Document validation defaults and cleanup roadmap

docs/cli.md

@@ -45,8 +45,10 @@ Flags:
   Defaults to `./notarius-output`.
 - `--diagnostics-dir path`: diagnostics work directory override for this
   invocation.
-- `--llm-profile id`: override every effective LLM-capable module binding to
+- `--llm-profile id`: override every effective LLM-capable pipeline module
-  use one Scriptorium profile ID.
+  binding to use one Scriptorium profile ID. Validator-specific profiles are
+  not overridden. Configured LLM-backed validators with explicit profiles are
+  validated against the configured Scriptorium profile source.
 - `--session-id id`: pass a stable prompt session identifier through LLM-backed
   module calls.
 - `--reference selector=path`: bind a reference path to a chunk, extractor,
@@ -207,8 +209,31 @@ The production CLI currently registers these module keys:
 - normalize: `noop`
 - output: `json`
 
-Configured validator module lists are reserved for a future validator-chain
+## Implemented Production Validators
-feature and are rejected by current configuration validation.
+
+The production CLI currently registers these validator keys:
+
+- `generic/always_accept`
+- `generic/always_reject`
+- `generic/valid_json`
+- `generic/valid_json_schema`
+- `extract/dnd/spells/shape`
+- `extract/dnd/spells/source_refs`
+- `extract/dnd/spells/source_relatedness`
+
+The production default chain for the `dnd/spells` extractor is:
+
+1. `generic/valid_json`
+2. `generic/valid_json_schema`
+3. `extract/dnd/spells/shape`
+4. `extract/dnd/spells/source_refs`
+5. `extract/dnd/spells/source_relatedness`
+
+Validator chain overrides are configured on `chunk`, lane `extract`, lane
+`merge`, and lane `normalize` bindings. Omitted overrides use production
+defaults, `validators: []` disables validation for that binding, and non-empty
+lists replace the default chain in configured order. Validator keys are resolved
+against the registered validator catalog.
 
 For YAML structure, Scriptorium profile sources, environment overrides, and
 module binding syntax, see [Configuration](config.md).

docs/config.md

@@ -131,8 +131,9 @@ Artifact lane fields:
 - `extract`: required module binding.
 - `merge`: optional module binding. Default module is `appendorder`.
 - `normalize`: optional module binding. Default module is `noop`.
-- `validators`: reserved for future configurable validator chains. Non-empty
+- `validators`: deprecated lane-level validator list. Non-empty lists are
-  lists are rejected by current configuration validation.
+  rejected; use `extract.validators`, `merge.validators`, or
+  `normalize.validators`.
 - `references`: optional compatibility alias for extractor reference bindings.
   Lane bindings override pipeline-level bindings for the same slot.
 
@@ -234,12 +235,32 @@ Binding fields:
 - `options`: optional module-specific settings.
 - `references`: optional reference bindings. Supported only for `chunk`,
   `extract`, `merge`, and `normalize` bindings. `input` and `output` bindings
-  reject this field during validation. Validator bindings are reserved for a
+  reject this field during validation.
-  future validator-chain feature and are rejected when configured.
+- `validators`: optional stage-local validator chain override. Supported only
+  for `chunk`, `extract`, `merge`, and `normalize` bindings. Omit the field to
+  use the production default chain; set `validators: []` to force an empty
+  chain; set a non-empty list to use exactly those validators in configured
+  order.
+
+Validator bindings use the same shorthand or object module-binding form, but
+only these fields are supported:
+
+- `module`: validator key.
+- `llm_profile`: optional Scriptorium profile ID for LLM-backed validators.
+- `options`: optional validator-specific settings.
+
+Validator bindings reject `references`, `retries`, and nested `validators`.
+During resolution, deterministic validators reject explicit `llm_profile`
+values.
 
 The `--llm-profile` run flag overrides every effective LLM-capable module
 binding to use one Scriptorium profile ID: chunk, every selected lane extract,
-merge, and normalize binding.
+merge, and normalize binding. It does not override validator-specific
+`llm_profile` values.
+
+Configured LLM-backed validators with explicit `llm_profile` values are
+validated against the configured Scriptorium profile source. Deterministic
+production validators do not call the LLM and must not set `llm_profile`.
 
 ## Implemented Production Modules
 
@@ -253,6 +274,32 @@ merge, and normalize binding.
 | normalize | `noop` | Passes merged raw outputs through unchanged. |
 | output | `json` | Produces JSON output files for normalized `application/json` lanes. |
 
+## Implemented Production Validators
+
+| Key | Execution | Notes |
+| --- | --- | --- |
+| `generic/always_accept` | deterministic | Accepts returned module output. |
+| `generic/always_reject` | deterministic | Rejects returned module output with reason `always_reject`. |
+| `generic/valid_json` | deterministic | Rejects payloads that are not syntactically valid JSON. |
+| `generic/valid_json_schema` | deterministic | Rejects invalid JSON or JSON that does not conform to the module response schema. |
+| `extract/dnd/spells/shape` | deterministic | Rejects malformed D&D spell-cast JSON payloads. |
+| `extract/dnd/spells/source_refs` | deterministic | Rejects missing or invalid D&D spell source references. |
+| `extract/dnd/spells/source_relatedness` | deterministic | Emits warnings when a spell name is not found near its cited source text. |
+
+The production default chain for the `dnd/spells` extractor is:
+
+```yaml
+validators:
+  - generic/valid_json
+  - generic/valid_json_schema
+  - extract/dnd/spells/shape
+  - extract/dnd/spells/source_refs
+  - extract/dnd/spells/source_relatedness
+```
+
+No other production module currently has a default validator chain. Empty
+chains approve output by default.
+
 The `generic` chunker accepts:
 
 - `max_units`: positive integer, default `50`;
@@ -309,6 +356,10 @@ Pipeline resolution additionally checks:
 - required module keys are present;
 - module keys are registered for the expected slot;
 - module capability requirements are satisfied;
+- non-empty validator overrides reference registered validator keys;
+- deterministic validators do not set `llm_profile`;
+- LLM-backed validators with explicit `llm_profile` values reference configured
+  Scriptorium profile IDs;
 - bound reference slots are declared by selected chunk, extractor, merger, or
   normalizer targets;
 - required reference slots are bound for selected targets.

docs/integrations/json-output.md

@@ -82,6 +82,35 @@ sanitizing the lane ID:
       "normalizer": "noop"
     }
   ],
+  "validator_chains": [
+    {
+      "stage": "extract",
+      "lane_id": "spells",
+      "module_key": "dnd/spells",
+      "validators": [
+        {
+          "key": "generic/valid_json",
+          "execution_class": "deterministic"
+        },
+        {
+          "key": "generic/valid_json_schema",
+          "execution_class": "deterministic"
+        },
+        {
+          "key": "extract/dnd/spells/shape",
+          "execution_class": "deterministic"
+        },
+        {
+          "key": "extract/dnd/spells/source_refs",
+          "execution_class": "deterministic"
+        },
+        {
+          "key": "extract/dnd/spells/source_relatedness",
+          "execution_class": "deterministic"
+        }
+      ]
+    }
+  ],
   "validation_status": "approved",
   "started_at": "2026-01-01T00:00:00Z",
   "completed_at": "2026-01-01T00:00:01Z"
@@ -103,6 +132,12 @@ references.
 `validation_status` is `approved` when no raw outputs were rejected and
 `rejected` when one or more raw outputs were rejected.
 
+`validator_chains` records the resolved validator chain for each validation
+point. Entries include stage, lane ID when applicable, module key, and validators
+with key and execution class. Empty chains are recorded with an empty
+`validators` array, including chains resolved from explicit empty config
+overrides.
+
 `normalized_outputs` summarizes each normalized lane output without embedding
 payload bytes. Entries include lane ID, normalizer module key, source ID, media
 type, and response schema provenance where available.

docs/internal/llm.md

@@ -19,9 +19,10 @@ structured output. The response also carries the raw structured output bytes
 returned by the runtime so modules can preserve raw payloads in pipeline stage
 outputs.
 
-Modules that call the LLM own their prompts, schemas, prompt IDs, validators,
+Modules that call the LLM own their prompts, schemas, prompt IDs, and
-and domain-specific interpretation. Provider adapters should not contain
+domain-specific interpretation. Validator packages own approve/reject policy,
-domain-specific prompt logic.
+and central catalog mappings decide which validators run by default. Provider
+adapters should not contain domain-specific prompt logic.
 
 Prompt input materials carry source or reference bytes with optional origin
 metadata. The Scriptorium-backed runtime receives them as named artifacts rather
@@ -43,9 +44,11 @@ The runtime records the actual selected Scriptorium profile, provider, and model
 used during execution. Manifest population does not rely on a precomputed
 profile ID before pipeline execution.
 
