )
 
@@ -340,6 +341,9 @@ func validateRecord(record pipeline.ChunkPlanRecord, requestedDigest string) err
 	if record.SchemaVersion != SchemaVersion {
 		return fmt.Errorf("schema_version %q is not supported", record.SchemaVersion)
 	}
+	if err := contracts.ValidateProducerDiagnostics(record.Diagnostics); err != nil {
+		return fmt.Errorf("diagnostics: %w", err)
+	}
 	if _, err := digestPathSegment(requestedDigest); err != nil {
 		return err
 	}

internal/framework/chunkplan/store_test.go

@@ -281,7 +281,7 @@ func TestFilesystemStoreReportsInvalidRecordsAsRecoverable(t *testing.T) {
 			return bytes.Replace(data, []byte(`{"schema_version"`), []byte(`{"SENTINEL_UNKNOWN_FIELD":true,"schema_version"`), 1)
 		}},
 		{name: "truncated JSON", mutate: func(data []byte) []byte { return data[:len(data)/2] }},
-		{name: "legacy v1 record", mutate: replaceJSON(`notarius.chunk-plan.v2`, `notarius.chunk-plan.v1`)},
+		{name: "legacy v2 record", mutate: replaceJSON(`notarius.chunk-plan.v3`, `notarius.chunk-plan.v2`)},
 		{name: "source mismatch", mutate: replaceJSON(testSourceDigest, "sha256:"+strings.Repeat("b", 64))},
 		{name: "plan digest mismatch", mutate: func(data []byte) []byte {
 			prefix := []byte(`"plan_digest":"sha256:`)
@@ -498,7 +498,13 @@ func testRecord(t *testing.T, value int) pipeline.ChunkPlanRecord {
 			References: []artifacts.ReferenceProvenance{{Stage: "chunk", SlotName: "guide", OriginType: "file", OriginURI: "file:///guide.txt", Digest: "sha256:reference"}},
 			Metadata:   map[string]any{"prompt_id": "test/prompt", "enabled": true},
 		},
-		Warnings:  []contracts.Warning{{Scope: "chunk/test", ReasonCode: "observed", Message: "warning"}},
+		Diagnostics: []contracts.ProducerDiagnostic{{
+			Disposition:     contracts.DiagnosticDispositionWarning,
+			Category:        contracts.DiagnosticCategoryConfiguration,
+			ReasonCode:      "empty_reference",
+			OccurrenceCount: 1,
+			Samples:         []contracts.DiagnosticSample{{Scope: "reference", Message: "Reference was empty."}},
+		}},
 		CreatedAt: time.Date(2026, 7, 18, 12, 0, 0, 0, time.UTC),
 	}
 }

internal/framework/contracts/contracts.go

@@ -162,9 +162,9 @@ type ChunkRequest struct {
 }
 
 type ChunkPlanResult struct {
-	Plan           source.ChunkPlan `json:"plan"`
-	Warnings       []Warning        `json:"warnings,omitempty"`
-	ModelCandidate *ModelCandidate  `json:"-"`
+	Plan           source.ChunkPlan     `json:"plan"`
+	Diagnostics    []ProducerDiagnostic `json:"diagnostics,omitempty"`
+	ModelCandidate *ModelCandidate      `json:"-"`
 }
 
 type Chunker interface {
@@ -283,25 +283,19 @@ const (
 )
 
 type ValidationResult struct {
-	Approved               bool      `json:"approved"`
-	ReasonCode             string    `json:"reason_code,omitempty"`
-	Message                string    `json:"message,omitempty"`
-	CorrectionGuidance     string    `json:"-"`
-	DiagnosticArtifactPath string    `json:"diagnostic_artifact_path,omitempty"`
-	Warnings               []Warning `json:"warnings,omitempty"`
-}
-
-type Warning struct {
-	Scope      string `json:"scope,omitempty"`
-	ReasonCode string `json:"reason_code"`
-	Message    string `json:"message"`
+	Approved               bool                 `json:"approved"`
+	ReasonCode             string               `json:"reason_code,omitempty"`
+	Message                string               `json:"message,omitempty"`
+	CorrectionGuidance     string               `json:"-"`
+	DiagnosticArtifactPath string               `json:"diagnostic_artifact_path,omitempty"`
+	Diagnostics            []ProducerDiagnostic `json:"diagnostics,omitempty"`
 }
 
 type OutputRequest struct {
 	Manifest                       artifacts.RunManifest `json:"manifest"`
 	NormalizeOutputs               []SerializedOutput    `json:"normalize_outputs,omitempty"`
 	Rejected                       []RejectedOutput      `json:"rejected,omitempty"`
-	Warnings                       []Warning             `json:"warnings,omitempty"`
+	Diagnostics                    DiagnosticCollection  `json:"diagnostics,omitempty"`
 	LLMProfile                     string                `json:"llm_profile,omitempty"`
 	StructuredOutputRepairAttempts *int                  `json:"structured_output_repair_attempts,omitempty"`
 	Metadata                       map[string]any        `json:"metadata,omitempty"`
@@ -316,8 +310,7 @@ type OutputFile struct {
 }
 
 type OutputResult struct {
-	Files    []OutputFile `json:"files,omitempty"`
-	Warnings []Warning    `json:"warnings,omitempty"`
+	Files []OutputFile `json:"files,omitempty"`
 }
 
 type OutputEncoder interface {

internal/framework/contracts/correction.go

@@ -126,6 +126,9 @@ func (candidate ModelCandidate) Validate() error {
 }
 
 func ValidateValidationResult(result ValidationResult) error {
+	if err := ValidateProducerDiagnostics(result.Diagnostics); err != nil {
+		return fmt.Errorf("validation diagnostics: %w", err)
+	}
 	if !result.Approved && result.ReasonCode == "" {
 		return errors.New("validation rejection reason code must not be empty")
 	}

internal/framework/contracts/correction_test.go

@@ -75,6 +75,9 @@ func TestCorrectionContractsRejectInvalidContent(t *testing.T) {
 		{"oversized correction guidance", func() error {
 			return ValidateValidationResult(ValidationResult{ReasonCode: "invalid", CorrectionGuidance: tooLongValidationGuidance})
 		}},
+		{"invalid diagnostics", func() error {
+			return ValidateValidationResult(ValidationResult{Approved: true, Diagnostics: []ProducerDiagnostic{{}}})
+		}},
 	} {
 		t.Run(test.name, func(t *testing.T) {
 			if err := test.call(); err == nil {

internal/framework/contracts/diagnostics.go

@@ -0,0 +1,401 @@
+package contracts
+
+import (
+	"errors"
+	"fmt"
+	"strings"
+	"unicode/utf8"
+)
+
+const (
+	MaxDiagnosticReasonCodeBytes = 128
+	MaxDiagnosticScopeBytes      = 512
+	MaxDiagnosticMessageBytes    = 4 * 1024
+	MaxDiagnosticSamples         = 3
+	MaxProducerDiagnosticGroups  = 64
+)
+
+// DiagnosticDisposition identifies the operator significance of a producer
+// finding. Warnings are reserved for process-level degradation or incomplete
+// configured work.
+type DiagnosticDisposition string
+
+const (
+	DiagnosticDispositionWarning     DiagnosticDisposition = "warning"
+	DiagnosticDispositionAdvisory    DiagnosticDisposition = "advisory"
+	DiagnosticDispositionObservation DiagnosticDisposition = "observation"
+)
+
+// DiagnosticCategory gives a stable, bounded classification for a producer
+// finding.
+type DiagnosticCategory string
+
+const (
+	DiagnosticCategoryConfiguration        DiagnosticCategory = "configuration"
+	DiagnosticCategoryDegradation          DiagnosticCategory = "degradation"
+	DiagnosticCategoryValidationIncomplete DiagnosticCategory = "validation_incomplete"
+	DiagnosticCategoryFallback             DiagnosticCategory = "fallback"
+	DiagnosticCategoryDataQuality          DiagnosticCategory = "data_quality"
+	DiagnosticCategoryNormalization        DiagnosticCategory = "normalization"
+)
+
+// DiagnosticOriginStage identifies the framework operation that promoted a
+// diagnostic. It is framework-owned rather than producer-owned.
+type DiagnosticOriginStage string
+
+const (
+	DiagnosticOriginStageReferences DiagnosticOriginStage = "references"
+	DiagnosticOriginStageChunk      DiagnosticOriginStage = "chunk"
+	DiagnosticOriginStageExtract    DiagnosticOriginStage = "extract"
+	DiagnosticOriginStageMerge      DiagnosticOriginStage = "merge"
+	DiagnosticOriginStageNormalize  DiagnosticOriginStage = "normalize"
+)
+
+// DiagnosticSample is a bounded, safe example of a diagnostic occurrence.
+// Chunk identity is attached by the framework when it promotes a producer
+// diagnostic into a final group.
+type DiagnosticSample struct {
+	Scope      string `json:"scope"`
+	Message    string `json:"message"`
+	ChunkID    string `json:"chunk_id,omitempty"`
+	ChunkIndex *int   `json:"chunk_index,omitempty"`
+}
+
+// ProducerDiagnostic is the locally grouped form returned by one producer or
+// validator. It intentionally has no pipeline origin.
+type ProducerDiagnostic struct {
+	Disposition        DiagnosticDisposition `json:"disposition"`
+	Category           DiagnosticCategory    `json:"category"`
+	ReasonCode         string                `json:"reason_code"`
+	OccurrenceCount    int                   `json:"occurrence_count"`
+	Samples            []DiagnosticSample    `json:"samples"`
+	OmittedSampleCount int                   `json:"omitted_sample_count"`
+}
+
+// DiagnosticOrigin is framework-owned context used to distinguish findings
+// from different pipeline locations during final aggregation.
+type DiagnosticOrigin struct {
+	Stage        DiagnosticOriginStage `json:"stage"`
+	StepID       string                `json:"step_id,omitempty"`
+	LaneID       string                `json:"lane_id,omitempty"`
+	ModuleKey    string                `json:"module_key,omitempty"`
+	ValidatorKey string                `json:"validator_key,omitempty"`
+}
+
+// DiagnosticGroup is a producer diagnostic after framework origin enrichment.
+type DiagnosticGroup struct {
+	Disposition        DiagnosticDisposition `json:"disposition"`
+	Category           DiagnosticCategory    `json:"category"`
+	ReasonCode         string                `json:"reason_code"`
+	Origin             DiagnosticOrigin      `json:"origin"`
+	OccurrenceCount    int                   `json:"occurrence_count"`
+	Samples            []DiagnosticSample    `json:"samples"`
+	OmittedSampleCount int                   `json:"omitted_sample_count"`
+}
+
+// DiagnosticCollection is the grouped collection supplied to later durable
+// and presentation boundaries. Global aggregation policy is applied by the
+// framework before it reaches those boundaries.
+type DiagnosticCollection struct {
+	Groups                       []DiagnosticGroup `json:"groups"`
+	Truncated                    bool              `json:"truncated"`
+	UnrepresentedOccurrenceCount int               `json:"unrepresented_occurrence_count"`
+}
+
+// DiagnosticProjection is the validated warning/non-warning view used by
+// durable and presentation boundaries. Occurrence totals are checked before
+// they leave the framework contract.
+type DiagnosticProjection struct {
+	Warnings                  []DiagnosticGroup
+	Diagnostics               []DiagnosticGroup
+	WarningOccurrenceCount    int
+	DiagnosticOccurrenceCount int
+}
+
+// Validate checks a producer-local diagnostic against the public safety and
+// classification contract.
+func (diagnostic ProducerDiagnostic) Validate() error {
+	return validateDiagnostic(
+		diagnostic.Disposition,
+		diagnostic.Category,
+		diagnostic.ReasonCode,
+		diagnostic.OccurrenceCount,
+		diagnostic.Samples,
+		diagnostic.OmittedSampleCount,
+		false,
+	)
+}
+
+// ValidateProducerDiagnostics validates the complete set returned by one
+// producer or validator result.
+func ValidateProducerDiagnostics(diagnostics []ProducerDiagnostic) error {
+	if len(diagnostics) > MaxProducerDiagnosticGroups {
+		return errors.New("producer diagnostics exceed maximum group count")
+	}
+	for index, diagnostic := range diagnostics {
+		if err := diagnostic.Validate(); err != nil {
+			return fmt.Errorf("producer diagnostic %d: %w", index, err)
+		}
+	}
+	return nil
+}
+
+// Validate checks framework-owned origin fields.
+func (origin DiagnosticOrigin) Validate() error {
+	switch origin.Stage {
+	case DiagnosticOriginStageReferences, DiagnosticOriginStageChunk, DiagnosticOriginStageExtract, DiagnosticOriginStageMerge, DiagnosticOriginStageNormalize:
+	default:
+		return errors.New("diagnostic origin stage is invalid")
+	}
+	for _, field := range []struct {
+		name  string
+		value string
+	}{
+		{name: "diagnostic origin step ID", value: origin.StepID},
+		{name: "diagnostic origin lane ID", value: origin.LaneID},
+		{name: "diagnostic origin module key", value: origin.ModuleKey},
+		{name: "diagnostic origin validator key", value: origin.ValidatorKey},
+	} {
+		if field.value != "" && (!utf8.ValidString(field.value) || strings.TrimSpace(field.value) == "") {
+			return fmt.Errorf("%s must be valid nonblank UTF-8 when present", field.name)
+		}
+	}
+	return nil
+}
+
+// Validate checks a final origin-enriched group.
+func (group DiagnosticGroup) Validate() error {
+	if err := group.Origin.Validate(); err != nil {
+		return err
+	}
+	return validateDiagnostic(
+		group.Disposition,
+		group.Category,
+		group.ReasonCode,
+		group.OccurrenceCount,
+		group.Samples,
+		group.OmittedSampleCount,
+		true,
+	)
+}
+
+// Validate checks the collection shape without imposing later global
+// aggregation limits.
+func (collection DiagnosticCollection) Validate() error {
+	if collection.UnrepresentedOccurrenceCount < 0 {
+		return errors.New("diagnostic collection unrepresented occurrence count must not be negative")
+	}
+	if !collection.Truncated && collection.UnrepresentedOccurrenceCount != 0 {
+		return errors.New("diagnostic collection has unrepresented occurrences without truncation")
+	}
+	seen := make(map[diagnosticGroupKey]struct{}, len(collection.Groups))
+	for index, group := range collection.Groups {
+		if err := group.Validate(); err != nil {
+			return fmt.Errorf("diagnostic group %d: %w", index, err)
+		}
+		key := diagnosticGroupKeyFromGroup(group)
+		if _, exists := seen[key]; exists {
+			return errors.New("diagnostic collection contains duplicate group identity")
+		}
+		seen[key] = struct{}{}
+	}
+	return nil
+}
+
+// ProjectDiagnosticCollection validates, partitions, and totals one finalized
+// collection. Unrepresented occurrences belong to the non-warning projection.
+func ProjectDiagnosticCollection(collection DiagnosticCollection) (DiagnosticProjection, error) {
+	if err := collection.Validate(); err != nil {
+		return DiagnosticProjection{}, err
+	}
+	projection := DiagnosticProjection{
+		Warnings:    make([]DiagnosticGroup, 0),
+		Diagnostics: make([]DiagnosticGroup, 0),
+	}
+	for _, group := range collection.Groups {
+		if group.Disposition == DiagnosticDispositionWarning {
+			count, err := addDiagnosticOccurrences(projection.WarningOccurrenceCount, group.OccurrenceCount)
+			if err != nil {
+				return DiagnosticProjection{}, fmt.Errorf("warning occurrences: %w", err)
+			}
+			projection.WarningOccurrenceCount = count
+			projection.Warnings = append(projection.Warnings, cloneDiagnosticGroup(group))
+			continue
+		}
+		count, err := addDiagnosticOccurrences(projection.DiagnosticOccurrenceCount, group.OccurrenceCount)
+		if err != nil {
+			return DiagnosticProjection{}, fmt.Errorf("diagnostic occurrences: %w", err)
+		}
+		projection.DiagnosticOccurrenceCount = count
+		projection.Diagnostics = append(projection.Diagnostics, cloneDiagnosticGroup(group))
+	}
+	count, err := addDiagnosticOccurrences(projection.DiagnosticOccurrenceCount, collection.UnrepresentedOccurrenceCount)
+	if err != nil {
+		return DiagnosticProjection{}, fmt.Errorf("diagnostic occurrences: %w", err)
+	}
+	projection.DiagnosticOccurrenceCount = count
+	return projection, nil
+}
+
+// CloneProducerDiagnostics returns independent diagnostic slice ownership.
+func CloneProducerDiagnostics(diagnostics []ProducerDiagnostic) []ProducerDiagnostic {
+	if len(diagnostics) == 0 {
+		return nil
+	}
+	cloned := make([]ProducerDiagnostic, len(diagnostics))
+	for index, diagnostic := range diagnostics {
+		cloned[index] = cloneProducerDiagnostic(diagnostic)
+	}
+	return cloned
+}
+
+// CloneDiagnosticCollection returns independent collection ownership.
+func CloneDiagnosticCollection(collection DiagnosticCollection) DiagnosticCollection {
+	groups := collection.Groups
+	collection.Groups = make([]DiagnosticGroup, len(groups))
+	for index, group := range groups {
+		collection.Groups[index] = cloneDiagnosticGroup(group)
+	}
+	return collection
+}
+
+func validateDiagnostic(disposition DiagnosticDisposition, category DiagnosticCategory, reasonCode string, occurrenceCount int, samples []DiagnosticSample, omittedSampleCount int, allowChunkContext bool) error {
+	if !diagnosticCategoryAllowed(disposition, category) {
+		return errors.New("diagnostic disposition and category combination is invalid")
+	}
+	if err := validateDiagnosticText(reasonCode, MaxDiagnosticReasonCodeBytes, "diagnostic reason code"); err != nil {
+		return err
+	}
+	if occurrenceCount <= 0 {
+		return errors.New("diagnostic occurrence count must be positive")
+	}
+	if len(samples) == 0 {
+		return errors.New("diagnostic samples must not be empty")
+	}
+	if len(samples) > MaxDiagnosticSamples {
+		return errors.New("diagnostic samples exceed maximum count")
+	}
+	seen := make(map[diagnosticSampleKey]struct{}, len(samples))
+	for index, sample := range samples {
+		if err := validateDiagnosticSample(sample, allowChunkContext); err != nil {
+			return fmt.Errorf("diagnostic sample %d: %w", index, err)
+		}
+		key := diagnosticSampleKeyFromSample(sample)
+		if _, exists := seen[key]; exists {
+			return errors.New("diagnostic samples must be distinct")
+		}
+		seen[key] = struct{}{}
+	}
+	if occurrenceCount < len(samples) {
+		return errors.New("diagnostic occurrence count is smaller than sample count")
+	}
+	if omittedSampleCount != occurrenceCount-len(samples) {
+		return errors.New("diagnostic omitted sample count is inconsistent")
+	}
+	return nil
+}
+
+func diagnosticCategoryAllowed(disposition DiagnosticDisposition, category DiagnosticCategory) bool {
+	switch disposition {
+	case DiagnosticDispositionWarning:
+		return category == DiagnosticCategoryConfiguration || category == DiagnosticCategoryDegradation || category == DiagnosticCategoryValidationIncomplete || category == DiagnosticCategoryFallback
+	case DiagnosticDispositionAdvisory:
+		return category == DiagnosticCategoryDataQuality
+	case DiagnosticDispositionObservation:
+		return category == DiagnosticCategoryNormalization
+	default:
+		return false
+	}
+}
+
+func validateDiagnosticSample(sample DiagnosticSample, allowChunkContext bool) error {
+	if err := validateDiagnosticText(sample.Scope, MaxDiagnosticScopeBytes, "diagnostic sample scope"); err != nil {
+		return err
+	}
+	if err := validateDiagnosticText(sample.Message, MaxDiagnosticMessageBytes, "diagnostic sample message"); err != nil {
+		return err
+	}
+	if !allowChunkContext && (sample.ChunkID != "" || sample.ChunkIndex != nil) {
+		return errors.New("producer diagnostic sample must not include framework chunk context")
+	}
+	if sample.ChunkID != "" && (!utf8.ValidString(sample.ChunkID) || strings.TrimSpace(sample.ChunkID) == "") {
+		return errors.New("diagnostic sample chunk ID must be valid nonblank UTF-8 when present")
+	}
+	if sample.ChunkIndex != nil && *sample.ChunkIndex < 0 {
+		return errors.New("diagnostic sample chunk index must not be negative")
+	}
+	return nil
+}
+
+func validateDiagnosticText(value string, maximum int, name string) error {
+	if !utf8.ValidString(value) {
+		return fmt.Errorf("%s must be valid UTF-8", name)
+	}
+	if strings.TrimSpace(value) == "" {
+		return fmt.Errorf("%s must not be blank", name)
+	}
+	if len(value) > maximum {
+		return fmt.Errorf("%s exceeds maximum length", name)
+	}
+	return nil
+}
+
+func cloneProducerDiagnostic(diagnostic ProducerDiagnostic) ProducerDiagnostic {
+	diagnostic.Samples = cloneDiagnosticSamples(diagnostic.Samples)
+	return diagnostic
+}
+
+func cloneDiagnosticGroup(group DiagnosticGroup) DiagnosticGroup {
+	group.Samples = cloneDiagnosticSamples(group.Samples)
+	return group
+}
+
+func addDiagnosticOccurrences(current, incoming int) (int, error) {
+	if incoming > int(^uint(0)>>1)-current {
+		return 0, errors.New("occurrence count overflow")
+	}
+	return current + incoming, nil
+}
+
+func cloneDiagnosticSamples(samples []DiagnosticSample) []DiagnosticSample {
+	if len(samples) == 0 {
+		return nil
+	}
+	cloned := make([]DiagnosticSample, len(samples))
+	for index, sample := range samples {
+		if sample.ChunkIndex != nil {
+			chunkIndex := *sample.ChunkIndex
+			sample.ChunkIndex = &chunkIndex
+		}
+		cloned[index] = sample
+	}
+	return cloned
+}
+
+type diagnosticSampleKey struct {
+	scope         string
+	message       string
+	chunkID       string
+	chunkIndex    int
+	hasChunkIndex bool
+}
+
+func diagnosticSampleKeyFromSample(sample DiagnosticSample) diagnosticSampleKey {
+	key := diagnosticSampleKey{scope: sample.Scope, message: sample.Message, chunkID: sample.ChunkID}
+	if sample.ChunkIndex != nil {
+		key.chunkIndex = *sample.ChunkIndex
+		key.hasChunkIndex = true
+	}
+	return key
+}
+
+type diagnosticGroupKey struct {
+	disposition DiagnosticDisposition
+	category    DiagnosticCategory
+	reasonCode  string
+	origin      DiagnosticOrigin
+}
+
+func diagnosticGroupKeyFromGroup(group DiagnosticGroup) diagnosticGroupKey {
+	return diagnosticGroupKey{disposition: group.Disposition, category: group.Category, reasonCode: group.ReasonCode, origin: group.Origin}
+}

internal/framework/contracts/diagnostics_test.go

@@ -0,0 +1,198 @@
+package contracts
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestProducerDiagnosticValidationAcceptsClassificationMatrix(t *testing.T) {
+	for _, test := range []struct {
+		name        string
+		disposition DiagnosticDisposition
+		category    DiagnosticCategory
+	}{
+		{name: "configuration warning", disposition: DiagnosticDispositionWarning, category: DiagnosticCategoryConfiguration},
+		{name: "degradation warning", disposition: DiagnosticDispositionWarning, category: DiagnosticCategoryDegradation},
+		{name: "incomplete validation warning", disposition: DiagnosticDispositionWarning, category: DiagnosticCategoryValidationIncomplete},
+		{name: "fallback warning", disposition: DiagnosticDispositionWarning, category: DiagnosticCategoryFallback},
+		{name: "quality advisory", disposition: DiagnosticDispositionAdvisory, category: DiagnosticCategoryDataQuality},
+		{name: "normalization observation", disposition: DiagnosticDispositionObservation, category: DiagnosticCategoryNormalization},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			diagnostic := validProducerDiagnostic()
+			diagnostic.Disposition = test.disposition
+			diagnostic.Category = test.category
+			if err := diagnostic.Validate(); err != nil {
+				t.Fatalf("Validate() error = %v", err)
+			}
+		})
+	}
+}
+
+func TestProducerDiagnosticValidationRejectsInvalidFieldsAndCounts(t *testing.T) {
+	tooLongReason := strings.Repeat("r", MaxDiagnosticReasonCodeBytes+1)
+	tooLongScope := strings.Repeat("s", MaxDiagnosticScopeBytes+1)
+	tooLongMessage := strings.Repeat("m", MaxDiagnosticMessageBytes+1)
+	for _, test := range []struct {
+		name   string
+		mutate func(*ProducerDiagnostic)
+	}{
+		{name: "invalid classification", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.Category = DiagnosticCategoryDataQuality }},
+		{name: "blank reason", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.ReasonCode = " \t" }},
+		{name: "invalid reason UTF-8", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.ReasonCode = string([]byte{0xff}) }},
+		{name: "oversized reason", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.ReasonCode = tooLongReason }},
+		{name: "blank scope", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.Samples[0].Scope = "\n" }},
+		{name: "invalid scope UTF-8", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.Samples[0].Scope = string([]byte{0xff}) }},
+		{name: "oversized scope", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.Samples[0].Scope = tooLongScope }},
+		{name: "blank message", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.Samples[0].Message = " " }},
+		{name: "invalid message UTF-8", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.Samples[0].Message = string([]byte{0xff}) }},
+		{name: "oversized message", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.Samples[0].Message = tooLongMessage }},
+		{name: "zero occurrences", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.OccurrenceCount = 0 }},
+		{name: "missing samples", mutate: func(diagnostic *ProducerDiagnostic) {
+			diagnostic.Samples = nil
+			diagnostic.OmittedSampleCount = diagnostic.OccurrenceCount
+		}},
+		{name: "too many samples", mutate: func(diagnostic *ProducerDiagnostic) {
+			diagnostic.OccurrenceCount = 4
+			diagnostic.Samples = []DiagnosticSample{{Scope: "one", Message: "one"}, {Scope: "two", Message: "two"}, {Scope: "three", Message: "three"}, {Scope: "four", Message: "four"}}
+			diagnostic.OmittedSampleCount = 0
+		}},
+		{name: "duplicate samples", mutate: func(diagnostic *ProducerDiagnostic) {
+			diagnostic.OccurrenceCount = 2
+			diagnostic.Samples = []DiagnosticSample{{Scope: "scope", Message: "message"}, {Scope: "scope", Message: "message"}}
+			diagnostic.OmittedSampleCount = 0
+		}},
+		{name: "inconsistent omission", mutate: func(diagnostic *ProducerDiagnostic) { diagnostic.OmittedSampleCount = 1 }},
+		{name: "producer chunk context", mutate: func(diagnostic *ProducerDiagnostic) { chunkIndex := 0; diagnostic.Samples[0].ChunkIndex = &chunkIndex }},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			diagnostic := validProducerDiagnostic()
+			test.mutate(&diagnostic)
+			if err := diagnostic.Validate(); err == nil {
+				t.Fatal("Validate() error = nil, want invalid diagnostic error")
+			}
+		})
+	}
+}
+
+func TestValidateProducerDiagnosticsEnforcesLocalGroupBound(t *testing.T) {
+	diagnostics := make([]ProducerDiagnostic, MaxProducerDiagnosticGroups)
+	for index := range diagnostics {
+		diagnostics[index] = validProducerDiagnostic()
+		diagnostics[index].ReasonCode = "reason-" + string(rune('a'+index))
+	}
+	if err := ValidateProducerDiagnostics(diagnostics); err != nil {
+		t.Fatalf("ValidateProducerDiagnostics() error = %v", err)
+	}
+	diagnostics = append(diagnostics, validProducerDiagnostic())
+	if err := ValidateProducerDiagnostics(diagnostics); err == nil {
+		t.Fatal("ValidateProducerDiagnostics() error = nil, want excessive-group error")
+	}
+}
+
+func TestDiagnosticGroupValidationPreservesChunkIndexZero(t *testing.T) {
+	chunkIndex := 0
+	group := DiagnosticGroup{
+		Disposition:     DiagnosticDispositionAdvisory,
+		Category:        DiagnosticCategoryDataQuality,
+		ReasonCode:      "unresolved",
+		Origin:          DiagnosticOrigin{Stage: DiagnosticOriginStageExtract, StepID: "extract", LaneID: "spells", ModuleKey: "dnd/spells", ValidatorKey: "dnd/spells/source-relatedness"},
+		OccurrenceCount: 1,
+		Samples:         []DiagnosticSample{{Scope: "spells[0]", Message: "Spell was not found", ChunkID: "chunk-1", ChunkIndex: &chunkIndex}},
+	}
+	if err := group.Validate(); err != nil {
+		t.Fatalf("Validate() error = %v", err)
+	}
+	collection := DiagnosticCollection{Groups: []DiagnosticGroup{group}}
+	if err := collection.Validate(); err != nil {
+		t.Fatalf("DiagnosticCollection.Validate() error = %v", err)
+	}
+}
+
+func TestDiagnosticGroupRejectsRepeatedSampleWithEqualChunkIndex(t *testing.T) {
+	firstIndex := 0
+	secondIndex := 0
+	group := DiagnosticGroup{
+		Disposition:     DiagnosticDispositionAdvisory,
+		Category:        DiagnosticCategoryDataQuality,
+		ReasonCode:      "unresolved",
+		Origin:          DiagnosticOrigin{Stage: DiagnosticOriginStageExtract, StepID: "extract", LaneID: "spells", ModuleKey: "dnd/spells"},
+		OccurrenceCount: 2,
+		Samples: []DiagnosticSample{
+			{Scope: "spells[0]", Message: "Spell was not found", ChunkID: "chunk-1", ChunkIndex: &firstIndex},
+			{Scope: "spells[0]", Message: "Spell was not found", ChunkID: "chunk-1", ChunkIndex: &secondIndex},
+		},
+	}
+	if err := group.Validate(); err == nil {
+		t.Fatal("Validate() error = nil, want duplicate sample error")
+	}
+}
+
+func TestCloneDiagnosticCollectionOwnsGroupsAndChunkIndex(t *testing.T) {
+	chunkIndex := 0
+	collection := DiagnosticCollection{Groups: []DiagnosticGroup{{
+		Disposition:     DiagnosticDispositionAdvisory,
+		Category:        DiagnosticCategoryDataQuality,
+		ReasonCode:      "unresolved",
+		Origin:          DiagnosticOrigin{Stage: DiagnosticOriginStageExtract, StepID: "extract", LaneID: "spells", ModuleKey: "dnd/spells"},
+		OccurrenceCount: 1,
+		Samples:         []DiagnosticSample{{Scope: "spells[0]", Message: "Spell was not found", ChunkIndex: &chunkIndex}},
+	}}}
+	cloned := CloneDiagnosticCollection(collection)
+	collection.Groups[0].Samples[0].Message = "changed"
+	*collection.Groups[0].Samples[0].ChunkIndex = 1
+	if got := cloned.Groups[0].Samples[0]; got.Message != "Spell was not found" || got.ChunkIndex == nil || *got.ChunkIndex != 0 {
+		t.Fatalf("cloned sample = %#v, want independently owned original", got)
+	}
+}
+
+func TestProjectDiagnosticCollectionPartitionsAndChecksTotals(t *testing.T) {
+	warning := validDiagnosticGroup(DiagnosticDispositionWarning, DiagnosticCategoryFallback, "fallback", 2)
+	diagnostic := validDiagnosticGroup(DiagnosticDispositionAdvisory, DiagnosticCategoryDataQuality, "quality", 3)
+	collection := DiagnosticCollection{
+		Groups:                       []DiagnosticGroup{warning, diagnostic},
+		Truncated:                    true,
+		UnrepresentedOccurrenceCount: 4,
+	}
+	projection, err := ProjectDiagnosticCollection(collection)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(projection.Warnings) != 1 || len(projection.Diagnostics) != 1 || projection.WarningOccurrenceCount != 2 || projection.DiagnosticOccurrenceCount != 7 {
+		t.Fatalf("projection = %#v, want partitioned exact totals", projection)
+	}
+	collection.Groups[0].Samples[0].Message = "mutated"
+	if projection.Warnings[0].Samples[0].Message != "message" {
+		t.Fatal("projection retained caller-owned sample storage")
+	}
+
+	overflow := DiagnosticCollection{Groups: []DiagnosticGroup{
+		validDiagnosticGroup(DiagnosticDispositionWarning, DiagnosticCategoryFallback, "first", int(^uint(0)>>1)),
+		validDiagnosticGroup(DiagnosticDispositionWarning, DiagnosticCategoryFallback, "second", 1),
+	}}
+	if _, err := ProjectDiagnosticCollection(overflow); err == nil {
+		t.Fatal("ProjectDiagnosticCollection() overflow error = nil")
+	}
+}
+
+func validDiagnosticGroup(disposition DiagnosticDisposition, category DiagnosticCategory, reason string, occurrences int) DiagnosticGroup {
+	return DiagnosticGroup{
+		Disposition:        disposition,
+		Category:           category,
+		ReasonCode:         reason,
+		Origin:             DiagnosticOrigin{Stage: DiagnosticOriginStageNormalize, StepID: "step", LaneID: "lane", ModuleKey: "module"},
+		OccurrenceCount:    occurrences,
+		Samples:            []DiagnosticSample{{Scope: "scope", Message: "message"}},
+		OmittedSampleCount: occurrences - 1,
+	}
+}
+
+func validProducerDiagnostic() ProducerDiagnostic {
+	return ProducerDiagnostic{
+		Disposition:     DiagnosticDispositionWarning,
+		Category:        DiagnosticCategoryConfiguration,
+		ReasonCode:      "empty_reference",
+		OccurrenceCount: 1,
+		Samples:         []DiagnosticSample{{Scope: "references.glossary", Message: "Reference is empty"}},
+	}
+}

internal/framework/contracts/typed_pipeline.go

@@ -48,7 +48,7 @@ type TypedExtractionRequest struct {
 
 type TypedExtractionResult[T any] struct {
 	Value          T
-	Warnings       []Warning
+	Diagnostics    []ProducerDiagnostic
 	ModelCandidate *ModelCandidate
 }
 
@@ -73,7 +73,7 @@ type TypedMergeRequest[T any] struct {
 
 type TypedMergeResult[T any] struct {
 	Value          T
-	Warnings       []Warning
+	Diagnostics    []ProducerDiagnostic
 	ModelCandidate *ModelCandidate
 }
 
@@ -97,24 +97,26 @@ type TypedNormalizeRequest[T any] struct {
 
 type TypedNormalizeResult[T any] struct {
 	Value          T
-	Warnings       []Warning
+	Diagnostics    []ProducerDiagnostic
 	Retry          *NormalizeRetry
 	ModelCandidate *ModelCandidate
 }
 
-// Normalize retry diagnostic limits bound module-provided values before the
-// framework persists them in debug artifacts.
+// Normalize retry limits bound module-provided control and diagnostic text
+// before the framework consumes or records it.
 const (
-	MaxNormalizeRetryReasonCodeBytes = 128
-	MaxNormalizeRetryMessageBytes    = 4096
+	MaxNormalizeRetryReasonCodeBytes         = 128
+	MaxNormalizeRetryMessageBytes            = 4096
+	MaxNormalizeRetryCorrectionGuidanceBytes = 4096
 )
 
 // NormalizeRetry asks the framework to retry normalization while retaining a
 // safe candidate for acceptance if the retry budget is exhausted.
 type NormalizeRetry struct {
-	ReasonCode       string
-	Message          string
-	FallbackWarnings []Warning
+	ReasonCode          string
+	Message             string
+	CorrectionGuidance  string
+	FallbackDiagnostics []ProducerDiagnostic
 }
 
 type Normalizer[T any] interface {

internal/framework/diagnostics/aggregator.go

@@ -0,0 +1,164 @@
+package diagnostics
+
+import (
+	"errors"
+	"fmt"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+)
+
+const (
+	MaxWarningGroups    = 128
+	MaxNonWarningGroups = 256
+)
+
+// Aggregator merges origin-enriched diagnostics in caller-supplied canonical
+// order. Its zero value is ready for use.
+type Aggregator struct {
+	groups                   []contracts.DiagnosticGroup
+	indices                  map[groupKey]int
+	warningGroups            int
+	nonWarningGroups         int
+	warningOccurrences       int
+	nonWarningOccurrences    int
+	unrepresentedOccurrences int
+}
+
+// Add validates and incorporates one final diagnostic group. Actionable
+// warnings cannot overflow; later non-warning groups are represented by exact
+// unrepresented-occurrence metadata once their fixed bound is reached.
+func (aggregator *Aggregator) Add(group contracts.DiagnosticGroup) error {
+	if err := group.Validate(); err != nil {
+		return fmt.Errorf("diagnostic group: %w", err)
+	}
+	if aggregator.indices == nil {
+		aggregator.indices = make(map[groupKey]int)
+	}
+	key := groupKeyFromGroup(group)
+	if index, exists := aggregator.indices[key]; exists {
+		if err := aggregator.checkOccurrenceTotal(group.Disposition, group.OccurrenceCount); err != nil {
+			return err
+		}
+		if err := aggregator.merge(index, group); err != nil {
+			return err
+		}
+		aggregator.addOccurrenceTotal(group.Disposition, group.OccurrenceCount)
+		return nil
+	}
+	if group.Disposition == contracts.DiagnosticDispositionWarning {
+		if aggregator.warningGroups >= MaxWarningGroups {
+			return errors.New("diagnostic warning groups exceed maximum count")
+		}
+	} else if aggregator.nonWarningGroups >= MaxNonWarningGroups {
+		if err := aggregator.checkOccurrenceTotal(group.Disposition, group.OccurrenceCount); err != nil {
+			return err
+		}
+		aggregator.addOccurrenceTotal(group.Disposition, group.OccurrenceCount)
+		return aggregator.addUnrepresented(group.OccurrenceCount)
+	}
+	if err := aggregator.checkOccurrenceTotal(group.Disposition, group.OccurrenceCount); err != nil {
+		return err
+	}
+	if group.Disposition == contracts.DiagnosticDispositionWarning {
+		aggregator.warningGroups++
+	} else {
+		aggregator.nonWarningGroups++
+	}
+	aggregator.addOccurrenceTotal(group.Disposition, group.OccurrenceCount)
+	aggregator.indices[key] = len(aggregator.groups)
+	aggregator.groups = append(aggregator.groups, contracts.CloneDiagnosticCollection(contracts.DiagnosticCollection{Groups: []contracts.DiagnosticGroup{group}}).Groups[0])
+	return nil
+}
+
+func (aggregator *Aggregator) checkOccurrenceTotal(disposition contracts.DiagnosticDisposition, count int) error {
+	current := aggregator.nonWarningOccurrences
+	if disposition == contracts.DiagnosticDispositionWarning {
+		current = aggregator.warningOccurrences
+	}
+	if count > maximumInt()-current {
+		return errors.New("diagnostic occurrence count overflow")
+	}
+	return nil
+}
+
+func (aggregator *Aggregator) addOccurrenceTotal(disposition contracts.DiagnosticDisposition, count int) {
+	if disposition == contracts.DiagnosticDispositionWarning {
+		aggregator.warningOccurrences += count
+		return
+	}
+	aggregator.nonWarningOccurrences += count
+}
+
+// Collection returns an independently owned grouped result in first-occurrence
+// order.
+func (aggregator *Aggregator) Collection() contracts.DiagnosticCollection {
+	if aggregator == nil {
+		return contracts.DiagnosticCollection{}
+	}
+	return contracts.CloneDiagnosticCollection(contracts.DiagnosticCollection{
+		Groups:                       aggregator.groups,
+		Truncated:                    aggregator.unrepresentedOccurrences > 0,
+		UnrepresentedOccurrenceCount: aggregator.unrepresentedOccurrences,
+	})
+}
+
+func (aggregator *Aggregator) merge(index int, incoming contracts.DiagnosticGroup) error {
+	current := &aggregator.groups[index]
+	if incoming.OccurrenceCount > maximumInt()-current.OccurrenceCount {
+		return errors.New("diagnostic occurrence count overflow")
+	}
+	current.OccurrenceCount += incoming.OccurrenceCount
+	for _, sample := range incoming.Samples {
+		if len(current.Samples) == contracts.MaxDiagnosticSamples || containsGroupSample(current.Samples, sample) {
+			continue
+		}
+		current.Samples = append(current.Samples, cloneGroupSample(sample))
+	}
+	current.OmittedSampleCount = current.OccurrenceCount - len(current.Samples)
+	return nil
+}
+
+func (aggregator *Aggregator) addUnrepresented(count int) error {
+	if count > maximumInt()-aggregator.unrepresentedOccurrences {
+		return errors.New("diagnostic unrepresented occurrence count overflow")
+	}
+	aggregator.unrepresentedOccurrences += count
+	return nil
+}
+
+func containsGroupSample(samples []contracts.DiagnosticSample, candidate contracts.DiagnosticSample) bool {
+	for _, sample := range samples {
+		if sample.Scope != candidate.Scope || sample.Message != candidate.Message || sample.ChunkID != candidate.ChunkID {
+			continue
+		}
+		if sample.ChunkIndex == nil || candidate.ChunkIndex == nil {
+			if sample.ChunkIndex == candidate.ChunkIndex {
+				return true
+			}
+			continue
+		}
+		if *sample.ChunkIndex == *candidate.ChunkIndex {
+			return true
+		}
+	}
+	return false
+}
+
+func cloneGroupSample(sample contracts.DiagnosticSample) contracts.DiagnosticSample {
+	if sample.ChunkIndex != nil {
+		value := *sample.ChunkIndex
+		sample.ChunkIndex = &value
+	}
+	return sample
+}
+
+type groupKey struct {
+	disposition contracts.DiagnosticDisposition
+	category    contracts.DiagnosticCategory
+	reasonCode  string
+	origin      contracts.DiagnosticOrigin
+}
+
+func groupKeyFromGroup(group contracts.DiagnosticGroup) groupKey {
+	return groupKey{disposition: group.Disposition, category: group.Category, reasonCode: group.ReasonCode, origin: group.Origin}
+}

internal/framework/diagnostics/aggregator_test.go

@@ -0,0 +1,89 @@
+package diagnostics
+
+import (
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+)
+
+func TestAggregatorMergesByOriginAndPreservesFirstOccurrenceOrder(t *testing.T) {
+	aggregator := Aggregator{}
+	for _, group := range []contracts.DiagnosticGroup{
+		groupForAggregation("first", "message one", contracts.DiagnosticDispositionAdvisory, contracts.DiagnosticCategoryDataQuality, "dnd/spells"),
+		groupForAggregation("second", "message two", contracts.DiagnosticDispositionWarning, contracts.DiagnosticCategoryFallback, "dnd/items"),
+		groupForAggregation("third", "message three", contracts.DiagnosticDispositionAdvisory, contracts.DiagnosticCategoryDataQuality, "dnd/spells"),
+	} {
+		if err := aggregator.Add(group); err != nil {
+			t.Fatalf("Add() error = %v", err)
+		}
+	}
+	collection := aggregator.Collection()
+	if len(collection.Groups) != 2 {
+		t.Fatalf("group count = %d, want 2", len(collection.Groups))
+	}
+	if got := []string{collection.Groups[0].Origin.ModuleKey, collection.Groups[1].Origin.ModuleKey}; !equalStrings(got, []string{"dnd/spells", "dnd/items"}) {
+		t.Fatalf("group order = %#v, want first occurrence order", got)
+	}
+	if collection.Groups[0].OccurrenceCount != 2 || collection.Groups[0].OmittedSampleCount != 0 {
+		t.Fatalf("merged group = %#v, want two represented occurrences", collection.Groups[0])
+	}
+	if got := []string{collection.Groups[0].Samples[0].Scope, collection.Groups[0].Samples[1].Scope}; !equalStrings(got, []string{"first", "third"}) {
+		t.Fatalf("merged samples = %#v, want first occurrence order", got)
+	}
+}
+
+func TestAggregatorEnforcesWarningBoundAndTruncatesOnlyNonWarnings(t *testing.T) {
+	warnings := Aggregator{}
+	for index := 0; index < MaxWarningGroups; index++ {
+		group := groupForAggregation("scope", "message", contracts.DiagnosticDispositionWarning, contracts.DiagnosticCategoryFallback, "module")
+		group.ReasonCode = "warning-" + string(rune('a'+index))
+		if err := warnings.Add(group); err != nil {
+			t.Fatalf("warning Add(%d) error = %v", index, err)
+		}
+	}
+	if err := warnings.Add(groupForAggregation("overflow", "overflow", contracts.DiagnosticDispositionWarning, contracts.DiagnosticCategoryFallback, "overflow")); err == nil {
+		t.Fatal("warning overflow error = nil, want error")
+	}
+
+	nonWarnings := Aggregator{}
+	for index := 0; index < MaxNonWarningGroups+3; index++ {
+		group := groupForAggregation("scope", "message", contracts.DiagnosticDispositionAdvisory, contracts.DiagnosticCategoryDataQuality, "module")
+		group.ReasonCode = "advisory-" + string(rune('a'+index))
+		if err := nonWarnings.Add(group); err != nil {
+			t.Fatalf("non-warning Add(%d) error = %v", index, err)
+		}
+	}
+	collection := nonWarnings.Collection()
+	if len(collection.Groups) != MaxNonWarningGroups || !collection.Truncated || collection.UnrepresentedOccurrenceCount != 3 {
+		t.Fatalf("collection = %#v, want bounded non-warning groups and three unrepresented occurrences", collection)
+	}
+}
+
+func TestAggregatorRejectsOccurrenceTotalOverflow(t *testing.T) {
+	aggregator := Aggregator{}
+	first := groupForAggregation("first", "first", contracts.DiagnosticDispositionAdvisory, contracts.DiagnosticCategoryDataQuality, "first")
+	first.OccurrenceCount = maximumInt()
+	first.OmittedSampleCount = maximumInt() - 1
+	if err := aggregator.Add(first); err != nil {
+		t.Fatal(err)
+	}
+	second := groupForAggregation("second", "second", contracts.DiagnosticDispositionAdvisory, contracts.DiagnosticCategoryDataQuality, "second")
+	if err := aggregator.Add(second); err == nil {
+		t.Fatal("Add() occurrence total overflow error = nil")
+	}
+	collection := aggregator.Collection()
+	if len(collection.Groups) != 1 || collection.Groups[0].OccurrenceCount != maximumInt() {
+		t.Fatalf("collection changed after rejected overflow = %#v", collection)
+	}
+}
+
+func groupForAggregation(scope string, message string, disposition contracts.DiagnosticDisposition, category contracts.DiagnosticCategory, module string) contracts.DiagnosticGroup {
+	return contracts.DiagnosticGroup{
+		Disposition:     disposition,
+		Category:        category,
+		ReasonCode:      "source_unrelated",
+		Origin:          contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageExtract, StepID: "extract", LaneID: "spells", ModuleKey: module},
+		OccurrenceCount: 1,
+		Samples:         []contracts.DiagnosticSample{{Scope: scope, Message: message}},
+	}
+}

internal/framework/diagnostics/collector.go

@@ -0,0 +1,93 @@
+// Package diagnostics provides bounded local grouping for producer and
+// validator diagnostic results.
+package diagnostics
+
+import (
+	"errors"
+	"fmt"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+)
+
+// Collector merges local producer diagnostics by their semantic identity. Its
+// zero value is ready for use.
+type Collector struct {
+	diagnostics []contracts.ProducerDiagnostic
+	indices     map[key]int
+}
+
+// NewCollector returns an empty local diagnostic collector.
+func NewCollector() *Collector {
+	return &Collector{}
+}
+
+// Add validates and merges one producer diagnostic. Every occurrence remains
+// counted, while the first three distinct samples in input order are retained.
+func (collector *Collector) Add(diagnostic contracts.ProducerDiagnostic) error {
+	if err := diagnostic.Validate(); err != nil {
+		return fmt.Errorf("producer diagnostic: %w", err)
+	}
+	if collector.indices == nil {
+		collector.indices = make(map[key]int)
+	}
+	diagnosticKey := key{disposition: diagnostic.Disposition, category: diagnostic.Category, reasonCode: diagnostic.ReasonCode}
+	index, exists := collector.indices[diagnosticKey]
+	if !exists {
+		if len(collector.diagnostics) >= contracts.MaxProducerDiagnosticGroups {
+			return errors.New("producer diagnostics exceed maximum group count")
+		}
+		collector.indices[diagnosticKey] = len(collector.diagnostics)
+		collector.diagnostics = append(collector.diagnostics, contracts.CloneProducerDiagnostics([]contracts.ProducerDiagnostic{diagnostic})[0])
+		return nil
+	}
+
+	current := &collector.diagnostics[index]
+	if diagnostic.OccurrenceCount > maximumInt()-current.OccurrenceCount {
+		return errors.New("producer diagnostic occurrence count overflow")
+	}
+	current.OccurrenceCount += diagnostic.OccurrenceCount
+	for _, sample := range diagnostic.Samples {
+		if len(current.Samples) == contracts.MaxDiagnosticSamples || containsSample(current.Samples, sample) {
+			continue
+		}
+		current.Samples = append(current.Samples, cloneSample(sample))
+	}
+	current.OmittedSampleCount = current.OccurrenceCount - len(current.Samples)
+	return nil
+}
+
+// Diagnostics returns an independently owned snapshot in first-occurrence
+// order.
+func (collector *Collector) Diagnostics() []contracts.ProducerDiagnostic {
+	if collector == nil {
+		return nil
+	}
+	return contracts.CloneProducerDiagnostics(collector.diagnostics)
+}
+
+func containsSample(samples []contracts.DiagnosticSample, candidate contracts.DiagnosticSample) bool {
+	for _, sample := range samples {
+		if sample.Scope == candidate.Scope && sample.Message == candidate.Message {
+			return true
+		}
+	}
+	return false
+}
+
+func cloneSample(sample contracts.DiagnosticSample) contracts.DiagnosticSample {
+	if sample.ChunkIndex != nil {
+		chunkIndex := *sample.ChunkIndex
+		sample.ChunkIndex = &chunkIndex
+	}
+	return sample
+}
+
+func maximumInt() int {
+	return int(^uint(0) >> 1)
+}
+
+type key struct {
+	disposition contracts.DiagnosticDisposition
+	category    contracts.DiagnosticCategory
+	reasonCode  string
+}

internal/framework/diagnostics/collector_test.go

@@ -0,0 +1,103 @@
+package diagnostics
+
+import (
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+)
+
+func TestCollectorCountsOccurrencesAndRetainsDistinctSamplesInOrder(t *testing.T) {
+	collector := NewCollector()
+	for _, diagnostic := range []contracts.ProducerDiagnostic{
+		advisory("one", "first"),
+		advisory("one", "first"),
+		advisory("two", "second"),
+		advisory("three", "third"),
+		advisory("four", "fourth"),
+	} {
+		if err := collector.Add(diagnostic); err != nil {
+			t.Fatalf("Add() error = %v", err)
+		}
+	}
+
+	diagnostics := collector.Diagnostics()
+	if len(diagnostics) != 1 {
+		t.Fatalf("group count = %d, want 1", len(diagnostics))
+	}
+	group := diagnostics[0]
+	if group.OccurrenceCount != 5 || group.OmittedSampleCount != 2 {
+		t.Fatalf("group counts = %#v, want five occurrences and two omitted samples", group)
+	}
+	if got := []string{group.Samples[0].Scope, group.Samples[1].Scope, group.Samples[2].Scope}; !equalStrings(got, []string{"one", "two", "three"}) {
+		t.Fatalf("sample order = %#v, want first three distinct samples", got)
+	}
+}
+
+func TestCollectorSeparatesGroupsAndRejectsInvalidOrExcessiveGroups(t *testing.T) {
+	collector := NewCollector()
+	if err := collector.Add(advisory("one", "first")); err != nil {
+		t.Fatalf("Add() error = %v", err)
+	}
+	warning := advisory("two", "second")
+	warning.Disposition = contracts.DiagnosticDispositionWarning
+	warning.Category = contracts.DiagnosticCategoryFallback
+	if err := collector.Add(warning); err != nil {
+		t.Fatalf("Add() error = %v", err)
+	}
+	if got := len(collector.Diagnostics()); got != 2 {
+		t.Fatalf("group count = %d, want 2", got)
+	}
+
+	invalid := advisory("bad", "bad")
+	invalid.ReasonCode = ""
+	if err := collector.Add(invalid); err == nil {
+		t.Fatal("Add() error = nil, want invalid diagnostic error")
+	}
+
+	limited := NewCollector()
+	for index := 0; index < contracts.MaxProducerDiagnosticGroups; index++ {
+		diagnostic := advisory("scope", "message")
+		diagnostic.ReasonCode = "reason-" + string(rune('a'+index))
+		if err := limited.Add(diagnostic); err != nil {
+			t.Fatalf("Add(%d) error = %v", index, err)
+		}
+	}
+	if err := limited.Add(advisory("overflow", "overflow")); err == nil {
+		t.Fatal("Add() error = nil, want local group limit error")
+	}
+}
+
+func TestCollectorReturnsIndependentSnapshots(t *testing.T) {
+	collector := NewCollector()
+	if err := collector.Add(advisory("scope", "message")); err != nil {
+		t.Fatalf("Add() error = %v", err)
+	}
+	first := collector.Diagnostics()
+	first[0].Samples[0].Message = "changed"
+	second := collector.Diagnostics()
+	if second[0].Samples[0].Message != "message" {
+		t.Fatalf("collector snapshot changed = %#v", second)
+	}
+}
+
+func advisory(scope string, message string) contracts.ProducerDiagnostic {
+	return contracts.ProducerDiagnostic{
+		Disposition:     contracts.DiagnosticDispositionAdvisory,
+		Category:        contracts.DiagnosticCategoryDataQuality,
+		ReasonCode:      "source_unrelated",
+		OccurrenceCount: 1,
+		Samples:         []contracts.DiagnosticSample{{Scope: scope, Message: message}},
+	}
+}
+
+func equalStrings(left []string, right []string) bool {
+	if len(left) != len(right) {
+		return false
+	}
+	for index := range left {
+		if left[index] != right[index] {
+			return false
+		}
+	}
+	return true
+}

internal/framework/pipeline/checkpoint.go

@@ -28,14 +28,14 @@ type CheckpointRecorder interface {
 	SourceSucceeded(moduleKey string, doc *source.SourceDocument) error
 	SourceFailed(moduleKey string, err error) error
 	ExtractRunning(laneID string, moduleKey string, dependencies []CheckpointFingerprint) error
-	ExtractSucceeded(laneID string, moduleKey string, dependencies []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput, warnings []contracts.Warning) error
+	ExtractSucceeded(laneID string, moduleKey string, dependencies []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput) error
 	ExtractFailed(laneID string, moduleKey string, dependencies []CheckpointFingerprint, err error) error
 	MergeRunning(laneID string, moduleKey string, dependencies []CheckpointFingerprint) error
-	MergeSucceeded(laneID string, moduleKey string, dependencies []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error
+	MergeSucceeded(laneID string, moduleKey string, dependencies []CheckpointFingerprint, output CheckpointArtifact) error
 	MergeRejected(laneID string, moduleKey string, dependencies []CheckpointFingerprint, rejected contracts.RejectedOutput) error
 	MergeFailed(laneID string, moduleKey string, dependencies []CheckpointFingerprint, err error) error
 	NormalizeRunning(laneID string, moduleKey string, dependencies []CheckpointFingerprint) error
-	NormalizeSucceeded(laneID string, moduleKey string, dependencies []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error
+	NormalizeSucceeded(laneID string, moduleKey string, dependencies []CheckpointFingerprint, output CheckpointArtifact) error
 	NormalizeRejected(laneID string, moduleKey string, dependencies []CheckpointFingerprint, rejected contracts.RejectedOutput) error
 	NormalizeFailed(laneID string, moduleKey string, dependencies []CheckpointFingerprint, err error) error
 }
@@ -45,14 +45,14 @@ type CheckpointRecorder interface {
 // remain available for callers that do not have step context.
 type StepCheckpointRecorder interface {
 	ExtractRunningForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint) error
-	ExtractSucceededForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput, warnings []contracts.Warning) error
+	ExtractSucceededForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput) error
 	ExtractFailedForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, err error) error
 	MergeRunningForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint) error
-	MergeSucceededForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error
+	MergeSucceededForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, output CheckpointArtifact) error
 	MergeRejectedForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, rejected contracts.RejectedOutput) error
 	MergeFailedForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, err error) error
 	NormalizeRunningForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint) error
-	NormalizeSucceededForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error
+	NormalizeSucceededForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, output CheckpointArtifact) error
 	NormalizeRejectedForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, rejected contracts.RejectedOutput) error
 	NormalizeFailedForStep(stepID, laneID string, moduleKey string, dependencies []CheckpointFingerprint, err error) error
 }
@@ -279,21 +279,27 @@ type CheckpointArtifact struct {
 	ChunkRef     source.SourceRef
 	Artifact     contracts.SerializedArtifact
 	SchemaDigest string
+	Diagnostics  []CheckpointDiagnostic
+}
+
+// CheckpointDiagnostic stores a producer-local diagnostic alongside a
+// reusable artifact. The runner supplies the current run's origin when it
+// promotes this value into a diagnostic group.
+type CheckpointDiagnostic struct {
+	Diagnostic   contracts.ProducerDiagnostic `json:"diagnostic"`
+	ValidatorKey string                       `json:"validator_key,omitempty"`
 }
 
 type ExtractCheckpoint struct {
 	Outputs  []CheckpointArtifact
 	Rejected []contracts.RejectedOutput
-	Warnings []contracts.Warning
 }
 
 type MergeCheckpoint struct {
-	Output   CheckpointArtifact
-	Warnings []contracts.Warning
+	Output CheckpointArtifact
 }
 type NormalizeCheckpoint struct {
-	Output   CheckpointArtifact
-	Warnings []contracts.Warning
+	Output CheckpointArtifact
 }
 
 type CheckpointLoader interface {
@@ -326,14 +332,14 @@ func (noopCheckpointRecorder) SourceFailed(string, error) error
 func (noopCheckpointRecorder) ExtractRunning(string, string, []CheckpointFingerprint) error {
 	return nil
 }
-func (noopCheckpointRecorder) ExtractSucceeded(string, string, []CheckpointFingerprint, []CheckpointArtifact, []contracts.RejectedOutput, []contracts.Warning) error {
+func (noopCheckpointRecorder) ExtractSucceeded(string, string, []CheckpointFingerprint, []CheckpointArtifact, []contracts.RejectedOutput) error {
 	return nil
 }
 func (noopCheckpointRecorder) ExtractFailed(string, string, []CheckpointFingerprint, error) error {
 	return nil
 }
 func (noopCheckpointRecorder) MergeRunning(string, string, []CheckpointFingerprint) error { return nil }
-func (noopCheckpointRecorder) MergeSucceeded(string, string, []CheckpointFingerprint, CheckpointArtifact, []contracts.Warning) error {
+func (noopCheckpointRecorder) MergeSucceeded(string, string, []CheckpointFingerprint, CheckpointArtifact) error {
 	return nil
 }
 func (noopCheckpointRecorder) MergeRejected(string, string, []CheckpointFingerprint, contracts.RejectedOutput) error {
@@ -345,7 +351,7 @@ func (noopCheckpointRecorder) MergeFailed(string, string, []CheckpointFingerprin
 func (noopCheckpointRecorder) NormalizeRunning(string, string, []CheckpointFingerprint) error {
 	return nil
 }
-func (noopCheckpointRecorder) NormalizeSucceeded(string, string, []CheckpointFingerprint, CheckpointArtifact, []contracts.Warning) error {
+func (noopCheckpointRecorder) NormalizeSucceeded(string, string, []CheckpointFingerprint, CheckpointArtifact) error {
 	return nil
 }
 func (noopCheckpointRecorder) NormalizeRejected(string, string, []CheckpointFingerprint, contracts.RejectedOutput) error {
@@ -378,11 +384,11 @@ func checkpointExtractRunning(recorder CheckpointRecorder, stepID, laneID, modul
 	}
 	return recorder.ExtractRunning(laneID, moduleKey, deps)
 }
-func checkpointExtractSucceeded(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput, warnings []contracts.Warning) error {
+func checkpointExtractSucceeded(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput) error {
 	if stepAware, ok := recorder.(StepCheckpointRecorder); ok {
-		return stepAware.ExtractSucceededForStep(stepID, laneID, moduleKey, deps, outputs, rejected, warnings)
+		return stepAware.ExtractSucceededForStep(stepID, laneID, moduleKey, deps, outputs, rejected)
 	}
-	return recorder.ExtractSucceeded(laneID, moduleKey, deps, outputs, rejected, warnings)
+	return recorder.ExtractSucceeded(laneID, moduleKey, deps, outputs, rejected)
 }
 func checkpointExtractFailed(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, err error) error {
 	if stepAware, ok := recorder.(StepCheckpointRecorder); ok {
@@ -396,11 +402,11 @@ func checkpointMergeRunning(recorder CheckpointRecorder, stepID, laneID, moduleK
 	}
 	return recorder.MergeRunning(laneID, moduleKey, deps)
 }
-func checkpointMergeSucceeded(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error {
+func checkpointMergeSucceeded(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, output CheckpointArtifact) error {
 	if stepAware, ok := recorder.(StepCheckpointRecorder); ok {
-		return stepAware.MergeSucceededForStep(stepID, laneID, moduleKey, deps, output, warnings)
+		return stepAware.MergeSucceededForStep(stepID, laneID, moduleKey, deps, output)
 	}
-	return recorder.MergeSucceeded(laneID, moduleKey, deps, output, warnings)
+	return recorder.MergeSucceeded(laneID, moduleKey, deps, output)
 }
 func checkpointMergeRejected(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, rejected contracts.RejectedOutput) error {
 	if stepAware, ok := recorder.(StepCheckpointRecorder); ok {
@@ -420,11 +426,11 @@ func checkpointNormalizeRunning(recorder CheckpointRecorder, stepID, laneID, mod
 	}
 	return recorder.NormalizeRunning(laneID, moduleKey, deps)
 }
-func checkpointNormalizeSucceeded(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error {
+func checkpointNormalizeSucceeded(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, output CheckpointArtifact) error {
 	if stepAware, ok := recorder.(StepCheckpointRecorder); ok {
-		return stepAware.NormalizeSucceededForStep(stepID, laneID, moduleKey, deps, output, warnings)
+		return stepAware.NormalizeSucceededForStep(stepID, laneID, moduleKey, deps, output)
 	}
-	return recorder.NormalizeSucceeded(laneID, moduleKey, deps, output, warnings)
+	return recorder.NormalizeSucceeded(laneID, moduleKey, deps, output)
 }
 func checkpointNormalizeRejected(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, rejected contracts.RejectedOutput) error {
 	if stepAware, ok := recorder.(StepCheckpointRecorder); ok {

internal/framework/pipeline/chunk_plan_store.go

@@ -8,7 +8,7 @@ import (
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
 )
 