-Explicit profile validation and `--llm-profile` overrides apply to LLM-capable
+Explicit profile validation applies to LLM-capable pipeline stages: chunk,
-pipeline stages: chunk, extract, merge, and normalize. Input, output, and
+extract, merge, normalize, and LLM-backed validators with explicit
-validator bindings are not part of the current production LLM profile scope.
+`llm_profile` values. Input, output, and deterministic validators do not call
+the LLM. The `--llm-profile` run flag overrides effective chunk, extract, merge,
+and normalize bindings; it does not override validator-specific profiles.
 
 ## Scriptorium Adapter
 

docs/internal/modules.md

@@ -5,6 +5,9 @@ contract from `internal/framework/contracts`, exposes a `ModuleSpec`, and
 registers itself with the matching pipeline registry.
 
 The CLI production catalog currently registers only the modules listed here.
+Validator implementations live under `internal/validators` and are registered
+separately from modules. Production default validator chains are central CLI
+catalog policy; module packages do not own their default validation chains.
 
 ## Contract Pattern
 
@@ -181,6 +184,12 @@ metadata under `artifact_lanes[].metadata.extractor`. Durable raw output
 details belong in the
 [D&D spell raw output contract](../integrations/dnd-spell-artifacts.md).
 
+The production catalog validates `dnd/spells` raw extract output with generic
+JSON validators followed by D&D spell validators under
+`internal/validators/extract/dnd/spells`. The extractor itself remains
+responsible for prompt, schema, and raw output production rather than
+approve/reject policy.
+
 The `dnd/scenes` chunker and `dnd/spells` extractor declare optional `players`,
 `party`, and `glossary` reference slots accepting UTF-8 plain text, Markdown,
 YAML, or JSON. They also accept `roster` as a deprecated compatibility alias for

docs/internal/pipeline.md

@@ -75,7 +75,8 @@ prompt execution metadata.
 
 `pipeline.Registries` holds concrete constructors for execution. A
 `pipeline.ModuleCatalog` exposes module specs for config validation and
-resolution.
+resolution. The catalog also exposes validator specs and central default
+validator-chain mappings without constructing modules or validators.
 
 Every production module registers a `ModuleSpec` with:
 
@@ -91,6 +92,14 @@ instances. Input, validate, and output specs must not declare reference slots.
 
 Capability checks prevent incompatible pipeline composition before a run starts.
 
+Every production validator registers a `ValidatorSpec` with:
+
+- `Key`: validator key used in config and manifests;
+- `ExecutionClass`: `deterministic` or `llm_backed`.
+
+Default validator chains are keyed by workflow stage and module key. Production
+currently registers a default chain for `extract` module `dnd/spells` only.
+
 ## Runner Input And Output
 
 `pipeline.RunInput` carries:
@@ -118,8 +127,8 @@ The runner:
 2. builds the input adapter and parses the raw input into a source document;
 3. validates the source document;
 4. builds the chunker and produces source chunks, retrying when configured;
-5. validates source chunks against framework invariants and any registered raw
+5. validates source chunks against framework invariants and the resolved chunk
-   chunk validators;
+   validator chain;
 6. runs each selected artifact lane in sorted resolved order;
 7. builds the output encoder and validates logical output file names.
 8. passes accepted normalized raw outputs, rejected output records, warnings,
@@ -181,20 +190,38 @@ Within an artifact lane, the runner:
 
 ## Validators
 
-The current runner handoff is raw-output based. Extractors, mergers, and
+The runner handoff is raw-output based. Chunkers, extractors, mergers, and
 normalizers do not advertise validator chains through their module interfaces.
-Runner-side raw validation chains receive the raw module output plus
+Resolved validation chains receive the raw module output plus stage, lane,
-stage, lane, module, source, and chunk provenance. Empty raw validation chains
+module, source, chunk, schema, session, reference, LLM client/profile, binding
-approve output by default.
+option, and run metadata context. Chunk validators receive the chunk result
+collection, merge validators receive the ordered extract outputs used by the
+merge, and normalize validators receive the accepted merge output. Empty chains
+approve output by default. Response-schema provenance may include in-memory JSON
+schema bytes for validators. Those bytes are omitted from manifests,
+diagnostics, and encoded output files.
 
-Pipeline-configured validator lists are not part of the current runner
+Resolved validator chains come from central default mappings unless a
-contract. Non-empty configured validator lists are rejected during configuration
+stage-local config override is set on `chunk`, lane `extract`, lane `merge`, or
-validation or resolved-run validation so they cannot appear in manifests without
+lane `normalize`. Explicit empty overrides are valid and are recorded as empty
-executing.
+chains in manifests. Explicit non-empty overrides replace the default chain and
+preserve configured order.
+
+The production default chain for `extract` module `dnd/spells` is:
+
+1. `generic/valid_json`
+2. `generic/valid_json_schema`
+3. `extract/dnd/spells/shape`
+4. `extract/dnd/spells/source_refs`
+5. `extract/dnd/spells/source_relatedness`
+
+No other production module currently has a default validator chain.
 
 Validator rejection is a non-fatal run outcome: the rejected output is recorded
 in `RunOutput.Rejected` and does not pass to the next stage. Validator execution
 errors are framework-level errors and retry according to the relevant binding.
+Warning-only validators return approved results with warnings; those warnings
+are promoted only from successful attempts whose outputs are used.
 
 ## Warnings And Failures
 

docs/operations.md

@@ -110,9 +110,8 @@ stderr, and writes warnings to durable output and diagnostics when retained.
 The run manifest `validation_status` indicates whether raw outputs were
 approved or rejected after validation.
 
-Reference-related warnings include empty bound reference files and D&D spell
+Reference-related warnings include empty bound reference files. Empty references
-relatedness warnings such as `spell_not_near_source`. Empty references are still
+are still passed to extractors so optional slots can be intentionally blank.
-passed to extractors so optional slots can be intentionally blank.
 
 ## Cleanup
 

docs/policy/architecture.md

@@ -82,8 +82,9 @@ format-specific validation rules. They should not own extraction-domain
 decisions.
 
 Extract modules own artifact semantics, prompt usage, structured response schema
-selection, validator defaults, and domain-specific interpretation. They should
+selection, and domain-specific interpretation. They should depend on framework
-depend on framework contracts and core source/artifact types, not concrete input
+contracts and core source/artifact types, not concrete input module packages.
+Production validation defaults are central catalog policy, not behavior owned by
 module packages.
 
 Merge modules combine extracted candidates. Normalize modules reconcile merged
@@ -99,14 +100,19 @@ warnings.
 
 Validators should be independently testable and composable.
 
-Deterministic validators should run before LLM-backed validators when both are
+Validators evaluate immutable module outputs returned by `chunk`, `extract`,
-present. Validator decision semantics should be explicit: each candidate
+`merge`, and `normalize` stages. Validator decision semantics should be
-artifact evaluated by a validator should receive exactly one decision from that
+explicit: each validator call approves, rejects, or approves with warnings for
-validator.
+the whole module output it receives. Validator rejection records rejected raw
+output; validator execution errors are framework errors.
 
-LLM-backed review belongs in module-owned validator chains, not in an implicit
+Default validator chains belong in central production catalog mappings keyed by
-global review phase. Extract and normalize modules may both use deterministic
+stage and module key. Pipeline configuration may override those mappings at the
-and LLM-backed validators.
+stage-local module binding. Empty chains are valid and approve by default.
+
+Deterministic validators should run before LLM-backed validators in production
+defaults when both are present. Configured validator order is authoritative and
+must not be silently reordered.
 
 Shared validator runtime mechanics belong in framework code. Concrete validator
 behavior belongs in module or validator implementation packages.

docs/roadmap/future.md

@@ -20,7 +20,8 @@ future work only.
   if references become large enough to require preprocessing.
 - Cross-lane entity normalization.
 - Cross-chunk semantic deduplication.
-- Configurable validator chains with production validator modules.
+- Additional validator packages and production default chains for future
+  modules.
 - Parallel execution where it preserves deterministic manifests and diagnostics.
 - Additional output encoders.
 

docs/roadmap/implementation.md

@@ -1,486 +1,16 @@
-# Validation Refactor Implementation Plan
+# Validation Refactor Implementation
 
-This plan implements the target state described in
+The validation refactor described by this temporary implementation plan has
-[validation.md](validation.md). Follow the stages in order. Do not skip the
+been completed.
-focused tests for a stage before moving to the next one.
 