-const ChunkPlanSchemaVersion = "notarius.chunk-plan.v2"
+const ChunkPlanSchemaVersion = "notarius.chunk-plan.v3"
 
 type ChunkPlanProducer struct {
 	InputModule string                          `json:"input_module"`
@@ -19,13 +19,13 @@ type ChunkPlanProducer struct {
 }
 
 type ChunkPlanRecord struct {
-	SchemaVersion string              `json:"schema_version"`
-	SourceDigest  string              `json:"source_digest"`
-	PlanDigest    string              `json:"plan_digest"`
-	Plan          source.ChunkPlan    `json:"plan"`
-	Producer      ChunkPlanProducer   `json:"producer"`
-	Warnings      []contracts.Warning `json:"warnings,omitempty"`
-	CreatedAt     time.Time           `json:"created_at"`
+	SchemaVersion string                         `json:"schema_version"`
+	SourceDigest  string                         `json:"source_digest"`
+	PlanDigest    string                         `json:"plan_digest"`
+	Plan          source.ChunkPlan               `json:"plan"`
+	Producer      ChunkPlanProducer              `json:"producer"`
+	Diagnostics   []contracts.ProducerDiagnostic `json:"diagnostics,omitempty"`
+	CreatedAt     time.Time                      `json:"created_at"`
 }
 
 type ChunkPlanStore interface {

internal/framework/pipeline/debug.go

@@ -52,11 +52,10 @@ type debugTimedEnvelope struct {
 }
 
 type debugBinaryEnvelope struct {
-	ContentBase64 string              `json:"content_base64,omitempty"`
-	ContentDigest string              `json:"content_digest,omitempty"`
-	MediaType     string              `json:"media_type,omitempty"`
-	Metadata      map[string]any      `json:"metadata,omitempty"`
-	Warnings      []contracts.Warning `json:"warnings,omitempty"`
+	ContentBase64 string         `json:"content_base64,omitempty"`
+	ContentDigest string         `json:"content_digest,omitempty"`
+	MediaType     string         `json:"media_type,omitempty"`
+	Metadata      map[string]any `json:"metadata,omitempty"`
 }
 
 type debugSourceInput struct {
@@ -483,14 +482,13 @@ func writeProducerTerminalDebug(recorder DebugRecorder, name string, terminal pr
 	})
 }
 
-func debugContentEnvelope(content []byte, mediaType string, metadata map[string]any, warnings []contracts.Warning) debugBinaryEnvelope {
+func debugContentEnvelope(content []byte, mediaType string, metadata map[string]any, _ any) debugBinaryEnvelope {
 	content = redactSecretBytes(content)
 	return debugBinaryEnvelope{
 		ContentBase64: base64.StdEncoding.EncodeToString(content),
 		ContentDigest: debugContentDigest(content),
 		MediaType:     mediaType,
 		Metadata:      redactSensitiveMap(metadata),
-		Warnings:      cloneWarnings(warnings),
 	}
 }
 
@@ -731,20 +729,9 @@ func debugValidationResultEnvelope(result contracts.ValidationResult) contracts.
 	result.Message = string(redactSecretBytes([]byte(result.Message)))
 	result.CorrectionGuidance = ""
 	result.DiagnosticArtifactPath = string(redactSecretBytes([]byte(result.DiagnosticArtifactPath)))
-	for i := range result.Warnings {
-		result.Warnings[i].Message = string(redactSecretBytes([]byte(result.Warnings[i].Message)))
-	}
 	return result
 }
 
-func debugWarningEnvelopes(warnings []contracts.Warning) []contracts.Warning {
-	out := cloneWarnings(warnings)
-	for i := range out {
-		out[i].Message = string(redactSecretBytes([]byte(out[i].Message)))
-	}
-	return out
-}
-
 func debugRejectedOutputEnvelope(rejected contracts.RejectedOutput) contracts.RejectedOutput {
 	rejected.Message = string(redactSecretBytes([]byte(rejected.Message)))
 	rejected.DiagnosticArtifactPath = string(redactSecretBytes([]byte(rejected.DiagnosticArtifactPath)))

internal/framework/pipeline/diagnostics.go

@@ -0,0 +1,124 @@
+package pipeline
+
+import (
+	"fmt"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	frameworkdiagnostics "gitea.maximumdirect.net/eric/notarius/internal/framework/diagnostics"
+)
+
+func terminalDiagnosticGroups(terminal producerAttemptTerminal, origin contracts.DiagnosticOrigin, chunk *source.Chunk) ([]contracts.DiagnosticGroup, error) {
+	return promoteCheckpointDiagnostics(terminalCheckpointDiagnostics(terminal), origin, chunk)
+}
+
+func terminalCheckpointDiagnostics(terminal producerAttemptTerminal) []CheckpointDiagnostic {
+	diagnostics := make([]CheckpointDiagnostic, 0, len(terminal.Diagnostics))
+	for _, diagnostic := range terminal.Diagnostics {
+		diagnostics = append(diagnostics, CheckpointDiagnostic{Diagnostic: diagnostic})
+	}
+	for _, record := range terminal.Validation.Diagnostics() {
+		diagnostics = append(diagnostics, CheckpointDiagnostic{Diagnostic: record.diagnostic, ValidatorKey: record.validatorName})
+	}
+	if terminal.Action == producerTerminalIncompleteAccepted {
+		for _, record := range incompleteValidationDiagnostics(terminal.Validation) {
+			diagnostics = append(diagnostics, CheckpointDiagnostic{Diagnostic: record.diagnostic, ValidatorKey: record.validatorName})
+		}
+	}
+	return cloneCheckpointDiagnostics(diagnostics)
+}
+
+func promoteCheckpointDiagnostics(diagnostics []CheckpointDiagnostic, origin contracts.DiagnosticOrigin, chunk *source.Chunk) ([]contracts.DiagnosticGroup, error) {
+	groups := make([]contracts.DiagnosticGroup, 0, len(diagnostics))
+	for _, record := range diagnostics {
+		validatorOrigin := origin
+		validatorOrigin.ValidatorKey = record.ValidatorKey
+		promoted, err := promoteProducerDiagnostics([]contracts.ProducerDiagnostic{record.Diagnostic}, validatorOrigin, chunk)
+		if err != nil {
+			return nil, fmt.Errorf("checkpoint diagnostic: %w", err)
+		}
+		groups = append(groups, promoted...)
+	}
+	return groups, nil
+}
+
+func cloneCheckpointDiagnostics(diagnostics []CheckpointDiagnostic) []CheckpointDiagnostic {
+	if len(diagnostics) == 0 {
+		return nil
+	}
+	cloned := make([]CheckpointDiagnostic, len(diagnostics))
+	for index, diagnostic := range diagnostics {
+		cloned[index] = CheckpointDiagnostic{
+			Diagnostic:   contracts.CloneProducerDiagnostics([]contracts.ProducerDiagnostic{diagnostic.Diagnostic})[0],
+			ValidatorKey: diagnostic.ValidatorKey,
+		}
+	}
+	return cloned
+}
+
+func promoteProducerDiagnostics(diagnostics []contracts.ProducerDiagnostic, origin contracts.DiagnosticOrigin, chunk *source.Chunk) ([]contracts.DiagnosticGroup, error) {
+	if err := contracts.ValidateProducerDiagnostics(diagnostics); err != nil {
+		return nil, err
+	}
+	if len(diagnostics) == 0 {
+		return nil, nil
+	}
+	groups := make([]contracts.DiagnosticGroup, len(diagnostics))
+	for index, diagnostic := range diagnostics {
+		group := contracts.DiagnosticGroup{
+			Disposition:        diagnostic.Disposition,
+			Category:           diagnostic.Category,
+			ReasonCode:         diagnostic.ReasonCode,
+			Origin:             origin,
+			OccurrenceCount:    diagnostic.OccurrenceCount,
+			Samples:            cloneDiagnosticSamples(diagnostic.Samples, chunk),
+			OmittedSampleCount: diagnostic.OmittedSampleCount,
+		}
+		if err := group.Validate(); err != nil {
+			return nil, err
+		}
+		groups[index] = group
+	}
+	return groups, nil
+}
+
+func cloneDiagnosticSamples(samples []contracts.DiagnosticSample, chunk *source.Chunk) []contracts.DiagnosticSample {
+	if len(samples) == 0 {
+		return nil
+	}
+	cloned := make([]contracts.DiagnosticSample, len(samples))
+	for index, sample := range samples {
+		if chunk != nil {
+			chunkIndex := chunk.Index
+			sample.ChunkID = chunk.ID
+			sample.ChunkIndex = &chunkIndex
+		} else if sample.ChunkIndex != nil {
+			chunkIndex := *sample.ChunkIndex
+			sample.ChunkIndex = &chunkIndex
+		}
+		cloned[index] = sample
+	}
+	return cloned
+}
+
+func appendDiagnosticGroups(output *RunOutput, groups []contracts.DiagnosticGroup) {
+	if output == nil || len(groups) == 0 {
+		return
+	}
+	cloned := contracts.CloneDiagnosticCollection(contracts.DiagnosticCollection{Groups: groups}).Groups
+	output.diagnosticGroups = append(output.diagnosticGroups, cloned...)
+}
+
+func finalizeDiagnostics(output *RunOutput) error {
+	if output == nil {
+		return nil
+	}
+	aggregator := frameworkdiagnostics.Aggregator{}
+	for _, group := range output.diagnosticGroups {
+		if err := aggregator.Add(group); err != nil {
+			return err
+		}
+	}
+	output.Diagnostics = aggregator.Collection()
+	return nil
+}

internal/framework/pipeline/diagnostics_test.go

@@ -0,0 +1,75 @@
+package pipeline
+
+import (
+	"context"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+)
+
+func TestTerminalDiagnosticGroupsDiscardSupersededAttemptDiagnostics(t *testing.T) {
+	terminal, err := runProducerAttempts(context.Background(), producerAttemptConfig{Retries: 1, Policy: DefaultValidationPolicy()}, func(_ context.Context, request producerAttemptRequest) (producerAttemptOutput, error) {
+		if request.Number == 1 {
+			return producerAttemptOutput{Value: "first", Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("discarded", "discarded")}, Candidate: attemptCandidate(t, "first")}, nil
+		}
+		return producerAttemptOutput{Value: "second", Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("terminal", "terminal")}}, nil
+	}, func(_ context.Context, output producerAttemptOutput) (validationReport, error) {
+		if output.Value == "first" {
+			return validationReport{records: []validationRecord{{validatorName: "validator", outcome: validationRejected, reasonCode: "invalid", correctionGuidance: "Correct it."}}}, nil
+		}
+		return validationReport{}, nil
+	})
+	if err != nil {
+		t.Fatalf("runProducerAttempts() error = %v", err)
+	}
+	groups, err := terminalDiagnosticGroups(terminal, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageMerge, StepID: "step", LaneID: "lane", ModuleKey: "module"}, nil)
+	if err != nil {
+		t.Fatalf("terminalDiagnosticGroups() error = %v", err)
+	}
+	if len(groups) != 1 || groups[0].Samples[0].Scope != "terminal" {
+		t.Fatalf("groups = %#v, want only terminal attempt diagnostics", groups)
+	}
+}
+
+func TestTerminalDiagnosticGroupsPreserveValidatorOrigin(t *testing.T) {
+	terminal := producerAttemptTerminal{Validation: validationReport{records: []validationRecord{{
+		validatorName: "dnd/spells/source-relatedness",
+		outcome:       validationApproved,
+		diagnostics:   []contracts.ProducerDiagnostic{producerDiagnostic("spells[0]", "not found")},
+	}}}}
+	groups, err := terminalDiagnosticGroups(terminal, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageExtract, StepID: "extract", LaneID: "spells", ModuleKey: "dnd/spells"}, nil)
+	if err != nil {
+		t.Fatalf("terminalDiagnosticGroups() error = %v", err)
+	}
+	if len(groups) != 1 || groups[0].Origin.ValidatorKey != "dnd/spells/source-relatedness" {
+		t.Fatalf("groups = %#v, want validator origin", groups)
+	}
+}
+
+func TestAppendDiagnosticGroupsRetainsOnlyStructuredDiagnostics(t *testing.T) {
+	output := RunOutput{}
+	appendDiagnosticGroups(&output, []contracts.DiagnosticGroup{{
+		Disposition:     contracts.DiagnosticDispositionAdvisory,
+		Category:        contracts.DiagnosticCategoryDataQuality,
+		ReasonCode:      "source_unrelated",
+		Origin:          contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageExtract, StepID: "extract", LaneID: "spells", ModuleKey: "dnd/spells"},
+		OccurrenceCount: 2,
+		Samples:         []contracts.DiagnosticSample{{Scope: "spells[0]", Message: "not found"}, {Scope: "spells[1]", Message: "also not found"}},
+	}})
+	if err := finalizeDiagnostics(&output); err != nil {
+		t.Fatalf("finalizeDiagnostics() error = %v", err)
+	}
+	if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].OccurrenceCount != 2 {
+		t.Fatalf("diagnostics = %#v, want grouped collection", output.Diagnostics)
+	}
+}
+
+func producerDiagnostic(scope string, message string) contracts.ProducerDiagnostic {
+	return contracts.ProducerDiagnostic{
+		Disposition:     contracts.DiagnosticDispositionAdvisory,
+		Category:        contracts.DiagnosticCategoryDataQuality,
+		ReasonCode:      "source_unrelated",
+		OccurrenceCount: 1,
+		Samples:         []contracts.DiagnosticSample{{Scope: scope, Message: message}},
+	}
+}

internal/framework/pipeline/extractor_registry.go

@@ -71,7 +71,7 @@ func RegisterExtractorBuilder[T any](registry *ExtractorRegistry, spec ModuleSpe
 		if err != nil {
 			return erasedTypedResult{}, fmt.Errorf("clone extraction model candidate: %w", err)
 		}
-		return erasedTypedResult{Value: result.Value, Warnings: cloneWarnings(result.Warnings), ModelCandidate: candidate}, nil
+		return erasedTypedResult{Value: result.Value, Diagnostics: contracts.CloneProducerDiagnostics(result.Diagnostics), ModelCandidate: candidate}, nil
 	}}
 	if registry.typedEntries == nil {
 		registry.typedEntries = map[string]typedExtractorEntry{}

internal/framework/pipeline/merger_registry.go

@@ -97,7 +97,7 @@ func RegisterMergerBuilder[T any](registry *MergerRegistry, spec ModuleSpec, val
 			if err != nil {
 				return erasedTypedResult{}, fmt.Errorf("clone merge model candidate: %w", err)
 			}
-			return erasedTypedResult{Value: result.Value, Warnings: cloneWarnings(result.Warnings), ModelCandidate: candidate}, nil
+			return erasedTypedResult{Value: result.Value, Diagnostics: contracts.CloneProducerDiagnostics(result.Diagnostics), ModelCandidate: candidate}, nil
 		},
 	}
 	return nil

internal/framework/pipeline/normalize_retry.go

@@ -30,5 +30,14 @@ func validateNormalizeRetry(retry *contracts.NormalizeRetry) error {
 	if len(retry.Message) > contracts.MaxNormalizeRetryMessageBytes {
 		return errors.New("normalize retry directive message exceeds maximum length")
 	}
+	if !utf8.ValidString(retry.CorrectionGuidance) {
+		return errors.New("normalize retry directive correction guidance has invalid UTF-8")
+	}
+	if retry.CorrectionGuidance != "" && strings.TrimSpace(retry.CorrectionGuidance) == "" {
+		return errors.New("normalize retry directive correction guidance is blank")
+	}
+	if len(retry.CorrectionGuidance) > contracts.MaxNormalizeRetryCorrectionGuidanceBytes {
+		return errors.New("normalize retry directive correction guidance exceeds maximum length")
+	}
 	return nil
 }

internal/framework/pipeline/normalizer_registry.go

@@ -88,7 +88,7 @@ func RegisterNormalizerBuilder[T any](registry *NormalizerRegistry, spec ModuleS
 			if err != nil {
 				return erasedTypedResult{}, fmt.Errorf("clone normalize model candidate: %w", err)
 			}
-			return erasedTypedResult{Value: result.Value, Warnings: cloneWarnings(result.Warnings), Retry: cloneNormalizeRetry(result.Retry), ModelCandidate: candidate}, nil
+			return erasedTypedResult{Value: result.Value, Diagnostics: contracts.CloneProducerDiagnostics(result.Diagnostics), Retry: cloneNormalizeRetry(result.Retry), ModelCandidate: candidate}, nil
 		},
 	}
 	return nil
@@ -99,9 +99,10 @@ func cloneNormalizeRetry(retry *contracts.NormalizeRetry) *contracts.NormalizeRe
 		return nil
 	}
 	return &contracts.NormalizeRetry{
-		ReasonCode:       retry.ReasonCode,
-		Message:          retry.Message,
-		FallbackWarnings: cloneWarnings(retry.FallbackWarnings),
+		ReasonCode:          retry.ReasonCode,
+		Message:             retry.Message,
+		CorrectionGuidance:  retry.CorrectionGuidance,
+		FallbackDiagnostics: contracts.CloneProducerDiagnostics(retry.FallbackDiagnostics),
 	}
 }
 

internal/framework/pipeline/normalizer_registry_test.go

@@ -8,26 +8,24 @@ import (
 )
 
 type retryingNotesNormalizer struct {
-	warnings []contracts.Warning
-	retry    *contracts.NormalizeRetry
+	retry *contracts.NormalizeRetry
 }
 
 func (retryingNotesNormalizer) Key() string                               { return "test/retry-normalize" }
 func (retryingNotesNormalizer) ReferenceSlots() []contracts.ReferenceSlot { return nil }
 func (n retryingNotesNormalizer) Normalize(_ context.Context, req contracts.TypedNormalizeRequest[codecNotes]) (contracts.TypedNormalizeResult[codecNotes], error) {
-	return contracts.TypedNormalizeResult[codecNotes]{Value: req.MergeOutput.Value, Warnings: n.warnings, Retry: n.retry}, nil
+	return contracts.TypedNormalizeResult[codecNotes]{Value: req.MergeOutput.Value, Retry: n.retry}, nil
 }
 
 func TestNormalizerRegistryErasureClonesRetryDirective(t *testing.T) {
-	warnings := []contracts.Warning{{Scope: "attempt", ReasonCode: "ordinary", Message: "ordinary warning"}}
 	retry := &contracts.NormalizeRetry{
-		ReasonCode:       "retryable",
-		Message:          "safe fallback available",
-		FallbackWarnings: []contracts.Warning{{Scope: "fallback", ReasonCode: "omitted", Message: "fallback warning"}},
+		ReasonCode:         "retryable",
+		Message:            "safe fallback available",
+		CorrectionGuidance: "return a complete corrected proposal",
 	}
 	registry := NewNormalizerRegistry()
 	if err := RegisterNormalizer(registry, ModuleSpec{Key: "test/retry-normalize", Stage: StageNormalize, ExecutionClass: contracts.ExecutionClassDeterministic, ArtifactKind: "test/notes"}, func() (contracts.Normalizer[codecNotes], error) {
-		return retryingNotesNormalizer{warnings: warnings, retry: retry}, nil
+		return retryingNotesNormalizer{retry: retry}, nil
 	}); err != nil {
 		t.Fatalf("RegisterNormalizer() error = %v", err)
 	}
@@ -43,10 +41,9 @@ func TestNormalizerRegistryErasureClonesRetryDirective(t *testing.T) {
 	if err != nil {
 		t.Fatalf("normalize() error = %v", err)
 	}
-	warnings[0].Message = "mutated"
 	retry.Message = "mutated"
-	retry.FallbackWarnings[0].Message = "mutated"
-	if result.Retry == nil || result.Warnings[0].Message != "ordinary warning" || result.Retry.Message != "safe fallback available" || result.Retry.FallbackWarnings[0].Message != "fallback warning" {
-		t.Fatalf("erased retry result = %#v, want independent warning data", result)
+	retry.CorrectionGuidance = "mutated guidance"
+	if result.Retry == nil || result.Retry.Message != "safe fallback available" || result.Retry.CorrectionGuidance != "return a complete corrected proposal" {
+		t.Fatalf("erased retry result = %#v, want independent retry data", result)
 	}
 }

internal/framework/pipeline/producer_attempts.go

@@ -52,24 +52,28 @@ type producerAttemptRequest struct {
 // the current value as a safe fallback if its shared budget is exhausted.
 // Artifact-specific adapters are responsible for validating and populating it.
 type producerRetryDirective struct {
-	FallbackWarnings []contracts.Warning
+	CorrectionGuidance  string
+	FallbackDiagnostics []contracts.ProducerDiagnostic
 }
 
 func (directive *producerRetryDirective) clone() *producerRetryDirective {
 	if directive == nil {
 		return nil
 	}
-	return &producerRetryDirective{FallbackWarnings: cloneWarnings(directive.FallbackWarnings)}
+	return &producerRetryDirective{
+		CorrectionGuidance:  directive.CorrectionGuidance,
+		FallbackDiagnostics: contracts.CloneProducerDiagnostics(directive.FallbackDiagnostics),
+	}
 }
 
 // producerAttemptOutput is intentionally artifact-neutral. Value remains
 // opaque to the state machine; Candidate is the attempt-local response that
 // may support semantic correction.
 type producerAttemptOutput struct {
-	Value     any
-	Candidate *contracts.ModelCandidate
-	Warnings  []contracts.Warning
-	Retry     *producerRetryDirective
+	Value       any
+	Candidate   *contracts.ModelCandidate
+	Diagnostics []contracts.ProducerDiagnostic
+	Retry       *producerRetryDirective
 }
 
 func (output producerAttemptOutput) clone() (producerAttemptOutput, error) {
@@ -78,10 +82,10 @@ func (output producerAttemptOutput) clone() (producerAttemptOutput, error) {
 		return producerAttemptOutput{}, fmt.Errorf("clone model candidate: %w", err)
 	}
 	return producerAttemptOutput{
-		Value:     output.Value,
-		Candidate: candidate,
-		Warnings:  cloneWarnings(output.Warnings),
-		Retry:     output.Retry.clone(),
+		Value:       output.Value,
+		Candidate:   candidate,
+		Diagnostics: contracts.CloneProducerDiagnostics(output.Diagnostics),
+		Retry:       output.Retry.clone(),
 	}, nil
 }
 
@@ -113,7 +117,7 @@ func (provenance producerAttemptProvenance) clone() producerAttemptProvenance {
 type producerAttemptTerminal struct {
 	Action               producerTerminalAction
 	Value                any
-	Warnings             []contracts.Warning
+	Diagnostics          []contracts.ProducerDiagnostic
 	Rejection            *contracts.RejectedOutput
 	Validation           validationReport
 	ValidationIncomplete bool
@@ -121,7 +125,7 @@ type producerAttemptTerminal struct {
 }
 
 func (terminal producerAttemptTerminal) clone() producerAttemptTerminal {
-	terminal.Warnings = cloneWarnings(terminal.Warnings)
+	terminal.Diagnostics = contracts.CloneProducerDiagnostics(terminal.Diagnostics)
 	if terminal.Rejection != nil {
 		rejection := *terminal.Rejection
 		rejection.Validation = cloneValidationSummaryPtr(rejection.Validation)
@@ -193,6 +197,10 @@ func runProducerAttempts(ctx context.Context, config producerAttemptConfig, prod
 			}
 			return failedProducerAttempt(provenance), fmt.Errorf("producer failed after %d attempt(s): %w", number, err)
 		}
+		if err := validateProducerAttemptDiagnostics(output); err != nil {
+			provenance = append(provenance, producerAttemptProvenance{Number: number, Kind: kind, Outcome: producerAttemptFailed})
+			return failedProducerAttempt(provenance), err
+		}
 
 		output, err = output.clone()
 		if err != nil {
@@ -200,12 +208,23 @@ func runProducerAttempts(ctx context.Context, config producerAttemptConfig, prod
 			return failedProducerAttempt(provenance), err
 		}
 		if output.Retry != nil && number < attemptLimit {
+			correction, err = moduleRetryCorrection(output)
+			if err != nil {
+				provenance = append(provenance, producerAttemptProvenance{Number: number, Kind: kind, Outcome: producerAttemptFailed})
+				return failedProducerAttempt(provenance), err
+			}
 			provenance = append(provenance, producerAttemptProvenance{Number: number, Kind: kind, Outcome: producerAttemptRetried})
-			kind, correction = producerAttemptModuleRetry, nil
+			kind = producerAttemptModuleRetry
 			continue
 		}
+		if output.Retry != nil && output.Retry.CorrectionGuidance != "" {
+			if _, err := moduleRetryCorrection(output); err != nil {
+				provenance = append(provenance, producerAttemptProvenance{Number: number, Kind: kind, Outcome: producerAttemptFailed})
+				return failedProducerAttempt(provenance), err
+			}
+		}
 		if output.Retry != nil {
-			output.Warnings = append(output.Warnings, cloneWarnings(output.Retry.FallbackWarnings)...)
+			output.Diagnostics = append(output.Diagnostics, contracts.CloneProducerDiagnostics(output.Retry.FallbackDiagnostics)...)
 		}
 		correctionCandidate, err := contracts.CloneModelCandidate(output.Candidate)
 		if err != nil {
@@ -253,20 +272,47 @@ func runProducerAttempts(ctx context.Context, config producerAttemptConfig, prod
 		if incomplete := firstIncompleteValidation(report); incomplete != nil {
 			provenance = append(provenance, producerAttemptProvenance{Number: number, Kind: kind, Outcome: producerAttemptIncompleteAccepted, Validation: report})
 			if config.Policy.ValidatorFailure == ValidatorFailureWarnContinue {
-				warnings := terminalWarnings(output, report)
-				warnings = append(warnings, incompleteValidationWarnings(report)...)
-				return producerAttemptTerminal{Action: producerTerminalIncompleteAccepted, Value: output.Value, Warnings: warnings, Validation: report, ValidationIncomplete: true, Provenance: cloneProducerAttemptProvenance(provenance)}, nil
+				return producerAttemptTerminal{Action: producerTerminalIncompleteAccepted, Value: output.Value, Diagnostics: cloneProducerDiagnostics(output.Diagnostics), Validation: report, ValidationIncomplete: true, Provenance: cloneProducerAttemptProvenance(provenance)}, nil
 			}
 			return failedProducerAttempt(provenance), validatorFailureError(*incomplete)
 		}
 
 		provenance = append(provenance, producerAttemptProvenance{Number: number, Kind: kind, Outcome: producerAttemptAccepted, Validation: report})
-		return producerAttemptTerminal{Action: producerTerminalAccepted, Value: output.Value, Warnings: terminalWarnings(output, report), Validation: report, Provenance: cloneProducerAttemptProvenance(provenance)}, nil
+		return producerAttemptTerminal{Action: producerTerminalAccepted, Value: output.Value, Diagnostics: cloneProducerDiagnostics(output.Diagnostics), Validation: report, Provenance: cloneProducerAttemptProvenance(provenance)}, nil
 	}
 
 	return failedProducerAttempt(provenance), errors.New("producer attempt budget was not exhausted deterministically")
 }
 