-## Stage 1: Replace Legacy Validator Contracts With One Raw Output Contract
+Current behavior is documented in:
 
-Goal: make the raw module-output validation boundary the only framework
+- [Configuration](../config.md)
-validator contract.
+- [CLI Reference](../cli.md)
+- [Pipeline Internals](../internal/pipeline.md)
+- [Modules](../internal/modules.md)
+- [JSON Output](../integrations/json-output.md)
+- [Troubleshooting](../troubleshooting.md)
 
-Tasks:
+Remaining future validation ideas, if any, belong in
-
+[Validation Roadmap](validation.md).
-- In `internal/framework/contracts`, replace the legacy candidate-oriented
-  validator contract with a module-output contract.
-- Remove these legacy types after updating callers:
-  - `ValidationDecision`;
-  - candidate-oriented `ValidationRequest`;
-  - candidate-oriented `ValidationResult`;
-  - legacy `Validator` behavior that works over `artifacts.ArtifactCandidate`.
-- Remove `RawValidator`, `RawValidationRequest`, and `RawValidationResult`.
-  The new `Validator` contract covers the current raw validation use case
-  directly.
-- Define execution class metadata:
-
-  ```go
-  type ExecutionClass string
-
-  const (
-      ExecutionClassDeterministic ExecutionClass = "deterministic"
-      ExecutionClassLLMBacked     ExecutionClass = "llm_backed"
-  )
-  ```
-
-- Define the new request shape around immutable module output. It should carry:
-  - `Stage` as a string-compatible stage identifier;
-  - `LaneID`;
-  - `ModuleKey`;
-  - `Source` and `SourceID`;
-  - `SourceInput`;
-  - `SessionID`;
-  - `References`;
-  - `LLMClient`;
-  - `LLMProfile`;
-  - `Options`;
-  - `Metadata`;
-  - `Schema`;
-  - `Payload`;
-  - chunk provenance for extract validation;
-  - `Chunks` for chunk validation;
-  - ordered `ExtractOutputs` for merge validation;
-  - `MergeOutput` for normalize validation.
-- Define the result shape as one decision over the whole module output:
-
-  ```go
-  type ValidationResult struct {
-      Approved bool
-      ReasonCode string
-      Message string
-      DiagnosticArtifactPath string
-      Warnings []Warning
-  }
-  ```
-
-- Define `Validator` as:
-
-  ```go
-  type Validator interface {
-      Name() string
-      ExecutionClass() ExecutionClass
-      Validate(ctx context.Context, req ValidationRequest) (ValidationResult, error)
-  }
-  ```
-
-- Update tests under `internal/framework/contracts` so fake validators exercise
-  the new contract and no longer reference artifact candidates.
-- Keep `artifacts.ArtifactCandidate` only if other non-validation code still
-  needs it. If it becomes unused after later stages, remove it in Stage 6.
-
-Design requirements:
-
-- Validators must not mutate request payloads, source documents, chunks, or
-  upstream outputs.
-- Validator execution errors are framework errors. Validator semantic rejection
-  is represented by `ValidationResult{Approved:false}`.
-- Empty validation chains still approve output by default.
-
-Focused tests:
-
-```sh
-go test ./internal/framework/contracts
-go test ./internal/framework/validate
-```
-
-## Stage 2: Add Validator Specs, Chain Mappings, And Manifest Provenance
-
-Goal: make validators centrally registered and make default chains explicit,
-reviewable, and auditable.
-
-Tasks:
-
-- Replace the current `pipeline.ValidatorRegistry` internals so it registers
-  validator specs, not generic module specs.
-- Add a `pipeline.ValidatorSpec` with at least:
-  - `Key`;
-  - `ExecutionClass`.
-- Keep validator construction lazy through `ValidatorRegistry.Build(key)`.
-- Add `ValidatorRegistry.Spec(key)` and sorted registered-spec accessors for
-  catalog and manifest use.
-- Remove `pipeline.RawValidationRegistry` after replacing its callers with the
-  new chain registry. The new chain registry should be keyed by:
-  - stage;
-  - module key.
-- Add a production/default mapping type such as:
-
-  ```go
-  type ValidatorChainMapping struct {
-      Stage ModuleStage
-      Module string
-      Validators []ModuleBinding
-  }
-  ```
-
-- Add a chain registry with these behaviors:
-  - duplicate stage/module mappings are rejected;
-  - unknown stages are rejected;
-  - empty default chains may be represented by absence of a mapping;
-  - lookup returns a defensive copy.
-- Extend `pipeline.ModuleCatalog` and `pipeline.Registries` to carry the chain
-  mapping registry alongside module and validator registries.
-- Add manifest types under `internal/core/artifacts` for resolved validator
-  chain provenance. Use a top-level manifest field so chunk, extract, merge, and
-  normalize chains can all be represented:
-
-  ```go
-  type ValidatorChainManifest struct {
-      Stage string `json:"stage"`
-      LaneID string `json:"lane_id,omitempty"`
-      ModuleKey string `json:"module_key"`
-      Validators []ValidatorManifest `json:"validators"`
-  }
-
-  type ValidatorManifest struct {
-      Key string `json:"key"`
-      ExecutionClass string `json:"execution_class"`
-  }
-  ```
-
-- Remove `ArtifactLaneManifest.Validators` and update affected tests and JSON
-  output documentation to use top-level validator-chain provenance instead.
-
-Design requirements:
-
-- The production chain mapping is central catalog policy, not module-owned
-  behavior.
-- A resolved run manifest must record the exact chain used for each validation
-  point, including explicit empty chains.
-
-Focused tests:
-
-```sh
-go test ./internal/framework/pipeline
-go test ./internal/core/artifacts
-```
-
-## Stage 3: Implement Stage-Scoped Config Overrides
-
-Goal: allow pipeline config to override default validator mappings for each
-validatable stage while preserving unset versus explicit empty semantics.
-
-Tasks:
-
-- Add stage-local validator override syntax to module bindings for only:
-  - `chunk`;
-  - artifact lane `extract`;
-  - artifact lane `merge`;
-  - artifact lane `normalize`.
-- Do not revive artifact-lane-level `validators` as an active chain. Keep that
-  field rejected with an error that directs users to
-  `extract.validators`, `merge.validators`, or `normalize.validators`.
-- Update file-config parsing so the implementation can distinguish:
-  - validators omitted;
-  - `validators: []`;
-  - `validators: [ ... ]`.
-- Represent this in pipeline profiles with this explicit override type:
-
-  ```go
-  type ValidatorOverride struct {
-      Set bool
-      Validators []ModuleBinding
-  }
-  ```
-
-- Validate configured validators during config validation:
-  - validator module key must be non-empty;
-  - `references` are not supported on validator bindings;
-  - nested validator overrides inside validator bindings are not supported;
-  - retries are not supported on validator bindings in this pass;
-  - explicit `llm_profile` is allowed only for LLM-backed validators and is
-    rejected for deterministic validators during resolution.
-- During pipeline resolution:
-  - unset override uses the central default chain;
-  - explicit empty override resolves to an empty chain and approves by default;
-  - explicit non-empty override resolves exactly the configured validators in
-    configured order;
-  - every referenced validator key must exist in the validator registry;
-  - configured order must not be silently reordered.
-- Update `--llm-profile` behavior:
-  - continue overriding chunk, extract, merge, and normalize module bindings;
-  - do not override validator bindings unless validator-specific profile
-    override behavior is explicitly added in a later roadmap.
-- Update Scriptorium explicit profile validation so it includes LLM-backed
-  validators with explicit `llm_profile` values, and ignores deterministic
-  validators.
-
-Design requirements:
-
-- Validator compatibility with a module or stage is not enforced during config
-  resolution.
-- Empty resolved chains are valid and should be recorded in manifest
-  provenance.
-- Config validation must fail before runtime if a non-empty override references
-  an unknown validator.
-
-Focused tests:
-
-```sh
-go test ./internal/core/config
-go test ./internal/framework/pipeline
-go test ./internal/cli
-```
-
-## Stage 4: Wire Validator Execution Into The Runner
-
-Goal: run resolved validator chains at chunk, extract, merge, and normalize
-validation points.
-
-Tasks:
-
-- Replace runner calls to the old raw validation registry with calls to the
-  resolved validator chain for the current validation point.
-- Build validators from `ValidatorRegistry` at runtime in resolved order.
-- For each validator call, populate the new `contracts.ValidationRequest` with:
-  - immutable module output payload;
-  - stage/module/lane/source/chunk provenance;
-  - schema metadata and in-memory schema content when available;
-  - source input material;
-  - session ID;
-  - resolved references for the target stage;
-  - LLM client/profile/options/metadata for the validator;
-  - upstream outputs needed by merge and normalize validators.
-- Preserve current retry behavior:
-  - semantic rejection returns a `RejectedOutput` and may be retried according
-    to the module binding's retry setting;
-  - validator execution errors are framework errors and are retried under the
-    same module binding retry setting;
-  - final rejection is recorded and does not pass downstream.
-- Preserve current warning behavior:
-  - warnings from accepted attempts are promoted;
-  - warnings from discarded retry attempts are not promoted;
-  - warning-only validators return `Approved:true` with warnings.
-- Continue enforcing framework-level chunk invariants in the runner before
-  chunk validators run.
-- Populate manifest validator-chain provenance after resolution and before
-  output encoding.
-- Keep rejected output manifest entries compatible with current `rejected.json`
-  and `manifest.json` shapes where practical.
-
-Design requirements:
-
-- Validators are read-only. Do not let a validator return rewritten payloads or
-  replacement stage outputs.
-- If a validator chain is empty, approve without building validators.
-- If a validator is mapped to an unsuitable output shape, the validator should
-  return a clear execution error or an explicit documented approval.
-
-Focused tests:
-
-```sh
-go test ./internal/framework/pipeline
-go test ./internal/cli
-```
-
-## Stage 5: Add Generic Validators
-
-Goal: provide reusable validators needed for incremental module development and
-production structural checks.
-
-Tasks:
-
-- Add `internal/validators/generic/always_accept`.
-  - Key: `generic/always_accept`.
-  - Execution class: deterministic.
-  - Always returns approved.
-- Add `internal/validators/generic/always_reject`.
-  - Key: `generic/always_reject`.
-  - Execution class: deterministic.
-  - Always returns rejected with stable reason code `always_reject`.
-- Add `internal/validators/generic/valid_json`.
-  - Key: `generic/valid_json`.
-  - Execution class: deterministic.
-  - Accepts syntactically valid JSON.
-  - Rejects invalid JSON with stable reason code `invalid_json`.
-- Add `internal/validators/generic/valid_json_schema`.
-  - Key: `generic/valid_json_schema`.
-  - Execution class: deterministic.
-  - Uses the module output's in-memory JSON schema content.
-  - Rejects invalid JSON with `invalid_json`.
-  - Rejects schema non-conformance with `json_schema_invalid`.
-  - Returns a validator execution error when no schema content is available.
-- Make `github.com/santhosh-tekuri/jsonschema/v6` a direct dependency for
-  schema validation.
-- Extend `contracts.ResponseSchema` with an in-memory-only field:
-
-  ```go
-  JSONSchema []byte `json:"-"`
-  ```
-
-  Module outputs should carry schema bytes to validators without serializing raw
-  schema content into manifests, diagnostics, or default output files.
-- Update D&D scene and D&D spell modules to populate in-memory schema content on
-  their output schema values.
-- Do not add a shared validator test helper package in this pass. Keep tests
-  local unless a later cleanup finds substantial duplication.
-
-Design requirements:
-
-- Generic validators must not depend on concrete production modules.
-- Tests must not assert exact embedded production prompt text.
-- Raw schema content must not be emitted in diagnostics or manifests.
-
-Focused tests:
-
-```sh
-go test ./internal/validators/generic/...
-go test ./internal/modules/chunk/dnd/scenes
-go test ./internal/modules/extract/dnd/spells
-go test ./internal/framework/pipeline
-```
-
-## Stage 6: Migrate D&D Spell Validators
-
-Goal: move D&D spell validation behavior out of the extractor package and into
-raw-output validators.
-
-Tasks:
-
-- Add `internal/validators/extract/dnd/spells/shape`.
-  - Key: `extract/dnd/spells/shape`.
-  - Execution class: deterministic.
-  - Parses raw spell JSON and rejects malformed spell-cast payloads or missing
-    required fields.
-- Add `internal/validators/extract/dnd/spells/source_refs`.
-  - Key: `extract/dnd/spells/source_refs`.
-  - Execution class: deterministic.
-  - Parses raw spell JSON and rejects missing or invalid source references.
-  - Use `internal/core/source.ValidateRef` for source reference validation.
-- Add `internal/validators/extract/dnd/spells/source_relatedness`.
-  - Key: `extract/dnd/spells/source_relatedness`.
-  - Execution class: deterministic.
-  - Warning-only validator.
-  - Approves output and emits `spell_not_near_source` warnings when a spell name
-    is not found in the cited source text.
-- Keep D&D-specific validator parsing local to validator packages or a validator
-  helper package. Do not import concrete extractor packages from validators.
-- Remove `internal/modules/extract/dnd/spells/validator.go` and its
-  candidate-oriented tests after equivalent validator-package tests exist.
-- Remove now-unused legacy candidate validation helpers and artifact candidate
-  types if they have no remaining callers.
-
-Design requirements:
-
-- The spell extractor should remain responsible for prompt/schema/provenance
-  and raw output production, not accept/reject policy.
-- `source_relatedness` must not reject output. It should only warn.
-- The default D&D spell chain must preserve the intended order:
-  `generic/valid_json`, `generic/valid_json_schema`,
-  `extract/dnd/spells/shape`, `extract/dnd/spells/source_refs`,
-  `extract/dnd/spells/source_relatedness`.
-
-Focused tests:
-
-```sh
-go test ./internal/validators/extract/dnd/spells/...
-go test ./internal/modules/extract/dnd/spells
-go test ./internal/framework/pipeline
-```
-
-## Stage 7: Register Production Validators And Defaults
-
-Goal: make production validation policy explicit and active.
-
-Tasks:
-
-- Register all generic validators in `internal/cli/catalog.go`.
-- Register all D&D spell validators in `internal/cli/catalog.go`.
-- Register default production chain mappings centrally near production module
-  registration.
-- Add the default chain for `extract` module `dnd/spells`:
-  - `generic/valid_json`;
-  - `generic/valid_json_schema`;
-  - `extract/dnd/spells/shape`;
-  - `extract/dnd/spells/source_refs`;
-  - `extract/dnd/spells/source_relatedness`.
-- Do not add default chains for other modules unless the stage has a meaningful
-  validator for that output shape.
-- Update catalog construction tests so available modules, available validators,
-  and default mappings can be reviewed together.
-- Add CLI tests proving:
-  - default D&D spell validators run;
-  - explicit empty override disables the default chain;
-  - explicit non-empty override replaces the default chain;
-  - configured validator order is preserved;
-  - unknown configured validator keys fail during validation/resolution;
-  - deterministic validators reject explicit `llm_profile`;
-  - LLM-backed validator profile validation checks explicit profile IDs.
-
-Design requirements:
-
-- Production defaults must be reviewable without constructing concrete modules.
-- Empty default chains approve by default.
-- The manifest records resolved chains for default chains, empty overrides, and
-  explicit override chains.
-
-Focused tests:
-
-```sh
-go test ./internal/cli
-go test ./internal/core/config
-go test ./internal/framework/pipeline
-```
-
-## Stage 8: Documentation, Examples, And Cleanup
-
-Goal: make implemented validation behavior canonical outside roadmap docs and
-remove stale roadmap instructions.
-
-Tasks:
-
-- Update `docs/policy/architecture.md` so validation policy reflects central
-  mappings and validator packages rather than module-owned validator chains.
-- Update `docs/config.md` to document stage-local validator overrides:
-  - unset uses defaults;
-  - explicit empty disables validators;
-  - explicit non-empty replaces defaults in configured order.
-- Update `docs/cli.md` to list production validators and describe validation
-  profile behavior.
-- Update `docs/internal/pipeline.md` with the implemented validator contract,
-  chain mapping, retry behavior, and manifest provenance.
-- Update `docs/internal/modules.md` to remove stale claims that modules own
-  validator defaults.
-- Update `docs/integrations/json-output.md` for any manifest shape changes.
-- Update `docs/troubleshooting.md` with validator-chain debugging guidance.
-- Leave maintained example configs unchanged unless a validator override example
-  is added intentionally with corresponding CLI/config tests. Keep examples
-  secret-free and loadable.
-- Replace `docs/roadmap/implementation.md` with a concise completed note after
-  all stages are implemented.
-- Update `docs/roadmap/validation.md` so it no longer describes completed work
-  as future work. Keep only deferred validation work, if any remains.
-
-Focused documentation checks:
-
-```sh
-rg -n "reserved for future configurable validator chains|configured validators are not supported|candidate-oriented|RawValidationRegistry" docs internal
-```
-
-Expected result after implementation: no stale current-behavior documentation
-describes configured validators as rejected, and no production code path depends
-on the legacy candidate-oriented validator contract.
-
-## Full Validation
-
-Run after every stage that changes shared contracts, and at the end:
-
-```sh
-go test ./...
-go vet ./...
-go build ./cmd/notarius
-```
-
-Also run focused packages added by this plan:
-
-```sh
-go test ./internal/validators/...
-```