+func moduleRetryCorrection(output producerAttemptOutput) (*contracts.SemanticCorrection, error) {
+	if output.Retry == nil || output.Retry.CorrectionGuidance == "" {
+		return nil, nil
+	}
+	if output.Candidate == nil {
+		return nil, errors.New("module retry correction guidance requires a model candidate")
+	}
+	if output.Candidate.Protocol != contracts.CorrectionProtocolSingleResponseV1 {
+		return nil, fmt.Errorf("module retry correction guidance requires protocol %q", contracts.CorrectionProtocolSingleResponseV1)
+	}
+	correction, err := contracts.NewSemanticCorrection(output.Candidate.Response, output.Retry.CorrectionGuidance)
+	if err != nil {
+		return nil, fmt.Errorf("construct module retry semantic correction: %w", err)
+	}
+	return correction, nil
+}
+
+func validateProducerAttemptDiagnostics(output producerAttemptOutput) error {
+	if err := contracts.ValidateProducerDiagnostics(output.Diagnostics); err != nil {
+		return fmt.Errorf("producer returned invalid diagnostics: %w", err)
+	}
+	if output.Retry != nil {
+		if err := contracts.ValidateProducerDiagnostics(output.Retry.FallbackDiagnostics); err != nil {
+			return fmt.Errorf("producer returned invalid retry fallback diagnostics: %w", err)
+		}
+	}
+	return nil
+}
+
 func isImmediateProducerFailure(err error) bool {
 	if errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded) {
 		return true
@@ -290,7 +336,7 @@ func applySemanticTerminalPolicy(policy ValidationPolicy, provenance []producerA
 	rejected := contracts.RejectedOutput{ValidatorName: rejection.validatorName, ReasonCode: rejection.reasonCode, Message: rejection.message, AttemptCount: number, DiagnosticArtifactPath: rejection.diagnosticPath}
 	switch policy.SemanticRejection {
 	case SemanticRejectionRejectOutput:
-		return producerAttemptTerminal{Action: producerTerminalRejected, Warnings: terminalWarnings(output, report), Rejection: &rejected, Validation: report, ValidationIncomplete: firstIncompleteValidation(report) != nil, Provenance: cloneProducerAttemptProvenance(provenance)}, nil
+		return producerAttemptTerminal{Action: producerTerminalRejected, Diagnostics: cloneProducerDiagnostics(output.Diagnostics), Rejection: &rejected, Validation: report, ValidationIncomplete: firstIncompleteValidation(report) != nil, Provenance: cloneProducerAttemptProvenance(provenance)}, nil
 	case SemanticRejectionFailRun:
 		return failedProducerAttempt(provenance), fmt.Errorf("producer candidate rejected after %d attempt(s): %s", number, rejection.message)
 	default:
@@ -308,28 +354,31 @@ func firstIncompleteValidation(report validationReport) *validationRecord {
 	return nil
 }
 
-func terminalWarnings(output producerAttemptOutput, report validationReport) []contracts.Warning {
-	warnings := cloneWarnings(output.Warnings)
-	warnings = append(warnings, report.Warnings()...)
-	return warnings
+func cloneProducerDiagnostics(diagnostics []contracts.ProducerDiagnostic) []contracts.ProducerDiagnostic {
+	return contracts.CloneProducerDiagnostics(diagnostics)
 }
 
-// incompleteValidationWarnings reports only validators that exhausted their
-// execution budget. It never reports rejected candidates, and it uses fixed
-// text so provider errors and correction content cannot cross this boundary.
-func incompleteValidationWarnings(report validationReport) []contracts.Warning {
-	warnings := make([]contracts.Warning, 0)
+// incompleteValidationDiagnostics reports every applicable validator that
+// could not complete under warn_continue. It uses fixed text so provider
+// errors and arbitrary validator prose cannot cross this boundary.
+func incompleteValidationDiagnostics(report validationReport) []validationDiagnosticRecord {
+	diagnostics := make([]validationDiagnosticRecord, 0)
 	for _, record := range report.records {
-		if record.outcome != validationFailed {
+		if record.outcome != validationFailed && record.outcome != validationSkipped {
 			continue
 		}
-		warnings = append(warnings, contracts.Warning{
-			Scope:      record.validatorName,
-			ReasonCode: "validator_execution_incomplete",
-			Message:    "Validator execution did not complete within its configured budget.",
-		})
+		diagnostics = append(diagnostics, validationDiagnosticRecord{validatorName: record.validatorName, diagnostic: contracts.ProducerDiagnostic{
+			Disposition:     contracts.DiagnosticDispositionWarning,
+			Category:        contracts.DiagnosticCategoryValidationIncomplete,
+			ReasonCode:      "validator_execution_incomplete",
+			OccurrenceCount: 1,
+			Samples: []contracts.DiagnosticSample{{
+				Scope:   record.validatorName,
+				Message: "Validator execution did not complete within its configured budget.",
+			}},
+		}})
 	}
-	return warnings
+	return diagnostics
 }
 
 // validationSummary projects a terminal state-machine result into the durable

internal/framework/pipeline/producer_attempts_test.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"errors"
 	"reflect"
+	"strings"
 	"testing"
 
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
@@ -217,6 +218,9 @@ func TestRunProducerAttemptsUsesModuleRetryBudgetAndFallback(t *testing.T) {
 		calls := 0
 		terminal, err := runProducerAttempts(context.Background(), producerAttemptConfig{Retries: 1, Policy: DefaultValidationPolicy()}, func(_ context.Context, request producerAttemptRequest) (producerAttemptOutput, error) {
 			calls++
+			if request.Correction != nil {
+				t.Fatalf("feedback-free module retry correction = %#v, want nil", request.Correction)
+			}
 			if request.Number == 1 {
 				return producerAttemptOutput{Value: "fallback", Retry: &producerRetryDirective{}}, nil
 			}
@@ -233,28 +237,137 @@ func TestRunProducerAttemptsUsesModuleRetryBudgetAndFallback(t *testing.T) {
 		}
 	})
 
-	t.Run("fallback", func(t *testing.T) {
-		fallbackWarning := contracts.Warning{ReasonCode: "fallback", Message: "fallback warning"}
-		terminal, err := runProducerAttempts(context.Background(), producerAttemptConfig{Policy: DefaultValidationPolicy()}, func(context.Context, producerAttemptRequest) (producerAttemptOutput, error) {
-			return producerAttemptOutput{Value: "fallback", Retry: &producerRetryDirective{FallbackWarnings: []contracts.Warning{fallbackWarning}}}, nil
+	t.Run("feedback retry", func(t *testing.T) {
+		const (
+			defective = `{"duplicate_groups":[{"candidate_ids":[1,99],"canonical_candidate_id":1}]}`
+			guidance  = "Use only candidate IDs from the supplied candidate list. Return one complete corrected response."
+		)
+		var observed *contracts.SemanticCorrection
+		terminal, err := runProducerAttempts(context.Background(), producerAttemptConfig{Retries: 1, Policy: DefaultValidationPolicy()}, func(_ context.Context, request producerAttemptRequest) (producerAttemptOutput, error) {
+			if request.Number == 1 {
+				return producerAttemptOutput{Value: "safe fallback", Candidate: attemptCandidate(t, defective), Retry: &producerRetryDirective{CorrectionGuidance: guidance}}, nil
+			}
+			observed = request.Correction
+			return producerAttemptOutput{Value: "corrected"}, nil
 		}, approveAttempt)
 		if err != nil {
 			t.Fatalf("runProducerAttempts() error = %v", err)
 		}
-		if terminal.Action != producerTerminalAccepted || terminal.Value != "fallback" || !reflect.DeepEqual(terminal.Warnings, []contracts.Warning{fallbackWarning}) {
+		if terminal.Action != producerTerminalAccepted || terminal.Value != "corrected" {
+			t.Fatalf("terminal = %#v, want corrected accepted value", terminal)
+		}
+		if observed == nil || string(observed.AssistantResponse) != defective || observed.UserGuidance != guidance {
+			t.Fatalf("module retry correction = %#v, want exact latest response and guidance", observed)
+		}
+		if got := attemptKinds(terminal.Provenance); !reflect.DeepEqual(got, []producerAttemptKind{producerAttemptInitial, producerAttemptModuleRetry}) {
+			t.Fatalf("attempt kinds = %v", got)
+		}
+	})
+
+	t.Run("feedback-free retry clears prior correction", func(t *testing.T) {
+		terminal, err := runProducerAttempts(context.Background(), producerAttemptConfig{Retries: 2, Policy: DefaultValidationPolicy()}, func(_ context.Context, request producerAttemptRequest) (producerAttemptOutput, error) {
+			switch request.Number {
+			case 1:
+				return producerAttemptOutput{Value: "first fallback", Candidate: attemptCandidate(t, "first defective response"), Retry: &producerRetryDirective{CorrectionGuidance: "Correct the first response."}}, nil
+			case 2:
+				if request.Correction == nil || string(request.Correction.AssistantResponse) != "first defective response" {
+					t.Fatalf("second attempt correction = %#v", request.Correction)
+				}
+				return producerAttemptOutput{Value: "second fallback", Retry: &producerRetryDirective{}}, nil
+			case 3:
+				if request.Correction != nil {
+					t.Fatalf("third attempt retained stale correction %#v", request.Correction)
+				}
+				return producerAttemptOutput{Value: "accepted"}, nil
+			default:
+				t.Fatalf("unexpected producer attempt %d", request.Number)
+				return producerAttemptOutput{}, nil
+			}
+		}, approveAttempt)
+		if err != nil || terminal.Action != producerTerminalAccepted || terminal.Value != "accepted" {
+			t.Fatalf("terminal = %#v, error = %v", terminal, err)
+		}
+	})
+
+	t.Run("fallback", func(t *testing.T) {
+		fallbackDiagnostic := contracts.ProducerDiagnostic{Disposition: contracts.DiagnosticDispositionWarning, Category: contracts.DiagnosticCategoryFallback, ReasonCode: "fallback", OccurrenceCount: 1, Samples: []contracts.DiagnosticSample{{Scope: "fallback", Message: "fallback warning"}}}
+		terminal, err := runProducerAttempts(context.Background(), producerAttemptConfig{Policy: DefaultValidationPolicy()}, func(context.Context, producerAttemptRequest) (producerAttemptOutput, error) {
+			return producerAttemptOutput{Value: "fallback", Retry: &producerRetryDirective{FallbackDiagnostics: []contracts.ProducerDiagnostic{fallbackDiagnostic}}}, nil
+		}, approveAttempt)
+		if err != nil {
+			t.Fatalf("runProducerAttempts() error = %v", err)
+		}
+		if terminal.Action != producerTerminalAccepted || terminal.Value != "fallback" || !reflect.DeepEqual(terminal.Diagnostics, []contracts.ProducerDiagnostic{fallbackDiagnostic}) {
 			t.Fatalf("terminal = %#v", terminal)
 		}
 	})
 }
 
+func TestRunProducerAttemptsRejectsModuleCorrectionWithoutModelCandidate(t *testing.T) {
+	producerCalls := 0
+	validatorCalls := 0
+	terminal, err := runProducerAttempts(context.Background(), producerAttemptConfig{Retries: 1, Policy: DefaultValidationPolicy()}, func(context.Context, producerAttemptRequest) (producerAttemptOutput, error) {
+		producerCalls++
+		return producerAttemptOutput{Value: "safe", Retry: &producerRetryDirective{CorrectionGuidance: "Return a complete corrected response."}}, nil
+	}, func(context.Context, producerAttemptOutput) (validationReport, error) {
+		validatorCalls++
+		return validationReport{}, nil
+	})
+	if err == nil || !strings.Contains(err.Error(), "requires a model candidate") {
+		t.Fatalf("runProducerAttempts() error = %v, want model-candidate contract failure", err)
+	}
+	if terminal.Action != producerTerminalFailed || producerCalls != 1 || validatorCalls != 0 {
+		t.Fatalf("terminal = %#v, producer calls = %d, validator calls = %d", terminal, producerCalls, validatorCalls)
+	}
+}
+
+func TestRunProducerAttemptsRejectsInvalidDiagnosticsWithoutRetry(t *testing.T) {
+	tests := []struct {
+		name   string
+		output producerAttemptOutput
+		want   string
+	}{
+		{
+			name:   "producer diagnostics",
+			output: producerAttemptOutput{Diagnostics: []contracts.ProducerDiagnostic{{}}},
+			want:   "producer returned invalid diagnostics",
+		},
+		{
+			name:   "retry fallback diagnostics",
+			output: producerAttemptOutput{Retry: &producerRetryDirective{FallbackDiagnostics: []contracts.ProducerDiagnostic{{}}}},
+			want:   "producer returned invalid retry fallback diagnostics",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			producerCalls := 0
+			validatorCalls := 0
+			terminal, err := runProducerAttempts(context.Background(), producerAttemptConfig{Retries: 3, Policy: DefaultValidationPolicy()}, func(context.Context, producerAttemptRequest) (producerAttemptOutput, error) {
+				producerCalls++
+				return test.output, nil
+			}, func(context.Context, producerAttemptOutput) (validationReport, error) {
+				validatorCalls++
+				return validationReport{}, nil
+			})
+			if err == nil || !strings.Contains(err.Error(), test.want) {
+				t.Fatalf("runProducerAttempts() error = %v, want %q", err, test.want)
+			}
+			if terminal.Action != producerTerminalFailed || producerCalls != 1 || validatorCalls != 0 {
+				t.Fatalf("terminal = %#v, producer calls = %d, validator calls = %d; want immediate framework failure", terminal, producerCalls, validatorCalls)
+			}
+		})
+	}
+}
+
 func TestRunProducerAttemptsRejectionWinsOverValidatorFailure(t *testing.T) {
-	firstWarnings := []contracts.Warning{{ReasonCode: "discarded", Message: "discarded warning"}}
-	secondWarnings := []contracts.Warning{{ReasonCode: "accepted", Message: "accepted warning"}}
+	firstDiagnostics := []contracts.ProducerDiagnostic{producerDiagnostic("discarded", "discarded warning")}
+	secondDiagnostics := []contracts.ProducerDiagnostic{producerDiagnostic("accepted", "accepted warning")}
 	terminal, err := runProducerAttempts(context.Background(), producerAttemptConfig{Retries: 1, Policy: DefaultValidationPolicy()}, func(_ context.Context, request producerAttemptRequest) (producerAttemptOutput, error) {
 		if request.Number == 1 {
-			return producerAttemptOutput{Value: "first", Candidate: attemptCandidate(t, "defective"), Warnings: firstWarnings}, nil
+			return producerAttemptOutput{Value: "first", Candidate: attemptCandidate(t, "defective"), Diagnostics: firstDiagnostics}, nil
 		}
-		return producerAttemptOutput{Value: "second", Candidate: attemptCandidate(t, "corrected"), Warnings: secondWarnings}, nil
+		return producerAttemptOutput{Value: "second", Candidate: attemptCandidate(t, "corrected"), Diagnostics: secondDiagnostics}, nil
 	}, func(_ context.Context, output producerAttemptOutput) (validationReport, error) {
 		if output.Value == "first" {
 			report := rejectedAttemptReport("defect")
@@ -272,8 +385,8 @@ func TestRunProducerAttemptsRejectionWinsOverValidatorFailure(t *testing.T) {
 	if got := attemptKinds(terminal.Provenance); !reflect.DeepEqual(got, []producerAttemptKind{producerAttemptInitial, producerAttemptSemanticRetry}) {
 		t.Fatalf("attempt kinds = %v", got)
 	}
-	if got := terminal.Warnings; !reflect.DeepEqual(got, secondWarnings) {
-		t.Fatalf("terminal warnings = %#v, want %#v", got, secondWarnings)
+	if got := terminal.Diagnostics; !reflect.DeepEqual(got, secondDiagnostics) {
+		t.Fatalf("terminal diagnostics = %#v, want %#v", got, secondDiagnostics)
 	}
 	if len(terminal.Provenance[0].Validation.records) != 2 {
 		t.Fatalf("first validation records = %#v, want rejection and failure", terminal.Provenance[0].Validation.records)
@@ -302,9 +415,6 @@ func TestValidationSummaryIsBoundedAndCorrectedSuccessIsQuiet(t *testing.T) {
 	if len(summary.ReasonCodes) != 0 || len(summary.RejectingValidators) != 0 || len(summary.IncompleteValidators) != 0 {
 		t.Fatalf("corrected success summary retained prior findings: %#v", summary)
 	}
-	if len(terminal.Warnings) != 0 {
-		t.Fatalf("corrected success warnings = %#v, want none", terminal.Warnings)
-	}
 }
 
 func TestWarnContinueRecordsOneWarningForEachExhaustedValidator(t *testing.T) {
@@ -324,11 +434,17 @@ func TestWarnContinueRecordsOneWarningForEachExhaustedValidator(t *testing.T) {
 	if terminal.Action != producerTerminalIncompleteAccepted {
 		t.Fatalf("terminal action = %q", terminal.Action)
 	}
-	if got, want := terminal.Warnings, []contracts.Warning{
-		{Scope: "first", ReasonCode: "validator_execution_incomplete", Message: "Validator execution did not complete within its configured budget."},
-		{Scope: "third", ReasonCode: "validator_execution_incomplete", Message: "Validator execution did not complete within its configured budget."},
-	}; !reflect.DeepEqual(got, want) {
-		t.Fatalf("warnings = %#v, want %#v", got, want)
+	groups, groupErr := terminalDiagnosticGroups(terminal, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageNormalize, StepID: "step", LaneID: "lane", ModuleKey: "module"}, nil)
+	if groupErr != nil {
+		t.Fatalf("terminalDiagnosticGroups() error = %v", groupErr)
+	}
+	if len(groups) != 3 {
+		t.Fatalf("diagnostic groups = %#v, want every failed and skipped validator", groups)
+	}
+	for index, validator := range []string{"first", "second", "third"} {
+		if group := groups[index]; group.Origin.ValidatorKey != validator || group.Disposition != contracts.DiagnosticDispositionWarning || group.Category != contracts.DiagnosticCategoryValidationIncomplete || group.ReasonCode != "validator_execution_incomplete" {
+			t.Fatalf("diagnostic group %d = %#v, want incomplete warning for %q", index, group, validator)
+		}
 	}
 	summary := validationSummary(terminal, StageNormalize, "step", "lane", "module", "", 0)
 	if summary.Status != "incomplete" || !reflect.DeepEqual(summary.IncompleteValidators, []string{"first", "second", "third"}) || !reflect.DeepEqual(summary.ReasonCodes, []string{"missing_prerequisite"}) {

internal/framework/pipeline/references.go

@@ -30,45 +30,45 @@ type ReferenceMaterializationOptions struct {
 	WorkingDir string
 }
 
-func MaterializeReferences(resolved ResolvedPipeline, catalog ModuleCatalog, options ReferenceMaterializationOptions) (ResolvedPipeline, []contracts.Warning, error) {
+func MaterializeReferences(resolved ResolvedPipeline, catalog ModuleCatalog, options ReferenceMaterializationOptions) (ResolvedPipeline, []contracts.ProducerDiagnostic, error) {
 	out := resolved
 	out.ChunkReferences = CloneReferenceTarget(resolved.ChunkReferences)
-	chunkReferenceSet, chunkWarnings, err := materializeReferenceTarget(resolved.ID, resolved.ChunkReferences, "", catalog, options)
+	chunkReferenceSet, chunkDiagnostics, err := materializeReferenceTarget(resolved.ID, resolved.ChunkReferences, "", catalog, options)
 	if err != nil {
 		return ResolvedPipeline{}, nil, err
 	}
 	out.ChunkReferences.ReferenceSet = chunkReferenceSet
-	warnings := append([]contracts.Warning(nil), chunkWarnings...)
+	diagnostics := contracts.CloneProducerDiagnostics(chunkDiagnostics)
 	if len(resolved.Steps) == 0 {
-		return out, warnings, nil
+		return out, diagnostics, nil
 	}
-	materializeLane := func(lane ResolvedArtifactLane) (ResolvedArtifactLane, []contracts.Warning, error) {
+	materializeLane := func(lane ResolvedArtifactLane) (ResolvedArtifactLane, []contracts.ProducerDiagnostic, error) {
 		materializedLane := lane
-		var allWarnings []contracts.Warning
+		var allDiagnostics []contracts.ProducerDiagnostic
 		materializedLane.ExtractReferences = CloneReferenceTarget(lane.ExtractReferences)
 		materializedLane.MergeReferences = CloneReferenceTarget(lane.MergeReferences)
 		materializedLane.NormalizeReferences = CloneReferenceTarget(lane.NormalizeReferences)
-		extractReferenceSet, laneWarnings, err := materializeReferenceTarget(resolved.ID, lane.ExtractReferences, lane.ArtifactKind, catalog, options)
+		extractReferenceSet, laneDiagnostics, err := materializeReferenceTarget(resolved.ID, lane.ExtractReferences, lane.ArtifactKind, catalog, options)
 		if err != nil {
 			return ResolvedArtifactLane{}, nil, err
 		}
 		materializedLane.ExtractReferences.ReferenceSet = extractReferenceSet
-		allWarnings = append(allWarnings, laneWarnings...)
+		allDiagnostics = append(allDiagnostics, laneDiagnostics...)
 
-		mergeReferenceSet, laneWarnings, err := materializeReferenceTarget(resolved.ID, lane.MergeReferences, lane.ArtifactKind, catalog, options)
+		mergeReferenceSet, laneDiagnostics, err := materializeReferenceTarget(resolved.ID, lane.MergeReferences, lane.ArtifactKind, catalog, options)
 		if err != nil {
 			return ResolvedArtifactLane{}, nil, err
 		}
 		materializedLane.MergeReferences.ReferenceSet = mergeReferenceSet
-		allWarnings = append(allWarnings, laneWarnings...)
+		allDiagnostics = append(allDiagnostics, laneDiagnostics...)
 
-		normalizeReferenceSet, laneWarnings, err := materializeReferenceTarget(resolved.ID, lane.NormalizeReferences, lane.ArtifactKind, catalog, options)
+		normalizeReferenceSet, laneDiagnostics, err := materializeReferenceTarget(resolved.ID, lane.NormalizeReferences, lane.ArtifactKind, catalog, options)
 		if err != nil {
 			return ResolvedArtifactLane{}, nil, err
 		}
 		materializedLane.NormalizeReferences.ReferenceSet = normalizeReferenceSet
-		allWarnings = append(allWarnings, laneWarnings...)
-		return materializedLane, allWarnings, nil
+		allDiagnostics = append(allDiagnostics, laneDiagnostics...)
+		return materializedLane, allDiagnostics, nil
 	}
 	if len(resolved.Steps) > 0 {
 		out.Steps = make([]ResolvedPipelineStep, len(resolved.Steps))
@@ -76,16 +76,16 @@ func MaterializeReferences(resolved ResolvedPipeline, catalog ModuleCatalog, opt
 			out.Steps[i].ID = step.ID
 			out.Steps[i].ArtifactLanes = make([]ResolvedArtifactLane, len(step.ArtifactLanes))
 			for j, lane := range step.ArtifactLanes {
-				materializedLane, laneWarnings, err := materializeLane(lane)
+				materializedLane, laneDiagnostics, err := materializeLane(lane)
 				if err != nil {
 					return ResolvedPipeline{}, nil, err
 				}
 				out.Steps[i].ArtifactLanes[j] = materializedLane
-				warnings = append(warnings, laneWarnings...)
+				diagnostics = append(diagnostics, laneDiagnostics...)
 			}
 		}
 	}
-	return out, warnings, nil
+	return out, diagnostics, nil
 }
 
 func materializeReferenceTarget(
@@ -94,7 +94,7 @@ func materializeReferenceTarget(
 	artifactKind contracts.ArtifactKind,
 	catalog ModuleCatalog,
 	options ReferenceMaterializationOptions,
-) (contracts.ReferenceSet, []contracts.Warning, error) {
+) (contracts.ReferenceSet, []contracts.ProducerDiagnostic, error) {
 	if len(target.Bindings) == 0 {
 		return contracts.ReferenceSet{}, nil, nil
 	}
@@ -109,7 +109,7 @@ func materializeReferenceTarget(
 	}
 
 	set := contracts.ReferenceSet{Slots: make(map[string]contracts.ResolvedReferenceSlot, len(target.Bindings))}
-	var warnings []contracts.Warning
+	var diagnostics []contracts.ProducerDiagnostic
 	for _, binding := range target.Bindings {
 		slotName := strings.TrimSpace(binding.SlotName)
 		slot, ok := slotByName[slotName]
@@ -159,11 +159,7 @@ func materializeReferenceTarget(
 			return contracts.ReferenceSet{}, nil, fmt.Errorf("%s reference slot %q path %q media type %q is not accepted", referenceTargetContext(pipelineID, target), slotName, path, mediaType)
 		}
 		if len(content) == 0 {
-			warnings = append(warnings, contracts.Warning{
-				Scope:      referenceWarningScope(pipelineID, target, slotName),
-				ReasonCode: "empty_reference",
-				Message:    fmt.Sprintf("reference slot %q for %s is bound to an empty file", slotName, referenceTargetLabel(target)),
-			})
+			diagnostics = append(diagnostics, contracts.ProducerDiagnostic{Disposition: contracts.DiagnosticDispositionWarning, Category: contracts.DiagnosticCategoryConfiguration, ReasonCode: "empty_reference", OccurrenceCount: 1, Samples: []contracts.DiagnosticSample{{Scope: referenceWarningScope(pipelineID, target, slotName), Message: fmt.Sprintf("reference slot %q for %s is bound to an empty file", slotName, referenceTargetLabel(target))}}})
 		}
 
 		item := contracts.ReferenceItem{
@@ -180,7 +176,7 @@ func materializeReferenceTarget(
 			Items: []contracts.ReferenceItem{item},
 		}
 	}
-	return set, warnings, nil
+	return set, diagnostics, nil
 }
 
 type referenceSizeLimitError struct {

internal/framework/pipeline/references_test.go

@@ -445,14 +445,14 @@ func TestMaterializeReferencesWarnsForEmptyFiles(t *testing.T) {
 	writeReferenceFile(t, path, nil)
 
 	resolved := resolvedPipelineWithReference(t, "roster", "empty.txt", contracts.ReferenceBindingSourceConfig, contracts.ReferenceSlot{Name: "roster"})
-	materialized, warnings, err := MaterializeReferences(resolved, referenceCatalog(t, []contracts.ReferenceSlot{{Name: "roster"}}), ReferenceMaterializationOptions{
+	materialized, diagnostics, err := MaterializeReferences(resolved, referenceCatalog(t, []contracts.ReferenceSlot{{Name: "roster"}}), ReferenceMaterializationOptions{
 		ConfigPath: filepath.Join(configDir, "config.yml"),
 	})
 	if err != nil {
 		t.Fatalf("MaterializeReferences() error = %v, want nil", err)
 	}
-	if len(warnings) != 1 || warnings[0].ReasonCode != "empty_reference" {
-		t.Fatalf("warnings = %#v, want empty reference warning", warnings)
+	if len(diagnostics) != 1 || diagnostics[0].Disposition != contracts.DiagnosticDispositionWarning || diagnostics[0].Category != contracts.DiagnosticCategoryConfiguration || diagnostics[0].ReasonCode != "empty_reference" || diagnostics[0].OccurrenceCount != 1 || len(diagnostics[0].Samples) != 1 {
+		t.Fatalf("diagnostics = %#v, want structured empty-reference signal", diagnostics)
 	}
 	item := materialized.Steps[0].ArtifactLanes[0].ExtractReferences.ReferenceSet.Slots["roster"].Items[0]
 	if item.SizeBytes != 0 || item.Digest != referenceDigest(nil) {
@@ -483,13 +483,13 @@ func TestMaterializeReferencesWarningScopesIncludeTargetContext(t *testing.T) {
 		t.Fatalf("ResolvePipeline() error = %v, want nil", err)
 	}
 
-	_, warnings, err := MaterializeReferences(resolved, catalog, ReferenceMaterializationOptions{
+	_, diagnostics, err := MaterializeReferences(resolved, catalog, ReferenceMaterializationOptions{
 		ConfigPath: filepath.Join(configDir, "config.yml"),
 	})
 	if err != nil {
 		t.Fatalf("MaterializeReferences() error = %v, want nil", err)
 	}
-	got := warningScopes(warnings)
+	got := diagnosticScopes(diagnostics)
 	want := []string{
 		"pipeline.baseline.chunk.reference.scene_guide",
 		"pipeline.baseline.lane.events.extract.reference.roster",
@@ -621,10 +621,10 @@ func referenceCatalogForTargets(t *testing.T, chunkSlots, extractSlots, mergeSlo
 	)
 }
 