docs/roadmap/validation.md

@@ -1,463 +1,31 @@
-# Validation System Refactor
+# Validation Roadmap
-
+
-This roadmap defines the target state for making validation a first-class,
+The first-class raw-output validation system is implemented. Current behavior is
-composable pipeline concern.
+documented outside roadmap files, especially in [Configuration](../config.md),
-
+[CLI Reference](../cli.md), [Pipeline Internals](../internal/pipeline.md), and
-Current pipeline behavior is raw-output based. The runner can execute
+[JSON Output](../integrations/json-output.md).
-`contracts.RawValidator` chains from `pipeline.RawValidationRegistry` for
+
-`chunk`, `extract`, `merge`, and `normalize` outputs. Empty chains approve by
+Implemented behavior includes:
-default, validator rejection records a rejected raw output, and rejected output
+
-does not pass to the next stage. Production currently registers no raw
+- a single raw module-output validator contract for chunk, extract, merge, and
-validators, and non-empty pipeline-configured validator lists are rejected so
+  normalize outputs;
-they cannot appear in manifests without executing.
+- validator specs with deterministic and LLM-backed execution classes;
-
+- central production validator registration and default chain mappings;
-Legacy candidate validator contracts and D&D spell validators still exist under
+- stage-local config overrides that distinguish omitted, explicit empty, and
-`internal/modules/extract/dnd/spells`, but they are not part of the current
+  explicit non-empty validator chains;
-runner path. The desired end state is that validator implementations, validator
+- manifest provenance for resolved validator chains;
-registration, and default module-to-validator mappings are explicit, reviewable,
+- generic JSON validators and D&D spell validators under `internal/validators`;
-and independent of concrete module packages.
+- production defaults for the `dnd/spells` extractor.
-
+
-## Goals
+## Future Work
-
+
-- Move artifact and module-output validation behavior out of `internal/modules`
+- Add production LLM-backed validators when there is a concrete review policy
-  and into `internal/validators`.
+  that benefits from model judgment.
-- Retire or replace the legacy candidate-oriented `contracts.Validator`,
+- Add validator diagnostics and timing summaries if operators need more detail
-  `ValidationRequest`, and `ValidationResult` path after equivalent raw-output
+  than `manifest.json`, `rejected.json`, and `warnings.json` provide.
-  validators exist.
+- Add media-type validators for non-JSON module outputs when such modules are
-- Keep each validator in its own package.
+  introduced.
-- Mirror the stage and domain shape of `internal/modules` where a validator is
+- Add compatibility metadata only if real deployments need config-time
-  module-specific.
+  enforcement that a validator is suitable for a specific stage or module.
-- Support deterministic and LLM-backed validators through the same framework
+- Add batching or context-window controls for LLM-backed validators if validator
-  contract.
+  inputs become large enough to require them.
-- Allow validators to be mapped to modules at any pipeline stage that returns
-  module output for validation: `chunk`, `extract`, `merge`, or `normalize`.
-- Make default production module-to-validator mappings centralized and
-  human-readable.
-- Allow pipeline configuration to override default mappings for advanced use.
-- Preserve the distinction between an unset validator override and an explicit
-  empty validator override.
-- Treat an empty validator set as valid and equivalent to approval.
-- Preserve the rule that module output passes forward unless a validator rejects
-  it.
-- Make successfully returned module output the explicit validation boundary:
-  questions about output syntax, media type, schema conformance, and domain
-  acceptability should be answered by validators.
-
-## Non-Goals
-
-- Do not create a general workflow engine or arbitrary validation DAG.
-- Do not revive the legacy artifact-candidate validation model as the primary
-  runner path.
-- Do not enforce validator compatibility with a module or stage in this pass.
-- Do not move ordinary runtime invariant checks into validator packages.
-- Do not require every module to have validators.
-- Do not require LLM-backed validators for modules that can be checked
-  deterministically.
-- Do not silently reorder configured validator chains unless that behavior is
-  introduced deliberately and documented as part of the validator contract.
-
-## Validation Boundary
-
-Validation packages should own approve/reject/warning evaluation of successfully
-returned module outputs. This means logic that decides whether a chunk result or
-raw extract, merge, or normalize payload should continue through the pipeline
-belongs in `internal/validators`.
-
-The boundary is:
-
-- no module output was returned: execution failed, and the pipeline should report
-  a module or runtime error;
-- module output was returned: the validator chain decides whether that output is
-  acceptable, and an empty validator chain approves it.
-
-Scriptorium and provider errors are execution failures rather than validator
-rejections. This includes provider timeouts, authentication failures,
-transport/runtime failures, Scriptorium structured-output retry exhaustion, and
-malformed responses that Scriptorium rejects before returning module output.
-
-Other validation-like checks should remain with their owning packages:
-
-- input parsing and source-format validation stay in input modules;
-- source document and source reference invariants stay in `internal/core/source`;
-- generic framework chunk invariants that make extraction possible stay in the
-  runner, such as non-empty chunk content, valid unit ranges, and canonical
-  source-unit ordering;
-- config validation stays in `internal/core/config`;
-- registry, profile, and pipeline consistency checks stay in framework and CLI
-  code;
-- response schema loading stays in module asset code;
-- Scriptorium runtime errors stay in LLM runtime code.
-
-Domain validators may call reusable core helpers such as `source.ValidateRef`,
-but the module-output approval or rejection decision should be made by a
-validator.
-
-Validators should answer module-output questions such as:
-
-- is returned content syntactically valid JSON;
-- does returned JSON conform to the module's declared schema;
-- does the returned media type match the module or pipeline policy;
-- are required domain fields present and non-empty;
-- are source references valid and appropriately grounded;
-- does domain-specific output satisfy the configured policy.
-
-## Audita Patterns To Adapt
-
-The validator architecture should adapt useful patterns from
-[`audita`](https://gitea.maximumdirect.net/eric/audita) without copying its
-narrower transcript-correction shape directly.
-
-Useful patterns:
-
-- concrete validators live under `internal/validators`;
-- shared validator runtime mechanics live under a framework package;
-- built-in validator keys are stable and centrally registered;
-- built-in chains are centrally reviewable;
-- validators carry execution-class metadata;
-- deterministic and LLM-backed validators implement one contract;
-- LLM-backed validator runtime can share batching, diagnostics, structured
-  response handling, and malformed-response policy;
-- reports and manifests can classify validator decisions by execution class.
-
-Important Notarius differences:
-
-- mappings must be keyed by stage and module key, not module key alone;
-- mappings should be owned by the central production catalog, not resolved inside
-  concrete module constructors;
-- configured mapping order should be authoritative unless the config explicitly
-  opts into a different ordering policy;
-- validators must support chunk, extract, merge, and normalize outputs rather
-  than only one proposal shape.
-
-## Validator Package Layout
-
-Concrete validators should live under `internal/validators`. Module-specific
-validators should mirror the module tree and use one package per validator:
-
-```text
-internal/validators/extract/dnd/spells/shape
-internal/validators/extract/dnd/spells/source_refs
-internal/validators/extract/dnd/spells/source_relatedness
-```
-
-D&D spell validation policy belongs under
-`internal/validators/extract/dnd/spells`, split by concern rather than bundled
-inside the extractor module.
-
-Generic validators may live under stage-specific generic paths when they operate
-on a particular stage output shape:
-
-```text
-internal/validators/chunk/generic/...
-internal/validators/extract/generic/...
-internal/validators/merge/generic/...
-internal/validators/normalize/generic/...
-```
-
-Truly stage-independent validators may live under `internal/validators/generic`
-once there is a real shared validator that justifies that location. Generic JSON
-syntax and JSON schema validators are likely candidates for
-`internal/validators/generic/valid_json` and
-`internal/validators/generic/valid_json_schema`.
-
-Each validator package should expose:
-
-- a stable validator key;
-- execution-class metadata;
-- a constructor;
-- a validator spec suitable for registration;
-- a `Register` function;
-- focused tests for decisions, warnings, errors, and diagnostics behavior.
-
-Reusable validator runtime mechanics should live in framework code, such as
-`internal/framework/validators`, not in concrete validator packages. This package
-can own shared helpers for decision cardinality, approval/rejection construction,
-LLM validator batching, validator diagnostics, and Scriptorium request plumbing.
-
-The concrete validator packages should own policy: what they inspect, what they
-approve or reject, what warning reason codes they emit, and how they interpret
-domain-specific data.
-
-## Validator Design Policy
-
-Validators should follow a small-tool model: each validator should do one thing
-well. If a validator both rejects output and emits unrelated warnings, split
-those concerns into separate validators so production mappings can include,
-exclude, and order them independently.
-
-Validators are read-only. A validator must not mutate pipeline state, rewrite
-module output, materialize raw output into typed stage output, or enrich the
-`ModuleOutput` passed to later validators. A validator returns an
-accept/reject verdict for the output it evaluates, plus any warnings or
-diagnostic references. Any conversion from raw module output into a downstream
-representation is a separate materialization concern and must not be hidden
-inside a validator.
-
-The initial generic validator set should include:
-
-- `generic/always_accept`: accepts returned module output unchanged. This is
-  functionally equivalent to a no-op validator and is primarily useful for tests,
-  demonstrations, and explicit pass-through configurations.
-- `generic/always_reject`: rejects returned module output without inspecting it.
-  This is primarily useful for tests and for proving rejection plumbing,
-  manifests, and diagnostics.
-- `generic/valid_json`: inspects raw returned module output and accepts only