-func warningScopes(warnings []contracts.Warning) []string {
-	scopes := make([]string, 0, len(warnings))
-	for _, warning := range warnings {
-		scopes = append(scopes, warning.Scope)
+func diagnosticScopes(diagnostics []contracts.ProducerDiagnostic) []string {
+	scopes := make([]string, 0, len(diagnostics))
+	for _, diagnostic := range diagnostics {
+		scopes = append(scopes, diagnostic.Samples[0].Scope)
 	}
 	sort.Strings(scopes)
 	return scopes

internal/framework/pipeline/runner.go

@@ -50,7 +50,7 @@ type RunInput struct {
 	StartedAt        time.Time
 	LLMProfiles      []artifacts.LLMProfileManifest
 	Metadata         map[string]any
-	Warnings         []contracts.Warning
+	Diagnostics      []contracts.ProducerDiagnostic
 	ChunkCacheMode   ChunkCacheMode
 	ChunkPlans       ChunkPlanStore
 	Checkpoints      CheckpointRecorder
@@ -69,16 +69,17 @@ type RunInput struct {
 }
 
 type RunOutput struct {
-	Manifest            artifacts.RunManifest         `json:"manifest"`
-	ChunkPlan           *artifacts.ChunkPlanSummary   `json:"chunk_plan,omitempty"`
-	NormalizeOutputs    []contracts.SerializedOutput  `json:"normalize_outputs,omitempty"`
-	Rejected            []contracts.RejectedOutput    `json:"rejected,omitempty"`
-	Warnings            []contracts.Warning           `json:"warnings,omitempty"`
-	OutputFiles         []contracts.OutputFile        `json:"-"`
-	CheckpointEvents    []CheckpointEvent             `json:"checkpoint_events,omitempty"`
-	ValidationSummaries []artifacts.ValidationSummary `json:"validation_summaries,omitempty"`
+	Manifest            artifacts.RunManifest          `json:"manifest"`
+	ChunkPlan           *artifacts.ChunkPlanSummary    `json:"chunk_plan,omitempty"`
+	NormalizeOutputs    []contracts.SerializedOutput   `json:"normalize_outputs,omitempty"`
+	Rejected            []contracts.RejectedOutput     `json:"rejected,omitempty"`
+	Diagnostics         contracts.DiagnosticCollection `json:"diagnostics,omitempty"`
+	OutputFiles         []contracts.OutputFile         `json:"-"`
+	CheckpointEvents    []CheckpointEvent              `json:"checkpoint_events,omitempty"`
+	ValidationSummaries []artifacts.ValidationSummary  `json:"validation_summaries,omitempty"`
 
 	normalizeReuseEligibility map[generatedOutputKey]bool
+	diagnosticGroups          []contracts.DiagnosticGroup
 }
 
 func (r *Runner) Run(ctx context.Context, input RunInput) (output RunOutput, err error) {
@@ -129,7 +130,11 @@ func (r *Runner) Run(ctx context.Context, input RunInput) (output RunOutput, err
 	defer func() {
 		output.Manifest.LLMProfiles = mergeLLMProfileManifests(input.LLMProfiles, llmProfileManifests(input.llmClient))
 	}()
-	output.Warnings = append(output.Warnings, cloneWarnings(input.Warnings)...)
+	inputDiagnostics, diagnosticErr := promoteProducerDiagnostics(input.Diagnostics, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageReferences}, nil)
+	if diagnosticErr != nil {
+		return failOutput(output), fmt.Errorf("promote input diagnostics: %w", diagnosticErr)
+	}
+	appendDiagnosticGroups(&output, inputDiagnostics)
 	if err := writeDebugTimed(debugRecorder, "run.json", debugTimedEnvelope{
 		Stage:     "run",
 		StartedAt: startedTime(input.StartedAt),
@@ -248,13 +253,12 @@ func (r *Runner) Run(ctx context.Context, input RunInput) (output RunOutput, err
 	if chunkResult.validation != nil {
 		output.ValidationSummaries = append(output.ValidationSummaries, artifacts.CloneValidationSummary(*chunkResult.validation))
 	}
-	output.Warnings = append(output.Warnings, chunkResult.warnings...)
+	appendDiagnosticGroups(&output, chunkResult.diagnostics)
 	chunkDebugPayload := map[string]any{
 		"cache_mode":          chunkMode,
 		"lookup":              chunkResult.lookup,
 		"accepted":            chunkResult.accepted,
 		"materialized_chunks": debugSourceChunkEnvelopes(chunkResult.chunks),
-		"warnings":            chunkResult.warnings,
 	}
 	if chunkResult.plan != nil {
 		chunkDebugPayload["plan"] = debugChunkPlanEnvelope(*chunkResult.plan)
@@ -311,6 +315,9 @@ func (r *Runner) Run(ctx context.Context, input RunInput) (output RunOutput, err
 	}
 	populateOutputManifest(&output)
 	output.Manifest.CompletedAt = timePtr(time.Now().UTC())
+	if err := finalizeDiagnostics(&output); err != nil {
+		return failOutput(output), fmt.Errorf("aggregate diagnostics: %w", err)
+	}
 
 	encoder := input.Prepared.output
 	if err := attachModuleManifestMetadata(&output, "output", encoder); err != nil {
@@ -344,7 +351,6 @@ func (r *Runner) Run(ctx context.Context, input RunInput) (output RunOutput, err
 		"manifest":          output.Manifest,
 		"normalize_outputs": debugSerializedOutputEnvelopes(output.NormalizeOutputs),
 		"rejected":          debugRejectedOutputEnvelopes(output.Rejected),
-		"warnings":          output.Warnings,
 		"options":           redactSensitiveMap(input.pipeline.Output.Options),
 		"metadata":          redactSensitiveMap(input.Metadata),
 	}
@@ -373,7 +379,7 @@ func (r *Runner) Run(ctx context.Context, input RunInput) (output RunOutput, err
 		Manifest:                       output.Manifest,
 		NormalizeOutputs:               cloneSerializedOutputs(output.NormalizeOutputs),
 		Rejected:                       cloneRejectedOutputs(output.Rejected),
-		Warnings:                       output.Warnings,
+		Diagnostics:                    contracts.CloneDiagnosticCollection(output.Diagnostics),
 		LLMProfile:                     input.pipeline.Output.LLMProfile,
 		StructuredOutputRepairAttempts: cloneStructuredOutputRepairAttempts(input.pipeline.Output.StructuredOutputRepairAttempts),
 		Metadata:                       outputMetadata,
@@ -395,8 +401,7 @@ func (r *Runner) Run(ctx context.Context, input RunInput) (output RunOutput, err
 		ModuleKey: encoder.Key(),
 		StartedAt: outputStarted,
 		Payload: map[string]any{
-			"files":    debugOutputFiles(files),
-			"warnings": encoded.Warnings,
+			"files": debugOutputFiles(files),
 		},
 	}); err != nil {
 		return failOutput(output), fmt.Errorf("write output debug artifact: %w", err)
@@ -404,7 +409,6 @@ func (r *Runner) Run(ctx context.Context, input RunInput) (output RunOutput, err
 	if err := ctx.Err(); err != nil {
 		return failOutput(output), err
 	}
-	output.Warnings = append(output.Warnings, encoded.Warnings...)
 	output.OutputFiles = files
 
 	return output, nil
@@ -444,7 +448,6 @@ func (r *Runner) runPreparedSteps(ctx context.Context, input RunInput, checkpoin
 type retryAttemptResult struct {
 	accepted  bool
 	rejection *contracts.RejectedOutput
-	warnings  []contracts.Warning
 }
 
 func runSimpleRetry(ctx context.Context, retries int, run func(attempt int) (retryAttemptResult, error)) (retryAttemptResult, error) {
@@ -474,7 +477,7 @@ func runSimpleRetry(ctx context.Context, retries int, run func(attempt int) (ret
 		if result.rejection != nil {
 			rejection := *result.rejection
 			rejection.AttemptCount = attempt
-			last = retryAttemptResult{rejection: &rejection, warnings: cloneWarnings(result.warnings)}
+			last = retryAttemptResult{rejection: &rejection}
 		}
 		if ctxErr := ctx.Err(); ctxErr != nil {
 			return retryAttemptResult{}, ctxErr
@@ -697,6 +700,10 @@ func validatorChainManifests(chains []ResolvedValidatorChain) []artifacts.Valida
 }
 
 func failOutput(output RunOutput) RunOutput {
+	// Preserve any diagnostics already accepted by the pipeline when a later
+	// operation fails. Producers and validators validate each diagnostic before
+	// it is appended, so finalization here cannot introduce a new failure path.
+	_ = finalizeDiagnostics(&output)
 	if output.Manifest.PipelineID != "" {
 		populateOutputManifest(&output)
 		output.Manifest.ValidationStatus = "failed"
@@ -971,13 +978,6 @@ func manifestMetadataWithSessionID(metadata map[string]any, sessionID string) (m
 	return out, nil
 }
 
-func cloneWarnings(warnings []contracts.Warning) []contracts.Warning {
-	if len(warnings) == 0 {
-		return nil
-	}
-	return append([]contracts.Warning(nil), warnings...)
-}
-
 func cloneSourceChunkPtr(chunk *source.Chunk) (*source.Chunk, error) {
 	if chunk == nil {
 		return nil, nil

internal/framework/pipeline/runner_accepted_checkpoint_test.go

@@ -84,7 +84,8 @@ func TestRunnerHydratesRequiredNormalizedArtifact(t *testing.T) {
 	loader := newAcceptedCheckpointLoader()
 	producerKey := CheckpointLaneKey(producer.resolved.StepID, producer.resolved.ID)
 	consumerKey := CheckpointLaneKey(consumer.resolved.StepID, consumer.resolved.ID)
-	loader.accepted[producerKey] = NormalizeCheckpoint{Output: stored, Warnings: []contracts.Warning{{Scope: "normalize", ReasonCode: "stored-warning", Message: "stored normalize warning"}}}
+	stored.Diagnostics = []CheckpointDiagnostic{{Diagnostic: contracts.ProducerDiagnostic{Disposition: contracts.DiagnosticDispositionWarning, Category: contracts.DiagnosticCategoryFallback, ReasonCode: "stored-warning", OccurrenceCount: 1, Samples: []contracts.DiagnosticSample{{Scope: "normalize", Message: "stored normalize warning"}}}}}
+	loader.accepted[producerKey] = NormalizeCheckpoint{Output: stored}
 	loader.acceptedDecision[producerKey] = NewCheckpointDecision(CheckpointDecisionReused, CheckpointReasonAcceptedArtifactReused)
 	policy := CheckpointExecutionPolicy{
 		RequireReusableLanes: map[string]struct{}{producerKey: {}},
@@ -101,8 +102,8 @@ func TestRunnerHydratesRequiredNormalizedArtifact(t *testing.T) {
 	if string(item.Content) != string(stored.Artifact.Content) || item.Producer.StepID != producer.resolved.StepID || item.Producer.LaneID != producer.resolved.ID {
 		t.Fatalf("consumer generated reference = %#v, want exact hydrated producer bytes and identity", item)
 	}
-	if len(output.Warnings) != 1 || output.Warnings[0].ReasonCode != "stored-warning" {
-		t.Fatalf("hydrated warnings = %#v, want normalize checkpoint warnings only", output.Warnings)
+	if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].ReasonCode != "stored-warning" {
+		t.Fatalf("hydrated diagnostics = %#v, want normalize checkpoint diagnostics only", output.Diagnostics)
 	}
 	assertAcceptedNormalizeEvent(t, output.CheckpointEvents, producer.resolved.StepID, producer.resolved.ID, CheckpointDecisionReused, CheckpointReasonAcceptedArtifactReused)
 	for _, event := range output.CheckpointEvents {
@@ -225,7 +226,8 @@ func TestRunnerRetainsEarlierHydratedProducerWhenLaterRequiredProducerFails(t *t
 	loader := newAcceptedCheckpointLoader()
 	firstKey := CheckpointLaneKey(first.resolved.StepID, first.resolved.ID)
 	secondKey := CheckpointLaneKey(second.resolved.StepID, second.resolved.ID)
-	loader.accepted[firstKey] = NormalizeCheckpoint{Output: stored, Warnings: []contracts.Warning{{Scope: "normalize", ReasonCode: "retained-warning", Message: "retained warning"}}}
+	stored.Diagnostics = []CheckpointDiagnostic{{Diagnostic: contracts.ProducerDiagnostic{Disposition: contracts.DiagnosticDispositionWarning, Category: contracts.DiagnosticCategoryFallback, ReasonCode: "retained-warning", OccurrenceCount: 1, Samples: []contracts.DiagnosticSample{{Scope: "normalize", Message: "retained warning"}}}}}
+	loader.accepted[firstKey] = NormalizeCheckpoint{Output: stored}
 	loader.acceptedDecision[firstKey] = NewCheckpointDecision(CheckpointDecisionReused, CheckpointReasonAcceptedArtifactReused)
 	loader.acceptedDecision[secondKey] = NewCheckpointDecision(CheckpointDecisionExecuted, CheckpointReasonMissing)
 	policy := CheckpointExecutionPolicy{RequireReusableLanes: map[string]struct{}{firstKey: {}, secondKey: {}}}
@@ -240,8 +242,8 @@ func TestRunnerRetainsEarlierHydratedProducerWhenLaterRequiredProducerFails(t *t
 	if len(output.NormalizeOutputs) != 1 || output.NormalizeOutputs[0].LaneID != first.resolved.ID || string(output.NormalizeOutputs[0].Artifact.Content) != string(stored.Artifact.Content) {
 		t.Fatalf("retained normalize outputs = %#v, want first producer", output.NormalizeOutputs)
 	}
-	if len(output.Warnings) != 1 || output.Warnings[0].ReasonCode != "retained-warning" {
-		t.Fatalf("retained warnings = %#v", output.Warnings)
+	if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].ReasonCode != "retained-warning" {
+		t.Fatalf("retained diagnostics = %#v", output.Diagnostics)
 	}
 	type decisionExpectation struct {
 		step, lane string

internal/framework/pipeline/runner_attempt_debug_test.go

@@ -162,13 +162,13 @@ func TestRunnerWritesAttemptScopedMergeAndNormalizeDebug(t *testing.T) {
 		if err := callAttemptDebugLLM(ctx, client, "merge"); err != nil {
 			return erasedTypedResult{}, err
 		}
-		return erasedTypedResult{Value: codecNotes{Items: []string{"merged"}}, Warnings: []contracts.Warning{{Scope: "merge", ReasonCode: "observed", Message: "merge warning"}}}, nil
+		return erasedTypedResult{Value: codecNotes{Items: []string{"merged"}}, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("merge", "merge diagnostic")}}, nil
 	}
 	prepared.Steps[0].lanes[0].typed.normalize = func(ctx context.Context, _ any, _ contracts.TypedNormalizeRequest[any]) (erasedTypedResult, error) {
 		if err := callAttemptDebugLLM(ctx, client, "normalize"); err != nil {
 			return erasedTypedResult{}, err
 		}
-		return erasedTypedResult{Value: codecNotes{Items: []string{"normalized"}}, Warnings: []contracts.Warning{{Scope: "normalize", ReasonCode: "observed", Message: "normalize warning"}}}, nil
+		return erasedTypedResult{Value: codecNotes{Items: []string{"normalized"}}, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("normalize", "normalize diagnostic")}}, nil
 	}
 
 	output, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), Debug: debug})
@@ -213,7 +213,7 @@ func TestRunnerWritesAttemptScopedMergeAndNormalizeDebug(t *testing.T) {
 	}
 }
 
-func TestRunnerRecordsDistinctRetryAttemptsAndPromotesAcceptedWarningsOnly(t *testing.T) {
+func TestRunnerRecordsDistinctRetryAttemptsAndPromotesAcceptedDiagnosticsOnly(t *testing.T) {
 	for _, stage := range []ModuleStage{StageMerge, StageNormalize} {
 		t.Run(string(stage), func(t *testing.T) {
 			prepared := preparedAttemptDebugPipeline(t)
@@ -230,7 +230,7 @@ func TestRunnerRecordsDistinctRetryAttemptsAndPromotesAcceptedWarningsOnly(t *te
 				if attempts == 1 {
 					scope = "discarded"
 				}
-				return erasedTypedResult{Value: codecNotes{Items: []string{scope}}, Warnings: []contracts.Warning{{Scope: scope, ReasonCode: "observed", Message: scope}}, ModelCandidate: attemptCandidate(t, fmt.Sprintf(`{"items":["%s"]}`, scope))}, nil
+				return erasedTypedResult{Value: codecNotes{Items: []string{scope}}, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic(scope, scope)}, ModelCandidate: attemptCandidate(t, fmt.Sprintf(`{"items":["%s"]}`, scope))}, nil
 			}
 			validatorCalls := 0
 			validator := preparedValidator{
@@ -268,11 +268,11 @@ func TestRunnerRecordsDistinctRetryAttemptsAndPromotesAcceptedWarningsOnly(t *te
 			if len(first.LLMCalls) != 1 || len(second.LLMCalls) != 1 || first.LLMCalls[0].CallID == second.LLMCalls[0].CallID {
 				t.Fatalf("retry LLM calls = first %#v, second %#v; want distinct calls", first.LLMCalls, second.LLMCalls)
 			}
-			if !strings.Contains(string(debug.json[firstPath]), "discarded") || !strings.Contains(string(debug.json[firstPath]), "rejection") {
-				t.Fatalf("first attempt envelope = %s, want discarded warning and rejection", debug.json[firstPath])
+			if !strings.Contains(string(debug.json[firstPath]), "rejection") {
+				t.Fatalf("first attempt envelope = %s, want rejection", debug.json[firstPath])
 			}
-			if len(output.Warnings) != 1 || output.Warnings[0].Scope != "accepted" {
-				t.Fatalf("promoted warnings = %#v, want accepted attempt only", output.Warnings)
+			if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].Samples[0].Scope != "accepted" {
+				t.Fatalf("promoted diagnostics = %#v, want accepted attempt only", output.Diagnostics)
 			}
 		})
 	}

internal/framework/pipeline/runner_candidate_encoding_test.go

@@ -101,13 +101,13 @@ type candidateCheckpointRecorder struct {
 	normalizeOutput    CheckpointArtifact
 }
 
-func (r *candidateCheckpointRecorder) MergeSucceeded(_ string, _ string, _ []CheckpointFingerprint, output CheckpointArtifact, _ []contracts.Warning) error {
+func (r *candidateCheckpointRecorder) MergeSucceeded(_ string, _ string, _ []CheckpointFingerprint, output CheckpointArtifact) error {
 	r.mergeSucceeded++
 	r.mergeOutput = cloneCheckpointArtifact(output)
 	return nil
 }
 
-func (r *candidateCheckpointRecorder) NormalizeSucceeded(_ string, _ string, _ []CheckpointFingerprint, output CheckpointArtifact, _ []contracts.Warning) error {
+func (r *candidateCheckpointRecorder) NormalizeSucceeded(_ string, _ string, _ []CheckpointFingerprint, output CheckpointArtifact) error {
 	r.normalizeSucceeded++
 	r.normalizeOutput = cloneCheckpointArtifact(output)
 	return nil

internal/framework/pipeline/runner_chunk_plan.go

@@ -12,24 +12,24 @@ import (
 )
 
 type chunkPlanExecution struct {
-	accepted   bool
-	chunks     []source.Chunk
-	plan       *source.ChunkPlan
-	warnings   []contracts.Warning
-	rejection  *contracts.RejectedOutput
-	lookup     ChunkPlanDecision
-	record     *ChunkPlanRecord
-	action     string
-	summary    artifacts.ChunkPlanSummary
-	validation *artifacts.ValidationSummary
+	accepted    bool
+	chunks      []source.Chunk
+	plan        *source.ChunkPlan
+	diagnostics []contracts.DiagnosticGroup
+	rejection   *contracts.RejectedOutput
+	lookup      ChunkPlanDecision
+	record      *ChunkPlanRecord
+	action      string
+	summary     artifacts.ChunkPlanSummary
+	validation  *artifacts.ValidationSummary
 }
 
 type generatedChunkPlanCandidate struct {
-	plan             source.ChunkPlan
-	chunks           []source.Chunk
-	record           ChunkPlanRecord
-	producerWarnings []contracts.Warning
-	terminal         *attemptTerminalRecorder
+	plan                source.ChunkPlan
+	chunks              []source.Chunk
+	record              ChunkPlanRecord
+	producerDiagnostics []contracts.ProducerDiagnostic
+	terminal            *attemptTerminalRecorder
 }
 
 func effectiveChunkCacheMode(mode ChunkCacheMode) ChunkCacheMode {
@@ -69,7 +69,7 @@ func (r *Runner) runChunkPlan(ctx context.Context, input RunInput, doc *source.S
 			if validationErr == nil {
 				report, err := r.validateChunkReport(ctx, doc, chunker.Key(), chunks, sourceInput, sessionID, input.pipeline.ChunkReferences.ReferenceSet, input.Metadata, input.Prepared.chunkValidators, 1, input.Debug)
 				if err != nil {
-					result.setValidation(report.Warnings(), nil, err)
+					result.setValidation(nil, err)
 					return result, err
 				}
 				rejection := report.FirstRejection()
@@ -80,24 +80,28 @@ func (r *Runner) runChunkPlan(ctx context.Context, input RunInput, doc *source.S
 					}
 					result.plan = &plan
 					result.chunks = chunks
-					result.warnings = append(cloneWarnings(record.Warnings), report.Warnings()...)
 					result.accepted = true
-					result.setValidation(report.Warnings(), nil, nil)
-					cachedTerminal := producerAttemptTerminal{Action: producerTerminalAccepted, Validation: report}
+					result.setValidation(nil, nil)
+					cachedTerminal := producerAttemptTerminal{Action: producerTerminalAccepted, Diagnostics: contracts.CloneProducerDiagnostics(record.Diagnostics), Validation: report}
+					diagnostics, diagnosticErr := terminalDiagnosticGroups(cachedTerminal, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageChunk, ModuleKey: chunker.Key()}, nil)
+					if diagnosticErr != nil {
+						return result, fmt.Errorf("promote reused chunk diagnostics: %w", diagnosticErr)
+					}
+					result.diagnostics = diagnostics
 					summary := validationSummary(cachedTerminal, StageChunk, "", "", chunker.Key(), "", 0)
 					result.validation = &summary
 					return result, nil
 				}
 				if incomplete != nil && input.pipeline.ChunkValidationPolicy.ValidatorFailure == ValidatorFailureFailRun {
 					failure := validatorFailureError(*incomplete)
-					result.setValidation(report.Warnings(), nil, failure)
+					result.setValidation(nil, failure)
 					failedTerminal := producerAttemptTerminal{Action: producerTerminalFailed, Validation: report, ValidationIncomplete: true}
 					summary := validationSummary(failedTerminal, StageChunk, "", "", chunker.Key(), "", 0)
 					result.validation = &summary
 					return result, failure
 				}
 				// A cache hit is not model material. Its rejection is discarded and
-				// generation begins with the ordinary initial request below. Warnings
+				// generation begins with the ordinary initial request below. Diagnostics
 				// from this discarded candidate are intentionally not promoted.
 			}
 			result.lookup = ChunkPlanDecision{Status: ChunkPlanInvalid, Reason: chunkPlanLookupReason(ChunkPlanInvalid)}
@@ -138,7 +142,7 @@ func (r *Runner) runChunkPlan(ctx context.Context, input RunInput, doc *source.S
 		plan, chunks, validationErr := validateAndMaterializeChunkPlan(doc, chunkResult.Plan)
 		if validationErr != nil {
 			attemptErr := fmt.Errorf("validate chunk plan from chunker %q: %w", chunker.Key(), validationErr)
-			payload := map[string]any{"plan": debugChunkPlanEnvelope(chunkResult.Plan), "warnings": debugWarningEnvelopes(chunkResult.Warnings)}
+			payload := map[string]any{"plan": debugChunkPlanEnvelope(chunkResult.Plan)}
 			return producerAttemptOutput{}, attemptTerminal.record(payload, fmt.Errorf("%w: %v", contracts.ErrInvalidStructuredOutput, attemptErr))
 		}
 		planDigest, digestErr := source.DigestChunkPlan(plan)
@@ -161,19 +165,18 @@ func (r *Runner) runChunkPlan(ctx context.Context, input RunInput, doc *source.S
 				References: append([]artifacts.ReferenceProvenance(nil), referenceTargetProvenance(input.pipeline.ChunkReferences)...),
 				Metadata:   producerMetadata,
 			},
-			Warnings: cloneWarnings(chunkResult.Warnings), CreatedAt: time.Now().UTC(),
+			CreatedAt: time.Now().UTC(),
 		}
-		return producerAttemptOutput{Value: generatedChunkPlanCandidate{plan: plan, chunks: chunks, record: candidate, producerWarnings: cloneWarnings(chunkResult.Warnings), terminal: &attemptTerminal}, Candidate: chunkResult.ModelCandidate, Warnings: cloneWarnings(chunkResult.Warnings)}, nil
+		return producerAttemptOutput{Value: generatedChunkPlanCandidate{plan: plan, chunks: chunks, record: candidate, producerDiagnostics: contracts.CloneProducerDiagnostics(chunkResult.Diagnostics), terminal: &attemptTerminal}, Candidate: chunkResult.ModelCandidate, Diagnostics: contracts.CloneProducerDiagnostics(chunkResult.Diagnostics)}, nil
 	}, func(validationCtx context.Context, output producerAttemptOutput) (validationReport, error) {
 		candidate, ok := output.Value.(generatedChunkPlanCandidate)
 		if !ok {
 			return validationReport{}, fmt.Errorf("chunk attempt candidate has incompatible type")
 		}
 		report, validationErr := r.validateChunkReport(validationCtx, doc, chunker.Key(), candidate.chunks, sourceInput, sessionID, input.pipeline.ChunkReferences.ReferenceSet, input.Metadata, input.Prepared.chunkValidators, candidate.terminal.envelope.Attempt, input.Debug)
-		attemptWarnings := append(cloneWarnings(output.Warnings), report.Warnings()...)
 		payload := map[string]any{
 			"plan": debugChunkPlanEnvelope(candidate.plan), "materialized_chunks": debugSourceChunkEnvelopes(candidate.chunks),
-			"warnings": debugWarningEnvelopes(attemptWarnings), "rejection": debugRejectedOutputPtr(chunkRejection(report, candidate.terminal.envelope.Attempt, chunker.Key())),
+			"rejection": debugRejectedOutputPtr(chunkRejection(report, candidate.terminal.envelope.Attempt, chunker.Key())),
 		}
 		if validationErr != nil {
 			return report, candidate.terminal.record(payload, validationErr)
@@ -190,18 +193,22 @@ func (r *Runner) runChunkPlan(ctx context.Context, input RunInput, doc *source.S
 		}
 		return result, err
 	}
+	diagnostics, diagnosticErr := terminalDiagnosticGroups(terminal, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageChunk, ModuleKey: chunker.Key()}, nil)
+	if diagnosticErr != nil {
+		return result, fmt.Errorf("promote chunk diagnostics: %w", diagnosticErr)
+	}
+	result.diagnostics = diagnostics
 	if terminal.Action == producerTerminalRejected {
 		result.rejection = terminal.Rejection
 		if result.rejection != nil {
 			result.rejection.Stage = string(StageChunk)
 			result.rejection.ModuleKey = chunker.Key()
 		}
-		result.warnings = cloneWarnings(terminal.Warnings)
 		result.validation = &terminalSummary
 		if result.rejection != nil {
 			result.rejection.Validation = cloneValidationSummaryPtr(result.validation)
 		}
-		result.setValidation(terminal.Validation.Warnings(), result.rejection, nil)
+		result.setValidation(result.rejection, nil)
 		return result, nil
 	}
 	candidate, ok := terminal.Value.(generatedChunkPlanCandidate)
@@ -222,9 +229,8 @@ func (r *Runner) runChunkPlan(ctx context.Context, input RunInput, doc *source.S
 	result.accepted = true
 	result.plan = &candidate.plan
 	result.chunks = candidate.chunks
-	result.warnings = cloneWarnings(terminal.Warnings)
 	result.validation = &terminalSummary
-	result.setValidation(terminal.Validation.Warnings(), nil, nil)
+	result.setValidation(nil, nil)
 	if terminal.ValidationIncomplete {
 		result.summary.ValidationStatus = "incomplete"
 	}
@@ -234,7 +240,7 @@ func (r *Runner) runChunkPlan(ctx context.Context, input RunInput, doc *source.S
 		if cloneErr != nil {
 			return result, fmt.Errorf("clone chunk plan record for publication: %w", cloneErr)
 		}
-		record.Warnings = cloneWarnings(candidate.producerWarnings)
+		record.Diagnostics = contracts.CloneProducerDiagnostics(candidate.producerDiagnostics)
 		if err := input.ChunkPlans.Save(record); err != nil {
 			return result, fmt.Errorf("save chunk plan: %w", err)
 		}
@@ -290,14 +296,12 @@ func (result *chunkPlanExecution) setCandidate(record ChunkPlanRecord, action st
 	return nil
 }
 
-func (result *chunkPlanExecution) setValidation(warnings []contracts.Warning, rejection *contracts.RejectedOutput, err error) {
+func (result *chunkPlanExecution) setValidation(rejection *contracts.RejectedOutput, err error) {
 	switch {
 	case err != nil:
 		result.summary.ValidationStatus = "error"
 	case rejection != nil:
 		result.summary.ValidationStatus = "rejected"
-	case len(warnings) > 0:
-		result.summary.ValidationStatus = "approved_with_warnings"
 	default:
 		result.summary.ValidationStatus = "approved"
 	}
@@ -311,7 +315,7 @@ func cloneChunkPlanRecord(record ChunkPlanRecord) (ChunkPlanRecord, error) {
 		return ChunkPlanRecord{}, fmt.Errorf("clone chunk plan producer metadata: %w", err)
 	}
 	record.Producer.Metadata = metadata
-	record.Warnings = cloneWarnings(record.Warnings)
+	record.Diagnostics = contracts.CloneProducerDiagnostics(record.Diagnostics)
 	return record, nil
 }
 

internal/framework/pipeline/runner_chunk_plan_test.go

@@ -118,9 +118,9 @@ func (*retryingChunker) ReferenceSlots() []contracts.ReferenceSlot { return nil
 func (c *retryingChunker) Plan(context.Context, contracts.ChunkRequest) (contracts.ChunkPlanResult, error) {
 	c.calls++
 	if c.calls == 1 {
-		return contracts.ChunkPlanResult{Warnings: []contracts.Warning{{Scope: "discarded", ReasonCode: "retry", Message: "discarded warning"}}}, errors.New("retry generation")
+		return contracts.ChunkPlanResult{Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("discarded", "discarded diagnostic")}}, errors.New("retry generation")
 	}
-	return contracts.ChunkPlanResult{Plan: source.CloneChunkPlan(c.plan), Warnings: []contracts.Warning{{Scope: "accepted", ReasonCode: "observed", Message: "accepted warning"}}}, nil
+	return contracts.ChunkPlanResult{Plan: source.CloneChunkPlan(c.plan), Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("accepted", "accepted diagnostic")}}, nil
 }
 
 type dependencyLoader struct {
@@ -233,6 +233,37 @@ func TestRunnerChunkPlanHitUsesStoredProducerProvenance(t *testing.T) {
 	}
 }
 
+func TestRunnerReusesChunkPlanDiagnosticsWithCurrentValidatorDiagnostics(t *testing.T) {
+	prepared, plan := preparedTerminalDebugPipeline(t)
+	record := chunkPlanRecord(t, prepared, plan)
+	record.Diagnostics = []contracts.ProducerDiagnostic{{
+		Disposition:     contracts.DiagnosticDispositionWarning,
+		Category:        contracts.DiagnosticCategoryConfiguration,
+		ReasonCode:      "stored_chunk_diagnostic",
+		OccurrenceCount: 1,
+		Samples:         []contracts.DiagnosticSample{{Scope: "reference", Message: "Stored configuration signal."}},
+	}}
+	validator := &countingChunkValidator{result: contracts.ValidationResult{Approved: true, Diagnostics: []contracts.ProducerDiagnostic{{
+		Disposition:     contracts.DiagnosticDispositionAdvisory,
+		Category:        contracts.DiagnosticCategoryDataQuality,
+		ReasonCode:      "current_validator_diagnostic",
+		OccurrenceCount: 1,
+		Samples:         []contracts.DiagnosticSample{{Scope: "chunk", Message: "Current validator finding."}},
+	}}}}
+	prepared.chunkValidators.validators = []preparedValidator{{resolved: ResolvedValidator{Binding: Binding(validator.Name()), Target: ValidatorTargetChunk}, chunk: validator}}
+
+	output, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), ChunkCacheMode: ChunkCacheAuto, ChunkPlans: &recordingChunkPlanStore{record: record, decision: ChunkPlanDecision{Status: ChunkPlanHit}}})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if validator.calls != 1 || len(output.Diagnostics.Groups) != 2 {
+		t.Fatalf("validator calls = %d diagnostics = %#v", validator.calls, output.Diagnostics)
+	}
+	if output.Diagnostics.Groups[0].ReasonCode != "stored_chunk_diagnostic" || output.Diagnostics.Groups[1].ReasonCode != "current_validator_diagnostic" || output.Diagnostics.Groups[1].Origin.ValidatorKey != validator.Name() {
+		t.Fatalf("diagnostic groups = %#v", output.Diagnostics.Groups)
+	}
+}
+
 func TestRunnerProvidesAcceptedChunkMapToOutput(t *testing.T) {
 	for _, test := range []struct {
 		name      string
@@ -368,7 +399,7 @@ func TestRunnerRegeneratesValidationIncompleteChunkPlanHit(t *testing.T) {
 		}},
 	}}
 	record := chunkPlanRecord(t, prepared, plan)
-	record.Warnings = []contracts.Warning{{Scope: "stored", ReasonCode: "old", Message: "discarded stored warning"}}
+	record.Diagnostics = []contracts.ProducerDiagnostic{producerDiagnostic("stored", "discarded stored diagnostic")}
 	store := &recordingChunkPlanStore{record: record, decision: ChunkPlanDecision{Status: ChunkPlanHit}}
 	debug := newCapturedDebugRecorder()
 