-  syntactically valid JSON.
-- `generic/valid_json_schema`: compares raw returned JSON with the module's
-  configured response schema and accepts only schema-conformant output.
-
-The exact keys may be adjusted during implementation to match local naming
-conventions, but the validator set should preserve these four behaviors.
-
-For the current D&D spell behavior, the target split is:
-
-- `generic/valid_json`: rejects returned module output that is not syntactically
-  valid JSON.
-- `generic/valid_json_schema`: rejects returned JSON that does not conform to
-  the configured response schema.
-- `extract/dnd/spells/shape`: rejects malformed spell-cast payloads and missing
-  required spell fields.
-- `extract/dnd/spells/source_refs`: rejects missing or invalid source
-  references.
-- `extract/dnd/spells/source_relatedness`: warning-only validator that reports
-  when a spell name is not found in the cited source text.
-
-Production default mappings should generally list deterministic validators
-before LLM-backed validators. This keeps cheap structural failures from consuming
-model calls and keeps diagnostics easier to interpret. Pipeline-configured order
-should still be authoritative; if a user explicitly lists an LLM-backed
-validator before a deterministic validator, the framework should honor that
-order rather than silently reshuffling it.
-
-## Execution Classes
-
-Validator specs should declare an execution class:
-
-```go
-type ExecutionClass string
-
-const (
-    ExecutionClassDeterministic ExecutionClass = "deterministic"
-    ExecutionClassLLMBacked     ExecutionClass = "llm_backed"
-)
-```
-
-Execution class should be metadata on the validator spec or registered
-definition, not an ad hoc convention inferred from package paths. It should be
-used for:
-
-- human-readable catalog and manifest reporting;
-- diagnostics and timing summaries;
-- operational policy such as concurrency budgeting for LLM-backed validators;
-- default mapping review, where deterministic validators should usually appear
-  before LLM-backed validators.
-
-Execution class should not by itself imply compatibility with a stage or module.
-
-## Validator Contract
-
-The validator framework should support validation of outputs from `chunk`,
-`extract`, `merge`, and `normalize` stages. The contract should be generalized
-enough for stage-specific validators to inspect the output they care about while
-ignoring irrelevant fields.
-
-The current `contracts.RawValidationRequest` is the right starting point. It
-already carries stage, lane, module, source, source and chunk provenance,
-response schema metadata, raw payload, and run metadata. The final contract
-should evolve from that raw-output shape rather than from the legacy
-artifact-candidate `ValidationRequest`.
-
-Additional fields needed for the full validator system include:
-
-- source input material when a validator needs to compare module output to the
-  original source payload;
-- session ID;
-- resolved references for the validated target;
-- LLM client and profile for LLM-backed validators;
-- validator options;
-- chunk output collections when validating a chunk module;
-- ordered upstream output envelopes when validating merge or normalize behavior.
-
-A shared module-output envelope should represent the validation boundary.
-Validators may inspect raw returned content and any already-existing typed stage
-envelope, such as `SourceChunk` values for chunk validation, but they must not
-modify it.
-
-Conceptually:
-
-```go
-type ModuleOutput struct {
-    Stage     pipeline.ModuleStage
-    ModuleKey string
-    LaneID    string
-
-    RawContent     []byte
-    MediaType      string
-    ResponseSchema contracts.ResponseSchema
-
-    SourceID   string
-    ChunkID    string
-    ChunkIndex int
-    Chunks     []contracts.SourceChunk
-    Warnings   []contracts.Warning
-}
-```
-
-The final implementation does not need to use this exact shape, but it should
-preserve the boundary: returned raw module output enters validation as immutable
-module output. Validators may parse raw bytes internally to decide approve,
-reject, or warn, but parsing inside a validator must not create or replace the
-payload passed to later stages.
-
-The result should continue to express validator identity, warnings, and explicit
-decisions. For output collections, the implementation should define an explicit
-decision shape rather than silently mutating lists. Validator decisions reject or
-approve output; validators do not rewrite output.
-
-An empty validator list is always valid. With no validators, the framework should
-pass module output forward unchanged and treat the output as approved for that
-validation point. If the approved output cannot be consumed by a later stage
-because its media type or envelope shape is unsuitable, that failure should be
-reported at the downstream boundary that requires a different shape, not as an
-implicit pre-validation rejection.
-
-## Module Development Workflow
-
-The validation system should make iterative module development easier. Once
-pipeline overrides are implemented, a module author should be able to start with
-an explicit empty validator mapping and inspect returned raw LLM output without
-first satisfying JSON syntax, schema, media-type, or domain validators.
-
-A typical development path should be:
-
-1. Configure an empty validator set for the module and inspect raw returned
-   output.
-2. Add `generic/valid_json` and adjust prompts until the model reliably returns
-   syntactically valid JSON.
-3. Add `generic/valid_json_schema` and iterate on prompt/schema alignment.
-4. Add media-type or schema validators appropriate to the module's intended
-   output format.
-5. Add domain-specific validators one at a time until production policy is
-   represented explicitly in the chain.
-
-This workflow is a central reason for making output validation explicit and
-composable rather than hiding schema, shape, or domain checks inside module
-implementation code.
-
-## Central Production Mappings
-
-Production defaults should be defined in a central, human-readable location near
-the production module and validator registries. The mapping should be keyed by
-stage and module key, not only by module key, so future modules can share keys
-only when stage context makes their ownership unambiguous.
-
-Conceptually:
-
-```go
-{
-    Stage: pipeline.StageExtract,
-    Module: "dnd/spells",
-    Validators: []pipeline.ModuleBinding{
-        pipeline.Binding("generic/valid_json"),
-        pipeline.Binding("generic/valid_json_schema"),
-        pipeline.Binding("extract/dnd/spells/shape"),
-        pipeline.Binding("extract/dnd/spells/source_refs"),
-        pipeline.Binding("extract/dnd/spells/source_relatedness"),
-    },
-}
-```
-
-The production catalog should expose three related surfaces together:
-
-- available modules;
-- available validators;
-- default module-to-validator mappings.
-
-This makes production validation policy reviewable without constructing concrete
-modules or searching inside module implementation packages.
-
-The validator registry should expose registered validator specs without building
-validators, including key and execution class. Building a validator should still
-be available for runtime execution.
-
-The mapping surface should preserve the current useful behavior of stage/module
-lookup and empty-chain approval while adding validator specs, execution-class
-metadata, production registration, config override integration, and manifest
-reporting of the resolved chain.
-
-## Pipeline Overrides
-
-Pipeline configuration should be able to override the central default mapping
-for a module binding. Current configuration validation rejects non-empty
-validator lists, and the current config/profile structs do not preserve whether
-an empty list was explicitly configured or simply omitted. The target config
-model must preserve that distinction.
-
-Override semantics should distinguish three states:
-
-- unset validators: use the central production/default mapping;
-- explicit empty validators: run no validators and pass output forward;
-- explicit non-empty validators: run exactly the configured validators in the
-  configured order.
-
-This keeps the happy path concise while preserving advanced control for
-experimentation, debugging, and custom deployments.
-
-The run manifest should record the resolved validator chain for each validation
-point so completed runs remain auditable after defaults or configuration change.
-Each manifest entry should include at least validator key and execution class,
-and should preserve the resolved order actually used for the run.
-
-## LLM-Backed Validator Runtime
-
-LLM-backed validators should use the same validator contract as deterministic
-validators. Shared framework runtime should provide common support for:
-
-- Scriptorium request construction;
-- validator prompt and schema provenance;
-- diagnostics redaction;
-- optional batching or context-window controls when validator inputs are large;
-- mapping successful LLM validator responses into validator decisions and
-  warnings;
-- consistent handling of Scriptorium/runtime errors.
-
-Scriptorium errors raised during validator execution should be treated as
-validator execution errors unless a specific validator deliberately converts a
-successful response into reject/warn decisions. This keeps provider/runtime
-failure distinct from a validator's semantic rejection of module output.
-
-## Stage Coverage
-
-Validators should be composable across all LLM-eligible stages:
-
-- `chunk`: validators can evaluate chunk boundaries, coverage, overlap, metadata,
-  or module-specific chunk quality.
-- `extract`: validators can evaluate raw extracted output, source references,
-  payload shape, evidence quality, media type, or domain constraints.
-- `merge`: validators can evaluate merged output, cross-chunk consistency,
-  deduplication results, media type, or domain-specific reconciliation.
-- `normalize`: validators can evaluate normalized output, final shape,
-  post-processing results, media type, or domain-specific policy.
-
-The framework should not require compatibility declarations in this pass. A
-validator mapped to an unsuitable output shape should return a clear error, or
-approve unchanged only when that is explicitly the validator's documented
-behavior.
-
-## Documentation Impact
-
-Current-behavior docs and policy should describe the implemented validation
-system once the refactor is complete:
-
-- `docs/policy/architecture.md` should describe centralized validator mappings
-  rather than module-owned validator chains.
-- `docs/internal/modules.md` should remove claims that concrete modules own
-  validator defaults.
-- Internal validation docs should describe validator package ownership, mapping
-  precedence, empty-chain approval behavior, and LLM-backed validator support.
-- User/config docs should replace the current "configured validators are
-  reserved and rejected" language with the implemented pipeline override
-  contract.
-
-Roadmap docs should not remain the canonical description of implemented
-validation behavior after the refactor is complete.

docs/troubleshooting.md

@@ -118,9 +118,10 @@ Symptoms include:
 
 Fix:
 
-- Confirm the selected chunker, extractor, merger, or normalizer declares the slot. The
+- Confirm the selected chunker, extractor, merger, or normalizer declares the
-  implemented `dnd/scenes` chunker and `dnd/spells` extractor declare optional
+  slot. The implemented `dnd/scenes` chunker and `dnd/spells` extractor declare
-  `roster` and `glossary` slots.
+  optional `players`, `party`, and `glossary` slots, plus `roster` as a
+  deprecated compatibility alias for `party`.
 - Use a specific selector when more than one selected target declares the same
   slot: `chunk.context=./context.txt`,
   `spells.extract.context=./extract-context.txt`,
@@ -317,11 +318,31 @@ Explanation and fixes:
   pass to the next pipeline stage.
 - Check `rejected.json` for the stage, lane, module, chunk, validator, reason,
   message, and attempt count.
+- Check `manifest.json` `validator_chains` to see the exact resolved validators
+  and order used for the rejected validation point. The production `dnd/spells`
+  extractor runs JSON syntax, JSON schema, D&D spell shape, source-reference,
+  and source-relatedness validators by default.
+- If the configured chain is not what you expected, inspect the selected
+  binding in config. Omitted `validators` uses production defaults,
+  `validators: []` disables validators for that binding, and a non-empty list
+  replaces the default chain in configured order.
+- Run `notarius config validate --pipeline <id>` to catch unknown validator keys
+  and invalid validator `llm_profile` usage before running the pipeline.
 - Increase a module binding's `retries` only when re-running the same module
   input can reasonably produce an acceptable output.
 - If rejection is deterministic, fix the source input, module configuration, or
   validation policy rather than adding retries.
 
+Common production D&D spell validator reasons:
+
+- `invalid_json`: the raw output is not valid JSON.
+- `json_schema_invalid`: the raw JSON does not match the spell response schema.
+- `invalid_spell_shape`: required spell-cast fields are missing or malformed.
+- `invalid_source_refs`: source references are missing or do not point to valid
+  source units.
+- `spell_not_near_source`: warning-only; the spell name was not found near the
+  cited source text.
+
 ## Retry Exhaustion
 
 Symptoms include:

go.mod

@@ -4,10 +4,8 @@ go 1.25.5
 
 require (
 	gitea.maximumdirect.net/eric/scriptorium v0.11.0
+	github.com/santhosh-tekuri/jsonschema/v6 v6.0.2
 	gopkg.in/yaml.v3 v3.0.1
 )
 
-require (
+require golang.org/x/text v0.14.0 // indirect
-	github.com/santhosh-tekuri/jsonschema/v6 v6.0.2 // indirect
-	golang.org/x/text v0.14.0 // indirect
-)

internal/cli/catalog.go

@@ -16,6 +16,13 @@ import (
 	"gitea.maximumdirect.net/eric/notarius/internal/modules/merge/appendorder"
 	"gitea.maximumdirect.net/eric/notarius/internal/modules/normalize/noop"
 	jsonoutput "gitea.maximumdirect.net/eric/notarius/internal/modules/output/json"
+	spellshape "gitea.maximumdirect.net/eric/notarius/internal/validators/extract/dnd/spells/shape"
+	spellsourcerefs "gitea.maximumdirect.net/eric/notarius/internal/validators/extract/dnd/spells/source_refs"
+	spellrelatedness "gitea.maximumdirect.net/eric/notarius/internal/validators/extract/dnd/spells/source_relatedness"
+	alwaysaccept "gitea.maximumdirect.net/eric/notarius/internal/validators/generic/always_accept"
+	alwaysreject "gitea.maximumdirect.net/eric/notarius/internal/validators/generic/always_reject"
+	validjson "gitea.maximumdirect.net/eric/notarius/internal/validators/generic/valid_json"
+	validjsonschema "gitea.maximumdirect.net/eric/notarius/internal/validators/generic/valid_json_schema"
 )
 
 func productionRegistries() (pipeline.Registries, error) {
@@ -26,6 +33,7 @@ func productionRegistries() (pipeline.Registries, error) {
 		Mergers:         pipeline.NewMergerRegistry(),
 		Normalizers:     pipeline.NewNormalizerRegistry(),
 		Validators:      pipeline.NewValidatorRegistry(),
+		ValidatorChains: pipeline.NewValidatorChainRegistry(),
 		Outputs:         pipeline.NewOutputEncoderRegistry(),
 	}
 	if err := seriatim.Register(registries.Inputs); err != nil {
@@ -46,12 +54,56 @@ func productionRegistries() (pipeline.Registries, error) {
 	if err := noop.Register(registries.Normalizers); err != nil {
 		return pipeline.Registries{}, fmt.Errorf("register noop normalizer: %w", err)
 	}
+	if err := registerProductionValidators(registries.Validators); err != nil {
+		return pipeline.Registries{}, err
+	}
+	if err := registerProductionValidatorChains(registries.ValidatorChains); err != nil {
+		return pipeline.Registries{}, err
+	}
 	if err := jsonoutput.Register(registries.Outputs); err != nil {
 		return pipeline.Registries{}, fmt.Errorf("register json output encoder: %w", err)
 	}
 	return registries, nil
 }
 
+func registerProductionValidators(registry *pipeline.ValidatorRegistry) error {
+	registrations := []struct {
+		name     string
+		register func(*pipeline.ValidatorRegistry) error
+	}{
+		{name: "generic always accept validator", register: alwaysaccept.Register},
+		{name: "generic always reject validator", register: alwaysreject.Register},
+		{name: "generic valid json validator", register: validjson.Register},
+		{name: "generic valid json schema validator", register: validjsonschema.Register},
+		{name: "dnd spell shape validator", register: spellshape.Register},
+		{name: "dnd spell source references validator", register: spellsourcerefs.Register},
+		{name: "dnd spell source relatedness validator", register: spellrelatedness.Register},
+	}
+	for _, registration := range registrations {
+		if err := registration.register(registry); err != nil {
+			return fmt.Errorf("register %s: %w", registration.name, err)
+		}
+	}
+	return nil
+}
+
+func registerProductionValidatorChains(registry *pipeline.ValidatorChainRegistry) error {
+	if err := registry.Register(pipeline.ValidatorChainMapping{
+		Stage:  pipeline.StageExtract,
+		Module: spells.Key,
+		Validators: []pipeline.ModuleBinding{
+			pipeline.Binding(validjson.Key),
+			pipeline.Binding(validjsonschema.Key),
+			pipeline.Binding(spellshape.Key),
+			pipeline.Binding(spellsourcerefs.Key),
+			pipeline.Binding(spellrelatedness.Key),
+		},
+	}); err != nil {
+		return fmt.Errorf("register dnd spells validator chain: %w", err)
+	}
+	return nil
+}
+
 func productionCatalog() (pipeline.ModuleCatalog, error) {
 	registries, err := productionRegistries()
 	if err != nil {
@@ -99,6 +151,7 @@ func catalogFromRegistries(registries pipeline.Registries) pipeline.ModuleCatalo
 		Mergers:         registries.Mergers,
 		Normalizers:     registries.Normalizers,
 		Validators:      registries.Validators,
+		ValidatorChains: registries.ValidatorChains,
 		Outputs:         registries.Outputs,
 	}
 }
@@ -111,6 +164,7 @@ func registriesFromCatalog(catalog pipeline.ModuleCatalog) pipeline.Registries {
 		Mergers:         catalog.Mergers,
 		Normalizers:     catalog.Normalizers,
 		Validators:      catalog.Validators,
+		ValidatorChains: catalog.ValidatorChains,
 		Outputs:         catalog.Outputs,
 	}
 }
@@ -122,6 +176,7 @@ func isEmptyCatalog(catalog pipeline.ModuleCatalog) bool {
 		catalog.Mergers == nil &&
 		catalog.Normalizers == nil &&
 		catalog.Validators == nil &&
+		catalog.ValidatorChains == nil &&
 		catalog.Outputs == nil
 }
 