@@ -382,8 +413,8 @@ func TestRunnerRegeneratesValidationIncompleteChunkPlanHit(t *testing.T) {
 	if output.ChunkPlan == nil || output.ChunkPlan.Action != "generated" || output.ChunkPlan.LookupStatus != "invalid" || output.ChunkPlan.ValidationStatus != "approved" {
 		t.Fatalf("chunk plan summary = %#v", output.ChunkPlan)
 	}
-	if len(output.Warnings) != 0 {
-		t.Fatalf("warnings = %#v, want discarded cache warnings omitted", output.Warnings)
+	if len(output.Diagnostics.Groups) != 0 {
+		t.Fatalf("diagnostics = %#v, want discarded cache diagnostics omitted", output.Diagnostics)
 	}
 	assertAttemptEnvelopeSequence(t, debug, "chunk", 1)
 }
@@ -396,7 +427,7 @@ func TestRunnerKeepsStoredPlanWhenCacheAndGeneratedValidationAreIncomplete(t *te
 	validator := &countingChunkValidator{err: errors.New("validator unavailable")}
 	prepared.chunkValidators.validators = []preparedValidator{{resolved: ResolvedValidator{Binding: Binding(validator.Name()), Target: ValidatorTargetChunk}, chunk: validator}}
 	record := chunkPlanRecord(t, prepared, plan)
-	record.Warnings = []contracts.Warning{{Scope: "stored", ReasonCode: "old", Message: "discarded stored warning"}}
+	record.Diagnostics = []contracts.ProducerDiagnostic{producerDiagnostic("stored", "discarded stored diagnostic")}
 	store := &recordingChunkPlanStore{record: record, decision: ChunkPlanDecision{Status: ChunkPlanHit}}
 
 	output, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), ChunkCacheMode: ChunkCacheAuto, ChunkPlans: store})
@@ -409,8 +440,8 @@ func TestRunnerKeepsStoredPlanWhenCacheAndGeneratedValidationAreIncomplete(t *te
 	if output.ChunkPlan == nil || output.ChunkPlan.Action != "generated" || output.ChunkPlan.LookupStatus != "invalid" || output.ChunkPlan.ValidationStatus != "incomplete" {
 		t.Fatalf("chunk plan summary = %#v", output.ChunkPlan)
 	}
-	if len(output.Warnings) != 1 || output.Warnings[0].ReasonCode != "validator_execution_incomplete" {
-		t.Fatalf("warnings = %#v, want only generated incomplete warning", output.Warnings)
+	if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].ReasonCode != "validator_execution_incomplete" {
+		t.Fatalf("diagnostics = %#v, want only generated incomplete diagnostic", output.Diagnostics)
 	}
 	if !reflect.DeepEqual(store.record, record) {
 		t.Fatal("discarded incomplete candidates mutated the stored record")
@@ -580,8 +611,8 @@ func TestRunnerRetriesBeforePublishingAcceptedPlan(t *testing.T) {
 	if chunker.calls != 2 || store.saves != 1 {
 		t.Fatalf("module calls = %d saves = %d, want 2 and 1", chunker.calls, store.saves)
 	}
-	if len(output.Warnings) != 1 || output.Warnings[0].Scope != "accepted" || len(store.saved.Warnings) != 1 || store.saved.Warnings[0].Scope != "accepted" {
-		t.Fatalf("output warnings = %#v stored warnings = %#v", output.Warnings, store.saved.Warnings)
+	if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].Samples[0].Scope != "accepted" || len(store.saved.Diagnostics) != 1 || store.saved.Diagnostics[0].Samples[0].Scope != "accepted" {
+		t.Fatalf("output diagnostics = %#v stored diagnostics = %#v", output.Diagnostics, store.saved.Diagnostics)
 	}
 }
 
@@ -593,7 +624,7 @@ func TestRunnerAutoHitValidatesOnceWithoutRegenerationOrMutation(t *testing.T) {
 		wantError    string
 		wantReject   bool
 	}{
-		{name: "warning", result: contracts.ValidationResult{Approved: true, Warnings: []contracts.Warning{{Scope: "current", ReasonCode: "observed", Message: "current warning"}}}},
+		{name: "diagnostic", result: contracts.ValidationResult{Approved: true, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("current", "current diagnostic")}}},
 		{name: "rejection", result: contracts.ValidationResult{Approved: false, ReasonCode: "rejected", Message: "rejected hit", CorrectionGuidance: "return an acceptable chunk plan"}, wantReject: true},
 		{name: "error", validatorErr: errors.New("validator failed"), wantError: "validator failed"},
 	}
@@ -606,7 +637,7 @@ func TestRunnerAutoHitValidatesOnceWithoutRegenerationOrMutation(t *testing.T) {
 			validator := &countingChunkValidator{result: tc.result, err: tc.validatorErr}
 			prepared.chunkValidators.validators = []preparedValidator{{resolved: ResolvedValidator{Binding: Binding(validator.Name()), Target: ValidatorTargetChunk}, chunk: validator}}
 			record := chunkPlanRecord(t, prepared, plan)
-			record.Warnings = []contracts.Warning{{Scope: "stored", ReasonCode: "observed", Message: "stored warning"}}
+			record.Diagnostics = []contracts.ProducerDiagnostic{producerDiagnostic("stored", "stored diagnostic")}
 			store := &recordingChunkPlanStore{record: record, decision: ChunkPlanDecision{Status: ChunkPlanHit}}
 			debug := newCapturedDebugRecorder()
 			output, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), ChunkCacheMode: ChunkCacheAuto, ChunkPlans: store, Debug: debug})
@@ -642,8 +673,8 @@ func TestRunnerAutoHitValidatesOnceWithoutRegenerationOrMutation(t *testing.T) {
 			if tc.wantReject && (len(output.Rejected) != 1 || output.Rejected[0].ReasonCode != "rejected") {
 				t.Fatalf("rejected = %#v", output.Rejected)
 			}
-			if tc.wantError == "" && !tc.wantReject && len(output.Warnings) != 1+len(tc.result.Warnings) {
-				t.Fatalf("warnings = %#v, want stored warning once plus current warnings", output.Warnings)
+			if tc.wantError == "" && !tc.wantReject && len(output.Diagnostics.Groups) != 1+len(tc.result.Diagnostics) {
+				t.Fatalf("diagnostics = %#v, want stored diagnostic once plus current diagnostics", output.Diagnostics)
 			}
 		})
 	}
@@ -693,7 +724,7 @@ func TestRunnerPublishesOnlyAcceptedGeneratedPlans(t *testing.T) {
 	}
 }
 
-func TestRunnerStoresProducerProvenanceAndProducerWarnings(t *testing.T) {
+func TestRunnerStoresProducerProvenanceAndDiagnostics(t *testing.T) {
 	prepared, plan := preparedTerminalDebugPipeline(t)
 	prepared.resolved.Chunk.LLMProfile = "chunk-profile"
 	prepared.resolved.ChunkExecutionClass = contracts.ExecutionClassLLMBacked
@@ -703,8 +734,8 @@ func TestRunnerStoresProducerProvenanceAndProducerWarnings(t *testing.T) {
 			"guide": {Items: []contracts.ReferenceItem{{SlotName: "guide", Digest: "sha256:guide", Origin: contracts.ReferenceOrigin{Type: "file", URI: "file:///guide.txt"}, Content: []byte("sensitive")}}},
 		}},
 	}
-	prepared.chunker = manifestChunker{terminalChunker: terminalChunker{key: prepared.resolved.Chunk.Module, plan: plan, warnings: []contracts.Warning{{Scope: "producer", ReasonCode: "observed", Message: "producer warning"}}}, metadata: map[string]any{"prompt_id": "chunk/prompt"}}
-	validator := &countingChunkValidator{result: contracts.ValidationResult{Approved: true, Warnings: []contracts.Warning{{Scope: "validator", ReasonCode: "observed", Message: "validator warning"}}}}
+	prepared.chunker = manifestChunker{terminalChunker: terminalChunker{key: prepared.resolved.Chunk.Module, plan: plan, diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("producer", "producer diagnostic")}}, metadata: map[string]any{"prompt_id": "chunk/prompt"}}
+	validator := &countingChunkValidator{result: contracts.ValidationResult{Approved: true, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("validator", "validator diagnostic")}}}
 	prepared.chunkValidators.validators = []preparedValidator{{resolved: ResolvedValidator{Binding: Binding(validator.Name()), Target: ValidatorTargetChunk}, chunk: validator}}
 	store := &recordingChunkPlanStore{}
 	if _, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), ChunkCacheMode: ChunkCacheRefresh, ChunkPlans: store}); err != nil {
@@ -717,8 +748,8 @@ func TestRunnerStoresProducerProvenanceAndProducerWarnings(t *testing.T) {
 	if len(producer.References) != 1 || producer.References[0].Digest != "sha256:guide" {
 		t.Fatalf("producer references = %#v", producer.References)
 	}
-	if len(store.saved.Warnings) != 1 || store.saved.Warnings[0].Scope != "producer" {
-		t.Fatalf("stored warnings = %#v, want only producer warning", store.saved.Warnings)
+	if len(store.saved.Diagnostics) != 1 || store.saved.Diagnostics[0].Samples[0].Scope != "producer" {
+		t.Fatalf("stored diagnostics = %#v, want only producer diagnostic", store.saved.Diagnostics)
 	}
 }
 

internal/framework/pipeline/runner_concurrency_test.go

@@ -146,8 +146,7 @@ func (e *cancelingOutputEncoder) Encode(context.Context, contracts.OutputRequest
 	e.calls.Add(1)
 	e.cancel()
 	return contracts.OutputResult{
-		Files:    []contracts.OutputFile{{Name: "result.txt", ContentType: "text/plain", Bytes: []byte("result")}},
-		Warnings: []contracts.Warning{{ReasonCode: "returned-after-cancel"}},
+		Files: []contracts.OutputFile{{Name: "result.txt", ContentType: "text/plain", Bytes: []byte("result")}},
 	}, nil
 }
 
@@ -202,9 +201,9 @@ func TestRunnerDiscardsOutputReturnedAfterCancellation(t *testing.T) {
 	if len(output.OutputFiles) != 0 {
 		t.Fatalf("output files = %#v, want none", output.OutputFiles)
 	}
-	for _, warning := range output.Warnings {
-		if warning.ReasonCode == "returned-after-cancel" {
-			t.Fatalf("output warnings include encoder warning after cancellation")
+	for _, group := range output.Diagnostics.Groups {
+		if group.ReasonCode == "returned-after-cancel" {
+			t.Fatalf("output diagnostics include encoder diagnostic after cancellation")
 		}
 	}
 }
@@ -367,7 +366,8 @@ func TestRunnerBoundsExtractJobsAndStabilizesReverseCompletion(t *testing.T) {
 			case <-ctx.Done():
 				return erasedTypedResult{}, ctx.Err()
 			}
-			return erasedTypedResult{Value: typedValueForLane(lane, request.Chunk.Index), Warnings: []contracts.Warning{{Scope: fmt.Sprintf("lane-%d/chunk-%d", lane, request.Chunk.Index), ReasonCode: "observed", Message: "ordered"}}}, nil
+			scope := fmt.Sprintf("lane-%d/chunk-%d", lane, request.Chunk.Index)
+			return erasedTypedResult{Value: typedValueForLane(lane, request.Chunk.Index), Diagnostics: []contracts.ProducerDiagnostic{{Disposition: contracts.DiagnosticDispositionObservation, Category: contracts.DiagnosticCategoryNormalization, ReasonCode: "observed", OccurrenceCount: 1, Samples: []contracts.DiagnosticSample{{Scope: scope, Message: "ordered"}}}}}, nil
 		})
 	}
 
@@ -396,13 +396,23 @@ func TestRunnerBoundsExtractJobsAndStabilizesReverseCompletion(t *testing.T) {
 	if got := maximum.Load(); got != 2 {
 		t.Fatalf("maximum concurrent extract jobs = %d, want 2", got)
 	}
-	wantScopes := []string{"lane-0/chunk-0", "lane-0/chunk-1", "lane-0/chunk-2", "lane-1/chunk-0", "lane-1/chunk-1", "lane-1/chunk-2"}
-	gotScopes := make([]string, len(result.output.Warnings))
-	for i := range result.output.Warnings {
-		gotScopes[i] = result.output.Warnings[i].Scope
+	if len(result.output.Diagnostics.Groups) != 2 {
+		t.Fatalf("diagnostic groups = %#v, want one deterministic group per lane", result.output.Diagnostics.Groups)
 	}
-	if !reflect.DeepEqual(gotScopes, wantScopes) {
-		t.Fatalf("warning order = %#v, want %#v", gotScopes, wantScopes)
+	for lane, group := range result.output.Diagnostics.Groups {
+		if group.Origin.LaneID != prepared.Steps[0].lanes[lane].resolved.ID {
+			t.Fatalf("group origin = %#v, want configured lane %q", group.Origin, prepared.Steps[0].lanes[lane].resolved.ID)
+		}
+		indexes := make([]int, len(group.Samples))
+		for index, sample := range group.Samples {
+			if sample.ChunkIndex == nil {
+				t.Fatalf("sample = %#v, want chunk index", sample)
+			}
+			indexes[index] = *sample.ChunkIndex
+		}
+		if !reflect.DeepEqual(indexes, []int{0, 1, 2}) {
+			t.Fatalf("group sample chunk order = %#v, want canonical chunk order", indexes)
+		}
 	}
 }
 

internal/framework/pipeline/runner_concurrent.go

@@ -24,7 +24,7 @@ type laneExtractState struct {
 	reuseEligible       bool
 	values              []erasedExtractArtifact
 	serialized          []CheckpointArtifact
-	warnings            []contracts.Warning
+	diagnostics         []contracts.DiagnosticGroup
 	rejected            []contracts.RejectedOutput
 	incomplete          []int
 	validationSummaries []artifacts.ValidationSummary
@@ -38,7 +38,7 @@ type laneExtractState struct {
 type finalizedExtractResults struct {
 	accepted            []erasedExtractArtifact
 	serialized          []CheckpointArtifact
-	warnings            []contracts.Warning
+	diagnostics         []contracts.DiagnosticGroup
 	rejected            []contracts.RejectedOutput
 	incomplete          []int
 	validationSummaries []artifacts.ValidationSummary
@@ -53,8 +53,8 @@ func loadExtract(loader CheckpointLoader, stepID, laneID, moduleKey string, deps
 	return loader.Extract(laneID, moduleKey, deps)
 }
 
-func recordExtract(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput, warnings []contracts.Warning) error {
-	return checkpointExtractSucceeded(recorder, stepID, laneID, moduleKey, deps, outputs, rejected, warnings)
+func recordExtract(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput) error {
+	return checkpointExtractSucceeded(recorder, stepID, laneID, moduleKey, deps, outputs, rejected)
 }
 
 type extractJob struct {
@@ -67,7 +67,7 @@ type extractJobResult struct {
 	chunkIndex           int
 	value                erasedExtractArtifact
 	serialized           CheckpointArtifact
-	warnings             []contracts.Warning
+	diagnostics          []contracts.DiagnosticGroup
 	rejected             *contracts.RejectedOutput
 	validationIncomplete bool
 	validationSummary    *artifacts.ValidationSummary
@@ -353,7 +353,11 @@ func hydrateRequiredLane(input RunInput, loader CheckpointLoader, doc *source.So
 		return state, err
 	}
 	hydrated := resolution.artifacts[0]
-	local.Warnings = append(local.Warnings, cloneWarnings(checkpoint.Warnings)...)
+	diagnostics, diagnosticErr := promoteCheckpointDiagnostics(hydrated.Diagnostics, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageNormalize, StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Normalize.Module}, nil)
+	if diagnosticErr != nil {
+		return state, fmt.Errorf("promote reused accepted diagnostics: %w", diagnosticErr)
+	}
+	appendDiagnosticGroups(&local, diagnostics)
 	local.NormalizeOutputs = append(local.NormalizeOutputs, contracts.SerializedOutput{
 		StepID:        input.stepID,
 		LaneID:        lane.ID,
@@ -418,8 +422,14 @@ func prepareLaneExtract(input RunInput, loader CheckpointLoader, doc *source.Sou
 			}
 			state.values = append(state.values, artifact)
 			state.serialized = append(state.serialized, cloneCheckpointArtifact(stored))
+			chunk := source.Chunk{ID: stored.ChunkID, Index: stored.ChunkIndex}
+			diagnostics, diagnosticErr := promoteCheckpointDiagnostics(stored.Diagnostics, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageExtract, StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Extract.Module}, &chunk)
+			if diagnosticErr != nil {
+				return nil, fmt.Errorf("promote reused extract diagnostics: %w", diagnosticErr)
+			}
+			state.diagnostics = append(state.diagnostics, diagnostics...)
 		}
-		state.warnings, state.rejected = cloneWarnings(cp.Warnings), cloneRejectedOutputs(cp.Rejected)
+		state.rejected = cloneRejectedOutputs(cp.Rejected)
 	}
 	return state, nil
 }
@@ -450,15 +460,14 @@ func (r *Runner) runExtractJob(ctx context.Context, input RunInput, doc *source.
 			return producerAttemptOutput{}, terminal.record(nil, attemptErr)
 		}
 		artifact := erasedExtractArtifact{LaneID: lane.ID, ExtractorKey: lane.Extract.Module, SourceID: doc.ID, ChunkID: chunk.ID, ChunkIndex: chunk.Index, ChunkRef: chunk.Ref, Value: extracted.Value}
-		attemptWarnings := cloneWarnings(extracted.Warnings)
 		serializedCandidate, encodeErr := serializeCandidateArtifact(typed.codec, artifact.LaneID, artifact.ExtractorKey, artifact.SourceID, artifact.Value)
 		if encodeErr != nil {
 			attemptErr := fmt.Errorf("serialize extract candidate for lane %q chunk %q: %w", lane.ID, chunk.ID, encodeErr)
-			payload := map[string]any{"warnings": debugWarningEnvelopes(attemptWarnings)}
+			payload := map[string]any{}
 			return producerAttemptOutput{}, terminal.record(payload, attemptErr)
 		}
 		serializedCandidate.ChunkID, serializedCandidate.ChunkIndex, serializedCandidate.ChunkRef = artifact.ChunkID, artifact.ChunkIndex, artifact.ChunkRef
-		return producerAttemptOutput{Value: extractAttemptValue{artifact: artifact, serialized: serializedCandidate, terminal: &terminal}, Candidate: extracted.ModelCandidate, Warnings: attemptWarnings}, nil
+		return producerAttemptOutput{Value: extractAttemptValue{artifact: artifact, serialized: serializedCandidate, terminal: &terminal}, Candidate: extracted.ModelCandidate, Diagnostics: contracts.CloneProducerDiagnostics(extracted.Diagnostics)}, nil
 	}, func(validationCtx context.Context, output producerAttemptOutput) (validationReport, error) {
 		candidate, ok := output.Value.(extractAttemptValue)
 		if !ok {
@@ -467,7 +476,6 @@ func (r *Runner) runExtractJob(ctx context.Context, input RunInput, doc *source.
 		report, validationErr := r.validateTypedReport(validationCtx, typed.codec, typedValidationTarget{stage: StageExtract, stepID: input.stepID, laneID: lane.ID, moduleKey: lane.Extract.Module, source: doc, sourceID: doc.ID, sourceInput: chunkInputMaterial(sourceInput, chunk), sessionID: sessionID, references: operationReferenceSet(input, lane.ExtractReferences), metadata: input.Metadata, chunk: &chunk, ref: chunk.Ref, value: candidate.artifact.Value, candidate: &candidate.serialized}, state.prepared.extractValidators, candidate.terminal.envelope.Attempt, input.Debug)
 		payload := map[string]any{
 			"output":    debugCheckpointArtifact(candidate.serialized),
-			"warnings":  debugWarningEnvelopes(append(cloneWarnings(output.Warnings), report.Warnings()...)),
 			"rejection": debugRejectedOutputPtr(typedRejection(report, typedValidationTarget{stage: StageExtract, stepID: input.stepID, laneID: lane.ID, moduleKey: lane.Extract.Module, chunk: &chunk}, candidate.terminal.envelope.Attempt)),
 		}
 		if validationErr != nil {
@@ -481,6 +489,14 @@ func (r *Runner) runExtractJob(ctx context.Context, input RunInput, doc *source.
 		result.err = errors.Join(result.err, debugErr)
 		return result
 	}
+	if err == nil {
+		diagnostics, diagnosticErr := terminalDiagnosticGroups(terminalResult, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageExtract, StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Extract.Module}, &chunk)
+		if diagnosticErr != nil {
+			result.err = fmt.Errorf("promote extract diagnostics: %w", diagnosticErr)
+			return result
+		}
+		result.diagnostics = diagnostics
+	}
 	if err == nil && terminalResult.Action == producerTerminalRejected {
 		result.rejected = terminalResult.Rejection
 		if result.rejected != nil {
@@ -488,7 +504,6 @@ func (r *Runner) runExtractJob(ctx context.Context, input RunInput, doc *source.
 			result.validationSummary = &summary
 			result.rejected.Validation = cloneValidationSummaryPtr(result.validationSummary)
 		}
-		result.warnings = cloneWarnings(terminalResult.Warnings)
 		return result
 	}
 	if err == nil {
@@ -498,19 +513,19 @@ func (r *Runner) runExtractJob(ctx context.Context, input RunInput, doc *source.
 			return result
 		}
 		stored, encodeErr := checkpointArtifact(typed.codec, candidate.artifact.LaneID, candidate.artifact.ExtractorKey, candidate.artifact.SourceID, candidate.artifact.Value)
-		payload := map[string]any{"output": debugCheckpointArtifact(candidate.serialized), "warnings": debugWarningEnvelopes(terminalResult.Warnings), "rejection": debugRejectedOutputPtr(nil)}
+		payload := map[string]any{"output": debugCheckpointArtifact(candidate.serialized), "rejection": debugRejectedOutputPtr(nil)}
 		if encodeErr != nil {
 			attemptErr := fmt.Errorf("serialize accepted extract output for lane %q chunk %q: %w", lane.ID, chunk.ID, encodeErr)
 			result.err = candidate.terminal.record(payload, attemptErr)
 			return result
 		}
 		stored.ChunkID, stored.ChunkIndex, stored.ChunkRef = candidate.artifact.ChunkID, candidate.artifact.ChunkIndex, candidate.artifact.ChunkRef
+		stored.Diagnostics = terminalCheckpointDiagnostics(terminalResult)
 		if debugErr := candidate.terminal.record(payload, nil); debugErr != nil {
 			result.err = debugErr
 			return result
 		}
 		result.value, result.serialized = candidate.artifact, stored
-		result.warnings = cloneWarnings(terminalResult.Warnings)
 		result.validationIncomplete = terminalResult.ValidationIncomplete
 		result.validationSummary = &summary
 	}
@@ -531,12 +546,12 @@ func finalizeLaneExtract(checkpoints CheckpointRecorder, stepID string, state *l
 		}
 		if result.rejected != nil {
 			state.rejected = append(state.rejected, *result.rejected)
-			state.warnings = append(state.warnings, result.warnings...)
+			state.diagnostics = append(state.diagnostics, result.diagnostics...)
 			continue
 		}
 		state.values = append(state.values, result.value)
 		state.serialized = append(state.serialized, result.serialized)
-		state.warnings = append(state.warnings, result.warnings...)
+		state.diagnostics = append(state.diagnostics, result.diagnostics...)
 		if result.validationIncomplete {
 			state.incomplete = append(state.incomplete, result.chunkIndex)
 		}
@@ -549,7 +564,7 @@ func finalizeLaneExtract(checkpoints CheckpointRecorder, stepID string, state *l
 		state.reuseEligible = false
 	}
 	if !state.decision.Reused && state.reuseEligible {
-		if err := recordExtract(checkpoints, stepID, lane.ID, lane.Extract.Module, state.deps, state.serialized, state.rejected, state.warnings); err != nil {
+		if err := recordExtract(checkpoints, stepID, lane.ID, lane.Extract.Module, state.deps, state.serialized, state.rejected); err != nil {
 			return fmt.Errorf("write extract checkpoint for lane %q: %w", lane.ID, err)
 		}
 	}
@@ -562,20 +577,20 @@ func (r *Runner) continueLane(ctx context.Context, input RunInput, checkpoints C
 	results := finalizedExtractResults{
 		accepted:            state.values,
 		serialized:          state.serialized,
-		warnings:            state.warnings,
+		diagnostics:         state.diagnostics,
 		rejected:            state.rejected,
 		incomplete:          state.incomplete,
 		validationSummaries: state.validationSummaries,
 		decision:            state.decision,
 		reuseEligible:       state.reuseEligible,
 	}
-	local.Warnings = append(local.Warnings, cloneWarnings(results.warnings)...)
+	local.diagnosticGroups = append(local.diagnosticGroups, contracts.CloneDiagnosticCollection(contracts.DiagnosticCollection{Groups: results.diagnostics}).Groups...)
 	local.Rejected = append(local.Rejected, cloneRejectedOutputs(results.rejected)...)
 	local.ValidationSummaries = append(local.ValidationSummaries, cloneValidationSummaries(results.validationSummaries)...)
 	if err := writeDebugTimed(input.Debug, path.Join("extract", fileio.EncodePathComponent(lane.ID), "input.json"), debugTimedEnvelope{Stage: string(StageExtract), StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Extract.Module, StartedAt: time.Now().UTC(), Payload: map[string]any{"reused": results.decision.Reused, "decision": results.decision, "source": debugSourceDocumentEnvelope(doc), "chunks": debugSourceChunkEnvelopes(chunks), "options": redactSensitiveMap(lane.Extract.Options), "metadata": redactSensitiveMap(input.Metadata)}}); err != nil {
 		return local, &laneRunError{stage: StageExtract, err: err}
 	}
-	if err := writeDebugTimed(input.Debug, path.Join("extract", fileio.EncodePathComponent(lane.ID), "output.json"), debugTimedEnvelope{Stage: string(StageExtract), StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Extract.Module, StartedAt: time.Now().UTC(), Payload: map[string]any{"reused": results.decision.Reused, "outputs": debugCheckpointArtifacts(results.serialized), "rejected": debugRejectedOutputEnvelopes(results.rejected), "warnings": debugWarningEnvelopes(results.warnings), "validation_incomplete_chunks": append([]int(nil), results.incomplete...)}}); err != nil {
+	if err := writeDebugTimed(input.Debug, path.Join("extract", fileio.EncodePathComponent(lane.ID), "output.json"), debugTimedEnvelope{Stage: string(StageExtract), StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Extract.Module, StartedAt: time.Now().UTC(), Payload: map[string]any{"reused": results.decision.Reused, "outputs": debugCheckpointArtifacts(results.serialized), "rejected": debugRejectedOutputEnvelopes(results.rejected), "validation_incomplete_chunks": append([]int(nil), results.incomplete...)}}); err != nil {
 		return local, &laneRunError{stage: StageExtract, err: err}
 	}
 	if len(results.accepted) == 0 {
@@ -640,7 +655,7 @@ func mergeLaneOutput(dst *RunOutput, src RunOutput) error {
 	}
 	dst.NormalizeOutputs = append(dst.NormalizeOutputs, cloneSerializedOutputs(src.NormalizeOutputs)...)
 	dst.Rejected = append(dst.Rejected, cloneRejectedOutputs(src.Rejected)...)
-	dst.Warnings = append(dst.Warnings, cloneWarnings(src.Warnings)...)
+	appendDiagnosticGroups(dst, src.diagnosticGroups)
 	dst.CheckpointEvents = append(dst.CheckpointEvents, src.CheckpointEvents...)
 	dst.ValidationSummaries = append(dst.ValidationSummaries, cloneValidationSummaries(src.ValidationSummaries)...)
 	if len(src.normalizeReuseEligibility) > 0 {

internal/framework/pipeline/runner_extract_correction_test.go

@@ -4,7 +4,6 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
-	"reflect"
 	"strings"
 	"sync"
 	"testing"
@@ -85,15 +84,6 @@ func TestRunnerCorrectsExtractCandidatesIndependentlyPerChunk(t *testing.T) {
 	}
 }
 
-func containsWarnings(have, want []contracts.Warning) bool {
-	for index := 0; index+len(want) <= len(have); index++ {
-		if reflect.DeepEqual(have[index:index+len(want)], want) {
-			return true
-		}
-	}
-	return false
-}
-
 func containsValidationSummary(summaries []artifacts.ValidationSummary, status string) bool {
 	for _, summary := range summaries {
 		if summary.Status == status {
@@ -126,8 +116,8 @@ func TestRunnerContinuesValidationIncompleteExtractWithoutCheckpoint(t *testing.
 	if output.Manifest.ValidationStatus != "incomplete" || !containsValidationSummary(output.Manifest.ValidationSummaries, "incomplete") {
 		t.Fatalf("manifest validation = %#v, want incomplete extract provenance", output.Manifest)
 	}
-	if got, want := output.Warnings, []contracts.Warning{{Scope: "typed/check", ReasonCode: "validator_execution_incomplete", Message: "Validator execution did not complete within its configured budget."}}; !containsWarnings(got, want) {
-		t.Fatalf("warnings = %#v, want %#v", got, want)
+	if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].ReasonCode != "validator_execution_incomplete" {
+		t.Fatalf("diagnostics = %#v, want validation-incomplete group", output.Diagnostics)
 	}
 	if len(recorder.checkpoint.Outputs) != 0 || len(recorder.checkpoint.Rejected) != 0 {
 		t.Fatalf("extract checkpoint = %#v, want no persisted incomplete output", recorder.checkpoint)

internal/framework/pipeline/runner_extract_handoff_test.go

@@ -35,11 +35,10 @@ func (v *mutatingMetadataValidator) Validate(_ context.Context, request contract
 	return contracts.ValidationResult{Approved: true}, nil
 }
 
-func (r *extractCaptureRecorder) ExtractSucceeded(_ string, _ string, _ []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput, warnings []contracts.Warning) error {
+func (r *extractCaptureRecorder) ExtractSucceeded(_ string, _ string, _ []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput) error {
 	r.checkpoint = ExtractCheckpoint{
 		Outputs:  cloneCheckpointArtifacts(outputs),
 		Rejected: cloneRejectedOutputs(rejected),
-		Warnings: cloneWarnings(warnings),
 	}
 	return nil
 }
@@ -56,7 +55,6 @@ func (l *extractResultLoader) Extract(string, string, []CheckpointFingerprint) (
 	return ExtractCheckpoint{
 		Outputs:  cloneCheckpointArtifacts(l.checkpoint.Outputs),
 		Rejected: cloneRejectedOutputs(l.checkpoint.Rejected),
-		Warnings: cloneWarnings(l.checkpoint.Warnings),
 	}, l.decision
 }
 
@@ -156,8 +154,14 @@ func TestRunnerContinuesFromFreshAndReusedExtractResults(t *testing.T) {
 	installExtractOperation(prepared, 0, func(_ context.Context, request contracts.TypedExtractionRequest) (erasedTypedResult, error) {
 		extractCalls++
 		return erasedTypedResult{
-			Value:    typedValueForLane(0, request.Chunk.Index),
-			Warnings: []contracts.Warning{{Scope: "extract", ReasonCode: "observed", Message: "accepted extract"}},
+			Value: typedValueForLane(0, request.Chunk.Index),
+			Diagnostics: []contracts.ProducerDiagnostic{{
+				Disposition:     contracts.DiagnosticDispositionObservation,
+				Category:        contracts.DiagnosticCategoryNormalization,
+				ReasonCode:      "accepted_extract_normalized",
+				OccurrenceCount: 1,
+				Samples:         []contracts.DiagnosticSample{{Scope: "extract", Message: "accepted extract"}},
+			}},
 		}, nil
 	})
 
@@ -206,12 +210,12 @@ func TestRunnerContinuesFromFreshAndReusedExtractResults(t *testing.T) {
 	if !reflect.DeepEqual(reused.NormalizeOutputs, fresh.NormalizeOutputs) {
 		t.Fatalf("reused normalize outputs = %#v, want fresh outputs %#v", reused.NormalizeOutputs, fresh.NormalizeOutputs)
 	}
-	if !reflect.DeepEqual(reused.Warnings, fresh.Warnings) {
-		t.Fatalf("reused warnings = %#v, want fresh warnings %#v", reused.Warnings, fresh.Warnings)
+	if !reflect.DeepEqual(reused.Diagnostics, fresh.Diagnostics) {
+		t.Fatalf("reused diagnostics = %#v, want fresh diagnostics %#v", reused.Diagnostics, fresh.Diagnostics)
 	}
 }
 
-func TestRunnerPromotesOnlyAcceptedExtractRetryWarnings(t *testing.T) {
+func TestRunnerPromotesOnlyAcceptedExtractRetryDiagnostics(t *testing.T) {
 	prepared := preparedAttemptDebugPipeline(t)
 	prepared.Steps[0].lanes[0].resolved.Extract.Retries = 1
 	attempts := 0
@@ -223,7 +227,7 @@ func TestRunnerPromotesOnlyAcceptedExtractRetryWarnings(t *testing.T) {
 		}
 		return erasedTypedResult{
 			Value:          typedValueForLane(0, request.Chunk.Index),
-			Warnings:       []contracts.Warning{{Scope: scope, ReasonCode: "observed", Message: scope}},
+			Diagnostics:    []contracts.ProducerDiagnostic{producerDiagnostic(scope, scope)},
 			ModelCandidate: attemptCandidate(t, fmt.Sprintf(`{"items":["%s"]}`, scope)),
 		}, nil
 	})
@@ -240,8 +244,8 @@ func TestRunnerPromotesOnlyAcceptedExtractRetryWarnings(t *testing.T) {
 	if attempts != 2 {
 		t.Fatalf("extract attempts = %d, want 2", attempts)
 	}
-	if len(output.Warnings) != 1 || output.Warnings[0].Scope != "accepted" {
-		t.Fatalf("promoted warnings = %#v, want accepted attempt only", output.Warnings)
+	if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].Samples[0].Scope != "accepted" {
+		t.Fatalf("promoted diagnostics = %#v, want accepted attempt only", output.Diagnostics)
 	}
 	assertAttemptEnvelopeSequence(t, debug, "extract/notes/chunk-000001", 1, 2)
 	first := debug.envelope(t, "extract/notes/chunk-000001/attempt-01.json")

internal/framework/pipeline/runner_normalize_retry_test.go

@@ -17,7 +17,7 @@ func TestRunnerHandlesRetryableNormalizeFallbacks(t *testing.T) {
 		validator      *preparedValidator
 		wantCalls      int
 		wantItem       string
-		wantWarnings   []string
+		wantMessages   []string
 		wantRejected   int
 		wantDebug      []string
 		wantCheckpoint int
@@ -30,7 +30,7 @@ func TestRunnerHandlesRetryableNormalizeFallbacks(t *testing.T) {
 			},
 			wantCalls:      1,
 			wantItem:       "fallback",
-			wantWarnings:   []string{"ordinary", "fallback-warning"},
+			wantMessages:   []string{"ordinary", "fallback-warning"},
 			wantDebug:      []string{`"another_attempt":false`, `"fallback_accepted":true`},
 			wantCheckpoint: 1,
 		},
@@ -41,11 +41,11 @@ func TestRunnerHandlesRetryableNormalizeFallbacks(t *testing.T) {
 				if attempt == 1 {
 					return retryableNormalizeResult("discarded", "discarded-ordinary", "discarded-fallback")
 				}
-				return erasedTypedResult{Value: codecNotes{Items: []string{"accepted"}}, Warnings: []contracts.Warning{{Scope: "accepted", ReasonCode: "ordinary", Message: "accepted-warning"}}}
+				return erasedTypedResult{Value: codecNotes{Items: []string{"accepted"}}, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("accepted", "accepted-warning")}}
 			},
 			wantCalls:      2,
 			wantItem:       "accepted",
-			wantWarnings:   []string{"accepted-warning"},
+			wantMessages:   []string{"accepted-warning"},
 			wantDebug:      []string{`"another_attempt":true`, `"fallback_accepted":false`},
 			wantCheckpoint: 1,
 		},
@@ -57,7 +57,7 @@ func TestRunnerHandlesRetryableNormalizeFallbacks(t *testing.T) {
 			},
 			wantCalls:      2,
 			wantItem:       "fallback-2",
-			wantWarnings:   []string{"ordinary-2", "fallback-warning-2"},
+			wantMessages:   []string{"ordinary-2", "fallback-warning-2"},
 			wantDebug:      []string{`"another_attempt":false`, `"fallback_accepted":true`},
 			wantCheckpoint: 1,
 		},
@@ -140,17 +140,73 @@ func TestRunnerHandlesRetryableNormalizeFallbacks(t *testing.T) {
 			if got := firstNote(normalized); got != tc.wantItem {
 				t.Fatalf("normalized item = %q, want %q", got, tc.wantItem)
 			}
-			gotWarnings := make([]string, len(output.Warnings))
-			for index, warning := range output.Warnings {
-				gotWarnings[index] = warning.Message
+			var gotMessages []string
+			for _, group := range output.Diagnostics.Groups {
+				for _, sample := range group.Samples {
+					gotMessages = append(gotMessages, sample.Message)
+				}
 			}
-			if strings.Join(gotWarnings, "|") != strings.Join(tc.wantWarnings, "|") {
-				t.Fatalf("durable warnings = %#v, want %#v", gotWarnings, tc.wantWarnings)
+			if strings.Join(gotMessages, "|") != strings.Join(tc.wantMessages, "|") {
+				t.Fatalf("durable diagnostics = %#v, want %#v", gotMessages, tc.wantMessages)
 			}
 		})
 	}
 }
 
+func TestRunnerForwardsModuleRequestedNormalizeCorrection(t *testing.T) {
+	const (
+		defective = `{"duplicate_groups":[{"candidate_ids":[1,99],"canonical_candidate_id":1}]}`
+		guidance  = "Duplicate group 1 must use only supplied candidate IDs. Return one complete corrected response."
+	)
+	prepared := preparedAttemptDebugPipeline(t)
+	lane := &prepared.Steps[0].lanes[0]
+	lane.resolved.Normalize.Retries = 1
+	var observed *contracts.SemanticCorrection
+	calls := 0
+	lane.typed.normalize = func(_ context.Context, _ any, request contracts.TypedNormalizeRequest[any]) (erasedTypedResult, error) {
+		calls++
+		if request.Correction != nil {
+			clone, err := contracts.CloneSemanticCorrection(request.Correction)
+			if err != nil {
+				return erasedTypedResult{}, err
+			}
+			observed = clone
+		}
+		if calls == 1 {
+			return erasedTypedResult{
+				Value:          codecNotes{Items: []string{"safe fallback"}},
+				ModelCandidate: attemptCandidate(t, defective),
+				Retry: &contracts.NormalizeRetry{
+					ReasonCode:         "semantic_proposal_invalid",
+					Message:            "operator-facing proposal diagnostic",
+					CorrectionGuidance: guidance,
+				},
+			}, nil
+		}
+		return erasedTypedResult{Value: codecNotes{Items: []string{"corrected"}}}, nil
+	}
+	debug := newCapturedDebugRecorder()
+	output, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), Debug: debug})
+	if err != nil {
+		t.Fatalf("Run() error = %v", err)
+	}
+	if calls != 2 || observed == nil || string(observed.AssistantResponse) != defective || observed.UserGuidance != guidance {
+		t.Fatalf("normalize calls = %d correction = %#v", calls, observed)
+	}
+	if len(output.NormalizeOutputs) != 1 || len(output.Rejected) != 0 {
+		t.Fatalf("run output = %#v, want corrected accepted output", output)
+	}
+	var retryDebug strings.Builder
+	for _, name := range debug.names() {
+		if strings.HasPrefix(name, "normalize/notes/attempt-") && strings.HasSuffix(name, ".json") {
+			retryDebug.Write(debug.json[name])
+		}
+	}
+	if !strings.Contains(retryDebug.String(), `"correction_available":true`) || strings.Contains(retryDebug.String(), guidance) || strings.Contains(retryDebug.String(), defective) {
+		t.Fatalf("retry debug = %s, want safe correction metadata without content", retryDebug.String())
+	}
+}
+
 func TestRunnerValidatesNormalizeRetryDiagnostics(t *testing.T) {
 	const (
 		reasonSentinel  = "reason-diagnostic-sentinel"
@@ -158,9 +214,11 @@ func TestRunnerValidatesNormalizeRetryDiagnostics(t *testing.T) {
 	)
 	reasonOverLimit := strings.Repeat("r", contracts.MaxNormalizeRetryReasonCodeBytes-len(reasonSentinel)) + reasonSentinel + "x"
 	messageOverLimit := strings.Repeat("m", contracts.MaxNormalizeRetryMessageBytes-len(messageSentinel)) + messageSentinel + "x"
+	guidanceOverLimit := strings.Repeat("g", contracts.MaxNormalizeRetryCorrectionGuidanceBytes-len(messageSentinel)) + messageSentinel + "x"
 	tests := []struct {
 		name         string
 		retry        contracts.NormalizeRetry
+		candidate    *contracts.ModelCandidate
 		wantError    string
 		hiddenValues []string
 	}{
@@ -171,6 +229,15 @@ func TestRunnerValidatesNormalizeRetryDiagnostics(t *testing.T) {
 				Message:    strings.Repeat("m", contracts.MaxNormalizeRetryMessageBytes),
 			},
 		},
+		{
+			name: "accepts correction guidance byte limit with candidate",
+			retry: contracts.NormalizeRetry{
+				ReasonCode:         reasonSentinel,
+				Message:            messageSentinel,
+				CorrectionGuidance: strings.Repeat("g", contracts.MaxNormalizeRetryCorrectionGuidanceBytes),
+			},
+			candidate: attemptCandidate(t, `{"duplicate_groups":[]}`),
+		},
 		{
 			name: "rejects oversized reason code",
 			retry: contracts.NormalizeRetry{
@@ -207,6 +274,46 @@ func TestRunnerValidatesNormalizeRetryDiagnostics(t *testing.T) {
 			wantError:    "message has invalid UTF-8",
 			hiddenValues: []string{reasonSentinel, messageSentinel},
 		},
+		{
+			name: "rejects oversized correction guidance",
+			retry: contracts.NormalizeRetry{
+				ReasonCode:         reasonSentinel,
+				Message:            messageSentinel,
+				CorrectionGuidance: guidanceOverLimit,
+			},
+			wantError:    "correction guidance exceeds maximum length",
+			hiddenValues: []string{reasonSentinel, messageSentinel},
+		},
+		{
+			name: "rejects invalid correction guidance UTF-8",
+			retry: contracts.NormalizeRetry{
+				ReasonCode:         reasonSentinel,
+				Message:            messageSentinel,
+				CorrectionGuidance: messageSentinel + string([]byte{0xff}),
+			},
+			wantError:    "correction guidance has invalid UTF-8",
+			hiddenValues: []string{reasonSentinel, messageSentinel},
+		},
+		{
+			name: "rejects blank correction guidance",
+			retry: contracts.NormalizeRetry{
+				ReasonCode:         reasonSentinel,
+				Message:            messageSentinel,
+				CorrectionGuidance: " \t\n ",
+			},
+			wantError:    "correction guidance is blank",
+			hiddenValues: []string{reasonSentinel, messageSentinel},
+		},
+		{
+			name: "rejects correction guidance without candidate",
+			retry: contracts.NormalizeRetry{
+				ReasonCode:         reasonSentinel,
+				Message:            messageSentinel,
+				CorrectionGuidance: "Return one complete corrected response.",
+			},
+			wantError:    "requires a model candidate",
+			hiddenValues: []string{reasonSentinel, messageSentinel},
+		},
 		{
 			name: "rejects blank reason code",
 			retry: contracts.NormalizeRetry{
@@ -231,7 +338,7 @@ func TestRunnerValidatesNormalizeRetryDiagnostics(t *testing.T) {
 		t.Run(tc.name, func(t *testing.T) {
 			prepared := preparedAttemptDebugPipeline(t)
 			prepared.Steps[0].lanes[0].typed.normalize = func(context.Context, any, contracts.TypedNormalizeRequest[any]) (erasedTypedResult, error) {
-				return erasedTypedResult{Value: codecNotes{Items: []string{"safe"}}, Retry: &tc.retry}, nil
+				return erasedTypedResult{Value: codecNotes{Items: []string{"safe"}}, Retry: &tc.retry, ModelCandidate: tc.candidate}, nil
 			}
 			debug := newCapturedDebugRecorder()
 			output, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), Debug: debug})
@@ -262,12 +369,12 @@ func TestRunnerValidatesNormalizeRetryDiagnostics(t *testing.T) {
 
 func retryableNormalizeResult(item, ordinary, fallback string) erasedTypedResult {
 	return erasedTypedResult{
-		Value:    codecNotes{Items: []string{item}},
-		Warnings: []contracts.Warning{{Scope: "attempt", ReasonCode: "ordinary", Message: ordinary}},
+		Value:       codecNotes{Items: []string{item}},
+		Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("attempt", ordinary)},
 		Retry: &contracts.NormalizeRetry{
-			ReasonCode:       "retryable_normalization",
-			Message:          "safe fallback is available",
-			FallbackWarnings: []contracts.Warning{{Scope: "fallback", ReasonCode: "fallback", Message: fallback}},
+			ReasonCode:          "retryable_normalization",
+			Message:             "safe fallback is available",
+			FallbackDiagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("fallback", fallback)},
 		},
 	}
 }

internal/framework/pipeline/runner_rejection_warnings_test.go

@@ -11,19 +11,19 @@ import (
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
 )
 
-type warningChunker struct {
+type diagnosticChunker struct {
 	key   string
 	plan  source.ChunkPlan
 	calls int
 }
 
-func (c *warningChunker) Key() string                             { return c.key }
-func (*warningChunker) ReferenceSlots() []contracts.ReferenceSlot { return nil }
-func (c *warningChunker) Plan(context.Context, contracts.ChunkRequest) (contracts.ChunkPlanResult, error) {
+func (c *diagnosticChunker) Key() string                             { return c.key }
+func (*diagnosticChunker) ReferenceSlots() []contracts.ReferenceSlot { return nil }
+func (c *diagnosticChunker) Plan(context.Context, contracts.ChunkRequest) (contracts.ChunkPlanResult, error) {
 	c.calls++
 	return contracts.ChunkPlanResult{
-		Plan:     source.CloneChunkPlan(c.plan),
-		Warnings: []contracts.Warning{{Scope: fmt.Sprintf("operation-%d", c.calls), ReasonCode: "operation", Message: "operation warning"}},
+		Plan:        source.CloneChunkPlan(c.plan),
+		Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic(fmt.Sprintf("operation-%d", c.calls), "operation diagnostic")},
 	}, nil
 }
 
@@ -40,14 +40,14 @@ func (v chunkValidationFunc) Validate(ctx context.Context, request contracts.Chu
 	return v.validate(ctx, request)
 }
 
-func TestRunnerPromotesOnlyTerminalRejectionWarnings(t *testing.T) {
+func TestRunnerPromotesOnlyTerminalRejectionDiagnostics(t *testing.T) {
 	for _, target := range []ModuleStage{StageChunk, StageExtract, StageMerge, StageNormalize} {
 		t.Run(string(target), func(t *testing.T) {
 			prepared := preparedAttemptDebugPipeline(t)
 			lane := &prepared.Steps[0].lanes[0]
 			attempts := 0
 			first := func() contracts.ValidationResult {
-				return contracts.ValidationResult{Approved: true, Warnings: []contracts.Warning{{Scope: fmt.Sprintf("validator-%d", attempts), ReasonCode: "validator", Message: "validator warning"}}}
+				return contracts.ValidationResult{Approved: true, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic(fmt.Sprintf("validator-%d", attempts), "validator diagnostic")}}
 			}
 			reject := func() contracts.ValidationResult {
 				return contracts.ValidationResult{Approved: false, ReasonCode: "rejected", Message: "rejected", CorrectionGuidance: "return an acceptable candidate"}
@@ -59,7 +59,7 @@ func TestRunnerPromotesOnlyTerminalRejectionWarnings(t *testing.T) {
 			case StageChunk:
 				prepared.resolved.ChunkValidationPolicy.SemanticRejection = SemanticRejectionRejectOutput
 				chunker := prepared.chunker.(*typedTestChunker)
-				prepared.chunker = &warningChunker{key: prepared.resolved.Chunk.Module, plan: source.CloneChunkPlan(chunker.plan)}
+				prepared.chunker = &diagnosticChunker{key: prepared.resolved.Chunk.Module, plan: source.CloneChunkPlan(chunker.plan)}
 				prepared.resolved.Chunk.Retries = 1
 				prepared.chunkValidators.validators = []preparedValidator{
 					{resolved: ResolvedValidator{Binding: Binding("warning-approval"), Target: ValidatorTargetChunk}, chunk: chunkValidationFunc{name: "warning-approval", validate: func(context.Context, contracts.ChunkValidationRequest) (contracts.ValidationResult, error) {
@@ -69,34 +69,34 @@ func TestRunnerPromotesOnlyTerminalRejectionWarnings(t *testing.T) {
 						return reject(), nil
 					}}},
 				}
-				chunkerWithWarnings := prepared.chunker.(*warningChunker)
+				chunkerWithDiagnostics := prepared.chunker.(*diagnosticChunker)
 				first = func() contracts.ValidationResult {
-					return contracts.ValidationResult{Approved: true, Warnings: []contracts.Warning{{Scope: fmt.Sprintf("validator-%d", chunkerWithWarnings.calls), ReasonCode: "validator", Message: "validator warning"}}}
+					return contracts.ValidationResult{Approved: true, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic(fmt.Sprintf("validator-%d", chunkerWithDiagnostics.calls), "validator diagnostic")}}
 				}
 			case StageExtract:
 				lane.resolved.Extract.Retries = 1
 				lane.resolved.ExtractValidationPolicy.SemanticRejection = SemanticRejectionRejectOutput
 				installExtractOperation(prepared, 0, func(context.Context, contracts.TypedExtractionRequest) (erasedTypedResult, error) {
 					attempts++
-					return erasedTypedResult{Value: codecNotes{Items: []string{"extract"}}, Warnings: []contracts.Warning{{Scope: fmt.Sprintf("operation-%d", attempts), ReasonCode: "operation", Message: "operation warning"}}, ModelCandidate: attemptCandidate(t, fmt.Sprintf(`{"items":["%d"]}`, attempts))}, nil
+					return erasedTypedResult{Value: codecNotes{Items: []string{"extract"}}, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic(fmt.Sprintf("operation-%d", attempts), "operation diagnostic")}, ModelCandidate: attemptCandidate(t, fmt.Sprintf(`{"items":["%d"]}`, attempts))}, nil
 				})
-				lane.extractValidators.validators = rejectionWarningTypedValidators(first, reject)
+				lane.extractValidators.validators = rejectionDiagnosticTypedValidators(first, reject)
 			case StageMerge:
 				lane.resolved.Merge.Retries = 1
 				lane.resolved.MergeValidationPolicy.SemanticRejection = SemanticRejectionRejectOutput
 				lane.typed.merge = func(context.Context, any, contracts.TypedMergeRequest[any]) (erasedTypedResult, error) {
 					attempts++
-					return erasedTypedResult{Value: codecNotes{Items: []string{"merge"}}, Warnings: []contracts.Warning{{Scope: fmt.Sprintf("operation-%d", attempts), ReasonCode: "operation", Message: "operation warning"}}, ModelCandidate: attemptCandidate(t, fmt.Sprintf(`{"items":["%d"]}`, attempts))}, nil
+					return erasedTypedResult{Value: codecNotes{Items: []string{"merge"}}, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic(fmt.Sprintf("operation-%d", attempts), "operation diagnostic")}, ModelCandidate: attemptCandidate(t, fmt.Sprintf(`{"items":["%d"]}`, attempts))}, nil
 				}
-				lane.mergeValidators.validators = rejectionWarningTypedValidators(first, reject)
+				lane.mergeValidators.validators = rejectionDiagnosticTypedValidators(first, reject)
 			case StageNormalize:
 				lane.resolved.Normalize.Retries = 1
 				lane.resolved.NormalizeValidationPolicy.SemanticRejection = SemanticRejectionRejectOutput
 				lane.typed.normalize = func(context.Context, any, contracts.TypedNormalizeRequest[any]) (erasedTypedResult, error) {
 					attempts++
-					return erasedTypedResult{Value: codecNotes{Items: []string{"normalize"}}, Warnings: []contracts.Warning{{Scope: fmt.Sprintf("operation-%d", attempts), ReasonCode: "operation", Message: "operation warning"}}, ModelCandidate: attemptCandidate(t, fmt.Sprintf(`{"items":["%d"]}`, attempts))}, nil
+					return erasedTypedResult{Value: codecNotes{Items: []string{"normalize"}}, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic(fmt.Sprintf("operation-%d", attempts), "operation diagnostic")}, ModelCandidate: attemptCandidate(t, fmt.Sprintf(`{"items":["%d"]}`, attempts))}, nil
 				}
-				lane.normalizeValidators.validators = rejectionWarningTypedValidators(first, reject)
+				lane.normalizeValidators.validators = rejectionDiagnosticTypedValidators(first, reject)
 			}
 
 			output, err := New().Run(context.Background(), RunInput{Prepared: prepared, RawInput: []byte("input"), Checkpoints: recorder, Debug: debug})
@@ -109,29 +109,26 @@ func TestRunnerPromotesOnlyTerminalRejectionWarnings(t *testing.T) {
 				wantScopes = []string{"operation-1", "validator-1"}
 				wantAttempts = 1
 			}
-			if got := rejectionWarningScopes(output.Warnings); !reflect.DeepEqual(got, wantScopes) {
-				t.Fatalf("published warning scopes = %#v, want %#v", got, wantScopes)
+			if got := rejectionDiagnosticScopes(output.Diagnostics.Groups); !reflect.DeepEqual(got, wantScopes) {
+				t.Fatalf("published diagnostic scopes = %#v, want %#v", got, wantScopes)
 			}
 			if len(output.Rejected) != 1 || output.Rejected[0].AttemptCount != wantAttempts {
 				t.Fatalf("rejections = %#v, want terminal rejection after %d attempt(s)", output.Rejected, wantAttempts)
 			}
-			if target == StageExtract && !reflect.DeepEqual(rejectionWarningScopes(recorder.checkpoint.Warnings), wantScopes) {
-				t.Fatalf("extract checkpoint warnings = %#v, want %#v", recorder.checkpoint.Warnings, wantScopes)
-			}
 			attemptPath := fmt.Sprintf("%s/notes/attempt-01.json", target)
 			if target == StageChunk {
 				attemptPath = "chunk/attempt-01.json"
 			} else if target == StageExtract {
 				attemptPath = "extract/notes/chunk-000001/attempt-01.json"
 			}
-			if !strings.Contains(string(debug.json[attemptPath]), "operation-1") || !strings.Contains(string(debug.json[attemptPath]), "validator-1") {
-				t.Fatalf("first attempt debug = %s, want discarded warnings", debug.json[attemptPath])
+			if !strings.Contains(string(debug.json[attemptPath]), "rejection") {
+				t.Fatalf("first attempt debug = %s, want rejection", debug.json[attemptPath])
 			}
 		})
 	}
 }
 
-func rejectionWarningTypedValidators(first func() contracts.ValidationResult, reject func() contracts.ValidationResult) []preparedValidator {
+func rejectionDiagnosticTypedValidators(first func() contracts.ValidationResult, reject func() contracts.ValidationResult) []preparedValidator {
 	return []preparedValidator{
 		{resolved: ResolvedValidator{Binding: Binding("warning-approval"), Target: ValidatorTargetTyped, ArtifactKind: "test/notes"}, typedValidate: func(context.Context, any, typedValidationTarget) (contracts.ValidationResult, error) {
 			return first(), nil
@@ -142,10 +139,10 @@ func rejectionWarningTypedValidators(first func() contracts.ValidationResult, re
 	}
 }
 
-func rejectionWarningScopes(warnings []contracts.Warning) []string {
-	scopes := make([]string, len(warnings))
-	for index := range warnings {
-		scopes[index] = warnings[index].Scope
+func rejectionDiagnosticScopes(groups []contracts.DiagnosticGroup) []string {
+	scopes := make([]string, len(groups))
+	for index := range groups {
+		scopes[index] = groups[index].Samples[0].Scope
 	}
 	return scopes
 }

internal/framework/pipeline/runner_reuse_lineage_test.go

@@ -61,7 +61,7 @@ func (s *reuseLineageCheckpointSpy) ExtractRunning(laneID, _ string, _ []Checkpo
 	return nil
 }
 
-func (s *reuseLineageCheckpointSpy) ExtractSucceeded(laneID, _ string, _ []CheckpointFingerprint, _ []CheckpointArtifact, _ []contracts.RejectedOutput, _ []contracts.Warning) error {
+func (s *reuseLineageCheckpointSpy) ExtractSucceeded(laneID, _ string, _ []CheckpointFingerprint, _ []CheckpointArtifact, _ []contracts.RejectedOutput) error {
 	s.write("extract", "succeeded", laneID)
 	return nil
 }
@@ -76,7 +76,7 @@ func (s *reuseLineageCheckpointSpy) MergeRunning(laneID, _ string, _ []Checkpoin
 	return nil
 }
 
-func (s *reuseLineageCheckpointSpy) MergeSucceeded(laneID, _ string, _ []CheckpointFingerprint, _ CheckpointArtifact, _ []contracts.Warning) error {
+func (s *reuseLineageCheckpointSpy) MergeSucceeded(laneID, _ string, _ []CheckpointFingerprint, _ CheckpointArtifact) error {
 	s.write("merge", "succeeded", laneID)
 	return nil
 }
@@ -96,7 +96,7 @@ func (s *reuseLineageCheckpointSpy) NormalizeRunning(laneID, _ string, _ []Check
 	return nil
 }
 
-func (s *reuseLineageCheckpointSpy) NormalizeSucceeded(laneID, _ string, _ []CheckpointFingerprint, _ CheckpointArtifact, _ []contracts.Warning) error {
+func (s *reuseLineageCheckpointSpy) NormalizeSucceeded(laneID, _ string, _ []CheckpointFingerprint, _ CheckpointArtifact) error {
 	s.write("normalize", "succeeded", laneID)
 	return nil
 }

internal/framework/pipeline/runner_semantic_correction_test.go

@@ -120,7 +120,7 @@ func TestRunnerContinuesNormalizeAfterValidatorFailure(t *testing.T) {
 	if output.Manifest.ValidationStatus != "incomplete" || !containsValidationSummary(output.Manifest.ValidationSummaries, "incomplete") {
 		t.Fatalf("manifest validation = %#v, want incomplete normalization provenance", output.Manifest)
 	}
-	if got, want := output.Warnings, []contracts.Warning{{Scope: "unavailable", ReasonCode: "validator_execution_incomplete", Message: "Validator execution did not complete within its configured budget."}}; !containsWarnings(got, want) {
-		t.Fatalf("warnings = %#v, want %#v", got, want)
+	if len(output.Diagnostics.Groups) != 1 || output.Diagnostics.Groups[0].ReasonCode != "validator_execution_incomplete" {
+		t.Fatalf("diagnostics = %#v, want validation-incomplete group", output.Diagnostics)
 	}
 }

internal/framework/pipeline/runner_terminal_debug_test.go

@@ -14,11 +14,11 @@ import (
 )
 
 type terminalChunker struct {
-	key      string
-	plan     source.ChunkPlan
-	warnings []contracts.Warning
-	err      error
-	calls    *int
+	key         string
+	plan        source.ChunkPlan
+	diagnostics []contracts.ProducerDiagnostic
+	err         error
+	calls       *int
 }
 
 func (c terminalChunker) Key() string { return c.key }
@@ -29,7 +29,7 @@ func (c terminalChunker) Plan(context.Context, contracts.ChunkRequest) (contract
 	if c.calls != nil {
 		(*c.calls)++
 	}
-	return contracts.ChunkPlanResult{Plan: source.CloneChunkPlan(c.plan), Warnings: cloneWarnings(c.warnings)}, c.err
+	return contracts.ChunkPlanResult{Plan: source.CloneChunkPlan(c.plan), Diagnostics: contracts.CloneProducerDiagnostics(c.diagnostics)}, c.err
 }
 
 type terminalChunkValidator struct {
@@ -167,7 +167,7 @@ func TestRunnerRecordsChunkTerminalOutcomes(t *testing.T) {
 	for _, tc := range tests {
 		t.Run(tc.name, func(t *testing.T) {
 			prepared, plan := preparedTerminalDebugPipeline(t)
-			prepared.chunker = terminalChunker{key: prepared.resolved.Chunk.Module, plan: plan, warnings: []contracts.Warning{{Scope: "chunk", ReasonCode: "observed", Message: "chunk warning"}}, err: tc.moduleError}
+			prepared.chunker = terminalChunker{key: prepared.resolved.Chunk.Module, plan: plan, diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("chunk", "chunk diagnostic")}, err: tc.moduleError}
 			prepared.chunkValidators.validators = []preparedValidator{{resolved: ResolvedValidator{Binding: Binding("terminal/chunk-validator"), Target: ValidatorTargetChunk}, chunk: tc.validator}}
 			debug := newCapturedDebugRecorder()
 
@@ -336,7 +336,7 @@ func TestRunnerRecordsExtractTerminalOutcomes(t *testing.T) {
 				if tc.moduleError != nil {
 					return erasedTypedResult{}, tc.moduleError
 				}
-				return erasedTypedResult{Value: codecNotes{Items: []string{value}}, Warnings: []contracts.Warning{{Scope: "extract", ReasonCode: "observed", Message: "extract warning"}}}, nil
+				return erasedTypedResult{Value: codecNotes{Items: []string{value}}, Diagnostics: []contracts.ProducerDiagnostic{producerDiagnostic("extract", "extract diagnostic")}}, nil
 			})
 			prepared.Steps[0].lanes[0].extractValidators.validators = []preparedValidator{{
 				resolved: ResolvedValidator{Binding: Binding("terminal/extract-validator"), Target: ValidatorTargetTyped, ArtifactKind: "test/notes"},
@@ -368,9 +368,6 @@ func TestRunnerRecordsExtractTerminalOutcomes(t *testing.T) {
 					t.Fatalf("extract rejection = envelope %#v, outputs %#v", envelope, output.Rejected)
 				}
 			}
-			if tc.moduleError == nil && !strings.Contains(string(debug.json[attemptPath]), "extract warning") {
-				t.Fatalf("attempt envelope = %s, want attempt warning", debug.json[attemptPath])
-			}
 		})
 	}
 }

internal/framework/pipeline/runner_typed.go

@@ -29,14 +29,15 @@ func loadNormalize(loader CheckpointLoader, stepID, laneID, moduleKey string, de
 	}
 	return loader.Normalize(laneID, moduleKey, deps)
 }
-func recordMerge(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error {
-	return checkpointMergeSucceeded(recorder, stepID, laneID, moduleKey, deps, output, warnings)
+func recordMerge(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, output CheckpointArtifact) error {
+	return checkpointMergeSucceeded(recorder, stepID, laneID, moduleKey, deps, output)
 }
-func recordNormalize(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error {
-	return checkpointNormalizeSucceeded(recorder, stepID, laneID, moduleKey, deps, output, warnings)
+func recordNormalize(recorder CheckpointRecorder, stepID, laneID, moduleKey string, deps []CheckpointFingerprint, output CheckpointArtifact) error {
+	return checkpointNormalizeSucceeded(recorder, stepID, laneID, moduleKey, deps, output)
 }
 func cloneCheckpointArtifact(output CheckpointArtifact) CheckpointArtifact {
 	output.Artifact = contracts.CloneSerializedArtifact(output.Artifact)
+	output.Diagnostics = cloneCheckpointDiagnostics(output.Diagnostics)
 	return output
 }
 
@@ -264,12 +265,14 @@ func (r *Runner) runMergeStage(ctx context.Context, input RunInput, checkpoints
 	}
 	var merged erasedMergeArtifact
 	var serializedMerge CheckpointArtifact
-	var mergeWarnings []contracts.Warning
 	if mergeDecision.Reused {
 		merged = erasedMergeArtifact{LaneID: lane.ID, MergerKey: lane.Merge.Module, SourceID: doc.ID, Value: mergeResolution.values[0]}
 		serializedMerge = mergeResolution.artifacts[0]
-		mergeWarnings = cloneWarnings(mergeCP.Warnings)
-		output.Warnings = append(output.Warnings, mergeWarnings...)
+		diagnostics, diagnosticErr := promoteCheckpointDiagnostics(serializedMerge.Diagnostics, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageMerge, StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Merge.Module}, nil)
+		if diagnosticErr != nil {
+			return stageResult, fmt.Errorf("promote reused merge diagnostics: %w", diagnosticErr)
+		}
+		appendDiagnosticGroups(output, diagnostics)
 	} else {
 		if stageResult.reuseEligible {
 			if err := checkpointMergeRunning(checkpoints, input.stepID, lane.ID, lane.Merge.Module, mergeDeps); err != nil {
@@ -290,12 +293,11 @@ func (r *Runner) runMergeStage(ctx context.Context, input RunInput, checkpoints
 				return producerAttemptOutput{}, terminal.record(nil, fmt.Errorf("merge lane %q with merger %q: %w", lane.ID, lane.Merge.Module, callErr))
 			}
 			candidate := erasedMergeArtifact{LaneID: lane.ID, MergerKey: lane.Merge.Module, SourceID: doc.ID, Value: result.Value}
-			warnings := cloneWarnings(result.Warnings)
 			serializedCandidate, encodeErr := serializeCandidateArtifact(typed.codec, candidate.LaneID, candidate.MergerKey, candidate.SourceID, candidate.Value)
 			if encodeErr != nil {
-				return producerAttemptOutput{}, terminal.record(map[string]any{"warnings": debugWarningEnvelopes(warnings)}, fmt.Errorf("serialize merge candidate for lane %q: %w", lane.ID, encodeErr))
+				return producerAttemptOutput{}, terminal.record(nil, fmt.Errorf("serialize merge candidate for lane %q: %w", lane.ID, encodeErr))
 			}
-			return producerAttemptOutput{Value: mergeAttemptValue{artifact: candidate, candidate: serializedCandidate, terminal: &terminal}, Candidate: result.ModelCandidate, Warnings: warnings}, nil
+			return producerAttemptOutput{Value: mergeAttemptValue{artifact: candidate, candidate: serializedCandidate, terminal: &terminal}, Candidate: result.ModelCandidate, Diagnostics: contracts.CloneProducerDiagnostics(result.Diagnostics)}, nil
 		}, func(validationCtx context.Context, output producerAttemptOutput) (validationReport, error) {
 			candidate, ok := output.Value.(mergeAttemptValue)
 			if !ok {
@@ -303,7 +305,7 @@ func (r *Runner) runMergeStage(ctx context.Context, input RunInput, checkpoints
 			}
 			target := typedValidationTarget{stage: StageMerge, stepID: input.stepID, laneID: lane.ID, moduleKey: lane.Merge.Module, source: doc, sourceID: doc.ID, sourceInput: sourceInput.Clone(), sessionID: sessionID, references: mergeReferences, metadata: input.Metadata, value: candidate.artifact.Value, candidate: &candidate.candidate}
 			report, validationErr := r.validateTypedReport(validationCtx, typed.codec, target, prepared.mergeValidators, candidate.terminal.envelope.Attempt, input.Debug)
-			payload := map[string]any{"output": debugCheckpointArtifact(candidate.candidate), "warnings": debugWarningEnvelopes(append(cloneWarnings(output.Warnings), report.Warnings()...)), "rejection": debugRejectedOutputPtr(typedRejection(report, target, candidate.terminal.envelope.Attempt))}
+			payload := map[string]any{"output": debugCheckpointArtifact(candidate.candidate), "rejection": debugRejectedOutputPtr(typedRejection(report, target, candidate.terminal.envelope.Attempt))}
 			if validationErr != nil {
 				return report, candidate.terminal.record(payload, validationErr)
 			}
@@ -322,6 +324,10 @@ func (r *Runner) runMergeStage(ctx context.Context, input RunInput, checkpoints
 			}
 			return stageResult, runErr
 		}
+		terminalDiagnostics, diagnosticErr := terminalDiagnosticGroups(terminalResult, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageMerge, StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Merge.Module}, nil)
+		if diagnosticErr != nil {
+			return stageResult, fmt.Errorf("promote merge diagnostics: %w", diagnosticErr)
+		}
 		if terminalResult.Action == producerTerminalRejected {
 			rejected := terminalResult.Rejection
 			if rejected == nil {
@@ -330,7 +336,7 @@ func (r *Runner) runMergeStage(ctx context.Context, input RunInput, checkpoints
 			rejected.Stage, rejected.StepID, rejected.LaneID, rejected.ModuleKey = string(StageMerge), input.stepID, lane.ID, lane.Merge.Module
 			rejected.Validation = &terminalSummary
 			output.ValidationSummaries = append(output.ValidationSummaries, artifacts.CloneValidationSummary(terminalSummary))
-			output.Warnings = append(output.Warnings, terminalResult.Warnings...)
+			appendDiagnosticGroups(output, terminalDiagnostics)
 			output.Rejected = append(output.Rejected, *rejected)
 			if stageResult.reuseEligible {
 				if err := checkpointMergeRejected(checkpoints, input.stepID, lane.ID, lane.Merge.Module, mergeDeps, *rejected); err != nil {
@@ -345,7 +351,7 @@ func (r *Runner) runMergeStage(ctx context.Context, input RunInput, checkpoints
 			return stageResult, fmt.Errorf("merge attempt terminal has incompatible value")
 		}
 		stored, encodeErr := checkpointArtifact(typed.codec, candidate.artifact.LaneID, candidate.artifact.MergerKey, candidate.artifact.SourceID, candidate.artifact.Value)
-		payload := map[string]any{"output": debugCheckpointArtifact(candidate.candidate), "warnings": debugWarningEnvelopes(terminalResult.Warnings), "rejection": debugRejectedOutputPtr(nil)}
+		payload := map[string]any{"output": debugCheckpointArtifact(candidate.candidate), "rejection": debugRejectedOutputPtr(nil)}
 		if encodeErr != nil {
 			attemptErr := fmt.Errorf("serialize accepted merge output for lane %q: %w", lane.ID, encodeErr)
 			if stageResult.reuseEligible {
@@ -353,6 +359,7 @@ func (r *Runner) runMergeStage(ctx context.Context, input RunInput, checkpoints
 			}
 			return stageResult, candidate.terminal.record(payload, attemptErr)
 		}
+		stored.Diagnostics = terminalCheckpointDiagnostics(terminalResult)
 		if debugErr := candidate.terminal.record(payload, nil); debugErr != nil {
 			if stageResult.reuseEligible {
 				_ = checkpointMergeFailed(checkpoints, input.stepID, lane.ID, lane.Merge.Module, mergeDeps, debugErr)
@@ -360,20 +367,19 @@ func (r *Runner) runMergeStage(ctx context.Context, input RunInput, checkpoints
 			return stageResult, debugErr
 		}
 		merged, serializedMerge = candidate.artifact, stored
-		mergeWarnings = cloneWarnings(terminalResult.Warnings)
 		output.ValidationSummaries = append(output.ValidationSummaries, terminalSummary)
 		stageResult.validationIncomplete = terminalResult.ValidationIncomplete
 		if stageResult.validationIncomplete {
 			stageResult.reuseEligible = false
 		}
-		output.Warnings = append(output.Warnings, mergeWarnings...)
+		appendDiagnosticGroups(output, terminalDiagnostics)
 		if stageResult.reuseEligible {
-			if err := recordMerge(checkpoints, input.stepID, lane.ID, lane.Merge.Module, mergeDeps, serializedMerge, mergeWarnings); err != nil {
+			if err := recordMerge(checkpoints, input.stepID, lane.ID, lane.Merge.Module, mergeDeps, serializedMerge); err != nil {
 				return stageResult, err
 			}
 		}
 	}
-	if err := writeDebugTimed(input.Debug, path.Join("merge", fileio.EncodePathComponent(lane.ID), "output.json"), debugTimedEnvelope{Stage: string(StageMerge), StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Merge.Module, StartedAt: time.Now().UTC(), Payload: map[string]any{"reused": mergeDecision.Reused, "accepted": true, "output": debugCheckpointArtifact(serializedMerge), "warnings": debugWarningEnvelopes(mergeWarnings)}}); err != nil {
+	if err := writeDebugTimed(input.Debug, path.Join("merge", fileio.EncodePathComponent(lane.ID), "output.json"), debugTimedEnvelope{Stage: string(StageMerge), StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Merge.Module, StartedAt: time.Now().UTC(), Payload: map[string]any{"reused": mergeDecision.Reused, "accepted": true, "output": debugCheckpointArtifact(serializedMerge)}}); err != nil {
 		return stageResult, err
 	}
 	stageResult.artifact = merged
@@ -383,7 +389,6 @@ func (r *Runner) runMergeStage(ctx context.Context, input RunInput, checkpoints
 
 type normalizeStageResult struct {
 	serialized    CheckpointArtifact
-	warnings      []contracts.Warning
 	accepted      bool
 	reuseEligible bool
 }
@@ -409,11 +414,13 @@ func (r *Runner) runNormalizeStage(ctx context.Context, input RunInput, checkpoi
 		return stageResult, err
 	}
 	var serializedNormalize CheckpointArtifact
-	var normalizeWarnings []contracts.Warning
 	if normalizeDecision.Reused {
 		serializedNormalize = normalizeResolution.artifacts[0]
-		normalizeWarnings = cloneWarnings(normalizeCP.Warnings)
-		output.Warnings = append(output.Warnings, normalizeWarnings...)
+		diagnostics, diagnosticErr := promoteCheckpointDiagnostics(serializedNormalize.Diagnostics, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageNormalize, StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Normalize.Module}, nil)
+		if diagnosticErr != nil {
+			return stageResult, fmt.Errorf("promote reused normalize diagnostics: %w", diagnosticErr)
+		}
+		appendDiagnosticGroups(output, diagnostics)
 	} else {
 		if stageResult.reuseEligible {
 			if err := checkpointNormalizeRunning(checkpoints, input.stepID, lane.ID, lane.Normalize.Module, normalizeDeps); err != nil {
@@ -433,28 +440,27 @@ func (r *Runner) runNormalizeStage(ctx context.Context, input RunInput, checkpoi
 			if callErr != nil {
 				return producerAttemptOutput{}, terminal.record(nil, fmt.Errorf("normalize lane %q with normalizer %q: %w", lane.ID, lane.Normalize.Module, callErr))
 			}
-			warnings := cloneWarnings(result.Warnings)
 			serializedCandidate, encodeErr := serializeCandidateArtifact(typed.codec, lane.ID, lane.Normalize.Module, doc.ID, result.Value)
 			if encodeErr != nil {
-				return producerAttemptOutput{}, terminal.record(map[string]any{"warnings": debugWarningEnvelopes(warnings)}, fmt.Errorf("serialize normalize candidate for lane %q: %w", lane.ID, encodeErr))
+				return producerAttemptOutput{}, terminal.record(nil, fmt.Errorf("serialize normalize candidate for lane %q: %w", lane.ID, encodeErr))
 			}
 			attemptValue := normalizeAttemptValue{value: result.Value, candidate: serializedCandidate, terminal: &terminal}
 			var directive *producerRetryDirective
 			if result.Retry != nil {
 				if err := validateNormalizeRetry(result.Retry); err != nil {
-					return producerAttemptOutput{}, terminal.record(map[string]any{"output": debugCheckpointArtifact(serializedCandidate), "warnings": debugWarningEnvelopes(warnings)}, fmt.Errorf("normalize lane %q returned invalid retry directive: %w", lane.ID, err))
+					return producerAttemptOutput{}, terminal.record(map[string]any{"output": debugCheckpointArtifact(serializedCandidate)}, fmt.Errorf("normalize lane %q returned invalid retry directive: %w", lane.ID, err))
 				}
 				anotherAttempt := request.Number <= lane.Normalize.Retries
-				attemptValue.retry = map[string]any{"reason_code": result.Retry.ReasonCode, "message": result.Retry.Message, "another_attempt": anotherAttempt, "fallback_accepted": !anotherAttempt}
-				directive = &producerRetryDirective{FallbackWarnings: cloneWarnings(result.Retry.FallbackWarnings)}
+				attemptValue.retry = map[string]any{"reason_code": result.Retry.ReasonCode, "message": result.Retry.Message, "correction_available": result.Retry.CorrectionGuidance != "", "another_attempt": anotherAttempt, "fallback_accepted": !anotherAttempt}
+				directive = &producerRetryDirective{CorrectionGuidance: result.Retry.CorrectionGuidance, FallbackDiagnostics: contracts.CloneProducerDiagnostics(result.Retry.FallbackDiagnostics)}
 				if anotherAttempt {
-					payload := map[string]any{"output": debugCheckpointArtifact(serializedCandidate), "warnings": debugWarningEnvelopes(warnings), "retry": attemptValue.retry}
+					payload := map[string]any{"output": debugCheckpointArtifact(serializedCandidate), "retry": attemptValue.retry}
 					if debugErr := terminal.record(payload, nil); debugErr != nil {
 						return producerAttemptOutput{}, debugErr
 					}
 				}
 			}
-			return producerAttemptOutput{Value: attemptValue, Candidate: result.ModelCandidate, Warnings: warnings, Retry: directive}, nil
+			return producerAttemptOutput{Value: attemptValue, Candidate: result.ModelCandidate, Diagnostics: contracts.CloneProducerDiagnostics(result.Diagnostics), Retry: directive}, nil
 		}, func(validationCtx context.Context, output producerAttemptOutput) (validationReport, error) {
 			candidate, ok := output.Value.(normalizeAttemptValue)
 			if !ok {
@@ -462,7 +468,7 @@ func (r *Runner) runNormalizeStage(ctx context.Context, input RunInput, checkpoi
 			}
 			target := typedValidationTarget{stage: StageNormalize, stepID: input.stepID, laneID: lane.ID, moduleKey: lane.Normalize.Module, source: doc, sourceID: doc.ID, sourceInput: sourceInput.Clone(), sessionID: sessionID, references: normalizeReferences, metadata: input.Metadata, value: candidate.value, candidate: &candidate.candidate}
 			report, validationErr := r.validateTypedReport(validationCtx, typed.codec, target, prepared.normalizeValidators, candidate.terminal.envelope.Attempt, input.Debug)
-			payload := map[string]any{"output": debugCheckpointArtifact(candidate.candidate), "warnings": debugWarningEnvelopes(append(cloneWarnings(output.Warnings), report.Warnings()...)), "rejection": debugRejectedOutputPtr(typedRejection(report, target, candidate.terminal.envelope.Attempt))}
+			payload := map[string]any{"output": debugCheckpointArtifact(candidate.candidate), "rejection": debugRejectedOutputPtr(typedRejection(report, target, candidate.terminal.envelope.Attempt))}
 			if candidate.retry != nil {
 				payload["retry"] = candidate.retry
 			}
@@ -484,6 +490,10 @@ func (r *Runner) runNormalizeStage(ctx context.Context, input RunInput, checkpoi
 			}
 			return stageResult, runErr
 		}
+		terminalDiagnostics, diagnosticErr := terminalDiagnosticGroups(terminalResult, contracts.DiagnosticOrigin{Stage: contracts.DiagnosticOriginStageNormalize, StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Normalize.Module}, nil)
+		if diagnosticErr != nil {
+			return stageResult, fmt.Errorf("promote normalize diagnostics: %w", diagnosticErr)
+		}
 		if terminalResult.Action == producerTerminalRejected {
 			rejected := terminalResult.Rejection
 			if rejected == nil {
@@ -492,7 +502,7 @@ func (r *Runner) runNormalizeStage(ctx context.Context, input RunInput, checkpoi
 			rejected.Stage, rejected.StepID, rejected.LaneID, rejected.ModuleKey = string(StageNormalize), input.stepID, lane.ID, lane.Normalize.Module
 			rejected.Validation = &terminalSummary
 			output.ValidationSummaries = append(output.ValidationSummaries, artifacts.CloneValidationSummary(terminalSummary))
-			output.Warnings = append(output.Warnings, terminalResult.Warnings...)
+			appendDiagnosticGroups(output, terminalDiagnostics)
 			output.Rejected = append(output.Rejected, *rejected)
 			if stageResult.reuseEligible {
 				if err := checkpointNormalizeRejected(checkpoints, input.stepID, lane.ID, lane.Normalize.Module, normalizeDeps, *rejected); err != nil {
@@ -506,7 +516,7 @@ func (r *Runner) runNormalizeStage(ctx context.Context, input RunInput, checkpoi
 			return stageResult, fmt.Errorf("normalize attempt terminal has incompatible value")
 		}
 		stored, encodeErr := checkpointArtifact(typed.codec, lane.ID, lane.Normalize.Module, doc.ID, candidate.value)
-		payload := map[string]any{"output": debugCheckpointArtifact(candidate.candidate), "warnings": debugWarningEnvelopes(terminalResult.Warnings), "rejection": debugRejectedOutputPtr(nil)}
+		payload := map[string]any{"output": debugCheckpointArtifact(candidate.candidate), "rejection": debugRejectedOutputPtr(nil)}
 		if candidate.retry != nil {
 			payload["retry"] = candidate.retry
 		}
@@ -517,6 +527,7 @@ func (r *Runner) runNormalizeStage(ctx context.Context, input RunInput, checkpoi
 			}
 			return stageResult, candidate.terminal.record(payload, attemptErr)
 		}
+		stored.Diagnostics = terminalCheckpointDiagnostics(terminalResult)
 		if debugErr := candidate.terminal.record(payload, nil); debugErr != nil {
 			if stageResult.reuseEligible {
 				_ = checkpointNormalizeFailed(checkpoints, input.stepID, lane.ID, lane.Normalize.Module, normalizeDeps, debugErr)
@@ -524,23 +535,21 @@ func (r *Runner) runNormalizeStage(ctx context.Context, input RunInput, checkpoi
 			return stageResult, debugErr
 		}
 		serializedNormalize = stored
-		normalizeWarnings = cloneWarnings(terminalResult.Warnings)
 		output.ValidationSummaries = append(output.ValidationSummaries, terminalSummary)
-		output.Warnings = append(output.Warnings, normalizeWarnings...)
+		appendDiagnosticGroups(output, terminalDiagnostics)
 		if terminalResult.ValidationIncomplete {
 			stageResult.reuseEligible = false
 		}
 		if stageResult.reuseEligible {
-			if err := recordNormalize(checkpoints, input.stepID, lane.ID, lane.Normalize.Module, normalizeDeps, serializedNormalize, normalizeWarnings); err != nil {
+			if err := recordNormalize(checkpoints, input.stepID, lane.ID, lane.Normalize.Module, normalizeDeps, serializedNormalize); err != nil {
 				return stageResult, err
 			}
 		}
 	}
-	if err := writeDebugTimed(input.Debug, path.Join("normalize", fileio.EncodePathComponent(lane.ID), "output.json"), debugTimedEnvelope{Stage: string(StageNormalize), StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Normalize.Module, StartedAt: time.Now().UTC(), Payload: map[string]any{"reused": normalizeDecision.Reused, "accepted": true, "output": debugCheckpointArtifact(serializedNormalize), "warnings": debugWarningEnvelopes(normalizeWarnings)}}); err != nil {
+	if err := writeDebugTimed(input.Debug, path.Join("normalize", fileio.EncodePathComponent(lane.ID), "output.json"), debugTimedEnvelope{Stage: string(StageNormalize), StepID: input.stepID, LaneID: lane.ID, ModuleKey: lane.Normalize.Module, StartedAt: time.Now().UTC(), Payload: map[string]any{"reused": normalizeDecision.Reused, "accepted": true, "output": debugCheckpointArtifact(serializedNormalize)}}); err != nil {
 		return stageResult, err
 	}
 	stageResult.serialized = serializedNormalize
-	stageResult.warnings = normalizeWarnings
 	stageResult.accepted = true
 	return stageResult, nil
 }

internal/framework/pipeline/synchronized_collaborators.go

@@ -134,8 +134,8 @@ func (r *lockedCheckpointRecorder) SourceFailed(key string, err error) error {
 func (r *lockedCheckpointRecorder) ExtractRunning(lane, key string, deps []CheckpointFingerprint) error {
 	return r.call(func() error { return r.inner.ExtractRunning(lane, key, deps) })
 }
-func (r *lockedCheckpointRecorder) ExtractSucceeded(lane, key string, deps []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput, warnings []contracts.Warning) error {
-	return r.call(func() error { return r.inner.ExtractSucceeded(lane, key, deps, outputs, rejected, warnings) })
+func (r *lockedCheckpointRecorder) ExtractSucceeded(lane, key string, deps []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput) error {
+	return r.call(func() error { return r.inner.ExtractSucceeded(lane, key, deps, outputs, rejected) })
 }
 func (r *lockedCheckpointRecorder) ExtractFailed(lane, key string, deps []CheckpointFingerprint, err error) error {
 	return r.call(func() error { return r.inner.ExtractFailed(lane, key, deps, err) })
@@ -143,8 +143,8 @@ func (r *lockedCheckpointRecorder) ExtractFailed(lane, key string, deps []Checkp
 func (r *lockedCheckpointRecorder) MergeRunning(lane, key string, deps []CheckpointFingerprint) error {
 	return r.call(func() error { return r.inner.MergeRunning(lane, key, deps) })
 }
-func (r *lockedCheckpointRecorder) MergeSucceeded(lane, key string, deps []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error {
-	return r.call(func() error { return r.inner.MergeSucceeded(lane, key, deps, output, warnings) })
+func (r *lockedCheckpointRecorder) MergeSucceeded(lane, key string, deps []CheckpointFingerprint, output CheckpointArtifact) error {
+	return r.call(func() error { return r.inner.MergeSucceeded(lane, key, deps, output) })
 }
 func (r *lockedCheckpointRecorder) MergeRejected(lane, key string, deps []CheckpointFingerprint, rejected contracts.RejectedOutput) error {
 	return r.call(func() error { return r.inner.MergeRejected(lane, key, deps, rejected) })
@@ -155,8 +155,8 @@ func (r *lockedCheckpointRecorder) MergeFailed(lane, key string, deps []Checkpoi
 func (r *lockedCheckpointRecorder) NormalizeRunning(lane, key string, deps []CheckpointFingerprint) error {
 	return r.call(func() error { return r.inner.NormalizeRunning(lane, key, deps) })
 }
-func (r *lockedCheckpointRecorder) NormalizeSucceeded(lane, key string, deps []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error {
-	return r.call(func() error { return r.inner.NormalizeSucceeded(lane, key, deps, output, warnings) })
+func (r *lockedCheckpointRecorder) NormalizeSucceeded(lane, key string, deps []CheckpointFingerprint, output CheckpointArtifact) error {
+	return r.call(func() error { return r.inner.NormalizeSucceeded(lane, key, deps, output) })
 }
 func (r *lockedCheckpointRecorder) NormalizeRejected(lane, key string, deps []CheckpointFingerprint, rejected contracts.RejectedOutput) error {
 	return r.call(func() error { return r.inner.NormalizeRejected(lane, key, deps, rejected) })
@@ -173,12 +173,12 @@ func (r *lockedCheckpointRecorder) ExtractRunningForStep(step, lane, key string,
 		return r.inner.ExtractRunning(lane, key, deps)
 	})
 }
-func (r *lockedCheckpointRecorder) ExtractSucceededForStep(step, lane, key string, deps []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput, warnings []contracts.Warning) error {
+func (r *lockedCheckpointRecorder) ExtractSucceededForStep(step, lane, key string, deps []CheckpointFingerprint, outputs []CheckpointArtifact, rejected []contracts.RejectedOutput) error {
 	return r.call(func() error {
 		if v, ok := r.inner.(StepCheckpointRecorder); ok {
-			return v.ExtractSucceededForStep(step, lane, key, deps, outputs, rejected, warnings)
+			return v.ExtractSucceededForStep(step, lane, key, deps, outputs, rejected)
 		}
-		return r.inner.ExtractSucceeded(lane, key, deps, outputs, rejected, warnings)
+		return r.inner.ExtractSucceeded(lane, key, deps, outputs, rejected)
 	})
 }
 func (r *lockedCheckpointRecorder) ExtractFailedForStep(step, lane, key string, deps []CheckpointFingerprint, err error) error {
@@ -197,12 +197,12 @@ func (r *lockedCheckpointRecorder) MergeRunningForStep(step, lane, key string, d
 		return r.inner.MergeRunning(lane, key, deps)
 	})
 }
-func (r *lockedCheckpointRecorder) MergeSucceededForStep(step, lane, key string, deps []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error {
+func (r *lockedCheckpointRecorder) MergeSucceededForStep(step, lane, key string, deps []CheckpointFingerprint, output CheckpointArtifact) error {
 	return r.call(func() error {
 		if v, ok := r.inner.(StepCheckpointRecorder); ok {
-			return v.MergeSucceededForStep(step, lane, key, deps, output, warnings)
+			return v.MergeSucceededForStep(step, lane, key, deps, output)
 		}
-		return r.inner.MergeSucceeded(lane, key, deps, output, warnings)
+		return r.inner.MergeSucceeded(lane, key, deps, output)
 	})
 }
 func (r *lockedCheckpointRecorder) MergeRejectedForStep(step, lane, key string, deps []CheckpointFingerprint, rejected contracts.RejectedOutput) error {
@@ -229,12 +229,12 @@ func (r *lockedCheckpointRecorder) NormalizeRunningForStep(step, lane, key strin
 		return r.inner.NormalizeRunning(lane, key, deps)
 	})
 }
-func (r *lockedCheckpointRecorder) NormalizeSucceededForStep(step, lane, key string, deps []CheckpointFingerprint, output CheckpointArtifact, warnings []contracts.Warning) error {
+func (r *lockedCheckpointRecorder) NormalizeSucceededForStep(step, lane, key string, deps []CheckpointFingerprint, output CheckpointArtifact) error {
 	return r.call(func() error {
 		if v, ok := r.inner.(StepCheckpointRecorder); ok {
-			return v.NormalizeSucceededForStep(step, lane, key, deps, output, warnings)
+			return v.NormalizeSucceededForStep(step, lane, key, deps, output)
 		}
-		return r.inner.NormalizeSucceeded(lane, key, deps, output, warnings)
+		return r.inner.NormalizeSucceeded(lane, key, deps, output)
 	})
 }
 func (r *lockedCheckpointRecorder) NormalizeRejectedForStep(step, lane, key string, deps []CheckpointFingerprint, rejected contracts.RejectedOutput) error {

internal/framework/pipeline/typed_execution.go

@@ -23,7 +23,7 @@ type erasedMergeArtifact struct {
 
 type erasedTypedResult struct {
 	Value          any
-	Warnings       []contracts.Warning
+	Diagnostics    []contracts.ProducerDiagnostic
 	Retry          *contracts.NormalizeRetry
 	ModelCandidate *contracts.ModelCandidate
 }