@@ -132,6 +187,7 @@ func isEmptyRegistries(registries pipeline.Registries) bool {
 		registries.Mergers == nil &&
 		registries.Normalizers == nil &&
 		registries.Validators == nil &&
+		registries.ValidatorChains == nil &&
 		registries.Outputs == nil
 }
 

internal/cli/run.go

@@ -495,6 +495,13 @@ func effectiveLLMProfileIDs(resolved pipeline.ResolvedPipeline) []string {
 		add(lane.Merge)
 		add(lane.Normalize)
 	}
+	for _, chain := range resolved.ValidatorChains {
+		for _, validator := range chain.Validators {
+			if validator.ExecutionClass == contracts.ExecutionClassLLMBacked {
+				add(validator.Binding)
+			}
+		}
+	}
 	ids := make([]string, 0, len(seen))
 	for id := range seen {
 		ids = append(ids, id)

internal/cli/run_test.go

@@ -27,6 +27,13 @@ import (
 	"gitea.maximumdirect.net/eric/notarius/internal/modules/merge/appendorder"
 	"gitea.maximumdirect.net/eric/notarius/internal/modules/normalize/noop"
 	jsonoutput "gitea.maximumdirect.net/eric/notarius/internal/modules/output/json"
+	spellshape "gitea.maximumdirect.net/eric/notarius/internal/validators/extract/dnd/spells/shape"
+	spellsourcerefs "gitea.maximumdirect.net/eric/notarius/internal/validators/extract/dnd/spells/source_refs"
+	spellrelatedness "gitea.maximumdirect.net/eric/notarius/internal/validators/extract/dnd/spells/source_relatedness"
+	alwaysaccept "gitea.maximumdirect.net/eric/notarius/internal/validators/generic/always_accept"
+	alwaysreject "gitea.maximumdirect.net/eric/notarius/internal/validators/generic/always_reject"
+	validjson "gitea.maximumdirect.net/eric/notarius/internal/validators/generic/valid_json"
+	validjsonschema "gitea.maximumdirect.net/eric/notarius/internal/validators/generic/valid_json_schema"
 	"gitea.maximumdirect.net/eric/scriptorium"
 )
 
@@ -121,13 +128,13 @@ func TestRunConfigValidateSuccessWithFakeCatalog(t *testing.T) {
 	}
 }
 
-func TestProductionCatalogIncludesDefaultModules(t *testing.T) {
+func TestProductionCatalogIncludesProductionModulesValidatorsAndDefaults(t *testing.T) {
 	catalog, err := productionCatalog()
 	if err != nil {
 		t.Fatalf("productionCatalog() error = %v, want nil", err)
 	}
 
-	tests := []struct {
+	moduleTests := []struct {
 		name string
 		got  func() (pipeline.ModuleSpec, bool)
 		want pipeline.ModuleSpec
@@ -169,7 +176,7 @@ func TestProductionCatalogIncludesDefaultModules(t *testing.T) {
 		},
 	}
 
-	for _, test := range tests {
+	for _, test := range moduleTests {
 		t.Run(test.name, func(t *testing.T) {
 			got, ok := test.got()
 			if !ok {
@@ -180,6 +187,42 @@ func TestProductionCatalogIncludesDefaultModules(t *testing.T) {
 			}
 		})
 	}
+
+	validatorTests := []pipeline.ValidatorSpec{
+		alwaysaccept.Spec(),
+		alwaysreject.Spec(),
+		validjson.Spec(),
+		validjsonschema.Spec(),
+		spellshape.Spec(),
+		spellsourcerefs.Spec(),
+		spellrelatedness.Spec(),
+	}
+	for _, want := range validatorTests {
+		t.Run("validator "+want.Key, func(t *testing.T) {
+			got, ok := catalog.Validators.Spec(want.Key)
+			if !ok {
+				t.Fatalf("validator spec %q ok = false, want true", want.Key)
+			}
+			if !reflect.DeepEqual(got, want) {
+				t.Fatalf("validator spec = %#v, want %#v", got, want)
+			}
+		})
+	}
+
+	gotChain := catalog.ValidatorChains.Validators(pipeline.StageExtract, spells.Key)
+	wantChain := []pipeline.ModuleBinding{
+		pipeline.Binding(validjson.Key),
+		pipeline.Binding(validjsonschema.Key),
+		pipeline.Binding(spellshape.Key),
+		pipeline.Binding(spellsourcerefs.Key),
+		pipeline.Binding(spellrelatedness.Key),
+	}
+	if !reflect.DeepEqual(gotChain, wantChain) {
+		t.Fatalf("dnd spell default validator chain = %#v, want %#v", gotChain, wantChain)
+	}
+	if got := catalog.ValidatorChains.Validators(pipeline.StageChunk, generic.Key); len(got) != 0 {
+		t.Fatalf("generic chunker default validator chain = %#v, want empty", got)
+	}
 }
 
 func TestProductionPromptAssetsRegisterAndPrepareDndPrompts(t *testing.T) {
@@ -336,6 +379,50 @@ func TestRunConfigValidateUnknownProductionModuleIncludesContext(t *testing.T) {
 	}
 }
 
+func TestRunConfigValidateRejectsUnknownProductionValidator(t *testing.T) {
+	configPath := writeTestConfig(t, mvpConfigYAMLWithExtractValidators("dnd-session", "\n            - missing/validator\n"))
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+
+	code := RunWithOptions([]string{"config", "validate", "--config", configPath, "--pipeline", "dnd-session"}, &stdout, &stderr, Options{})
+
+	if code != 1 {
+		t.Fatalf("RunWithOptions() code = %d, want 1", code)
+	}
+	if got := stderr.String(); !strings.Contains(got, "unknown validator") || !strings.Contains(got, "missing/validator") {
+		t.Fatalf("stderr = %q, want unknown validator", got)
+	}
+}
+
+func TestRunConfigValidateRejectsLLMProfileForDeterministicProductionValidator(t *testing.T) {
+	configPath := writeTestConfig(t, `version: 2
+pipelines:
+  dnd-session:
+    input: seriatim
+    artifacts:
+      spells:
+        extract:
+          module: dnd/spells
+          validators:
+            - module: generic/valid_json
+              llm_profile: review
+`)
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+
+	code := RunWithOptions([]string{"config", "validate", "--config", configPath, "--pipeline", "dnd-session"}, &stdout, &stderr, Options{})
+
+	if code != 1 {
+		t.Fatalf("RunWithOptions() code = %d, want 1", code)
+	}
+	got := stderr.String()
+	for _, want := range []string{"llm_profile", "deterministic", validjson.Key} {
+		if !strings.Contains(got, want) {
+			t.Fatalf("stderr = %q, want substring %q", got, want)
+		}
+	}
+}
+
 func TestRunConfigValidateReportsParseErrors(t *testing.T) {
 	configPath := writeFile(t, "config.yml", "version: 1\n")
 	var stdout bytes.Buffer
@@ -743,7 +830,7 @@ func TestRunPipelineLLMFactoryFailure(t *testing.T) {
 	}
 }
 
-func TestRunPipelineCarriesInvalidLLMSourceRefsAsRawOutput(t *testing.T) {
+func TestRunPipelineDefaultDNDSpellValidatorsRejectInvalidSourceRefs(t *testing.T) {
 	configPath := writeTestConfig(t, mvpConfigYAML("dnd-session", "dnd/spells"))
 	inputPath := writeSeriatimInput(t)
 	outputDir := t.TempDir()
@@ -756,11 +843,107 @@ func TestRunPipelineCarriesInvalidLLMSourceRefsAsRawOutput(t *testing.T) {
 		LLMClientFactory: fakeLLMFactory(client, nil),
 	})
 
+	if code != 0 {
+		t.Fatalf("RunWithOptions() code = %d, stderr=%q", code, stderr.String())
+	}
+	if !strings.Contains(stdout.String(), "outputs=0") || !strings.Contains(stdout.String(), "rejected=1") {
+		t.Fatalf("stdout = %q, want rejected output count", stdout.String())
+	}
+
+	var manifest artifacts.RunManifest
+	readJSONFile(t, filepath.Join(onlyChildDir(t, outputDir), "manifest.json"), &manifest)
+	if manifest.ValidationStatus != "rejected" {
+		t.Fatalf("validation status = %q, want rejected", manifest.ValidationStatus)
+	}
+	if len(manifest.RejectedOutputs) != 1 {
+		t.Fatalf("rejected outputs = %#v, want one rejection", manifest.RejectedOutputs)
+	}
+	rejection := manifest.RejectedOutputs[0]
+	if rejection.ValidatorName != spellsourcerefs.Key || rejection.ReasonCode != spellsourcerefs.ReasonCode {
+		t.Fatalf("rejection = %#v, want source reference validator rejection", rejection)
+	}
+	gotChain := manifestValidatorChain(t, manifest, pipeline.StageExtract, "spells", spells.Key)
+	wantKeys := []string{validjson.Key, validjsonschema.Key, spellshape.Key, spellsourcerefs.Key, spellrelatedness.Key}
+	if got := manifestValidatorKeys(gotChain); !reflect.DeepEqual(got, wantKeys) {
+		t.Fatalf("validator chain keys = %#v, want %#v", got, wantKeys)
+	}
+}
+
+func TestRunPipelineExplicitEmptyValidatorOverrideDisablesDNDSpellDefaults(t *testing.T) {
+	configPath := writeTestConfig(t, mvpConfigYAMLWithExtractValidators("dnd-session", " []\n"))
+	inputPath := writeSeriatimInput(t)
+	outputDir := t.TempDir()
+	diagnosticsDir := t.TempDir()
+	client := newFakeRunLLMClient(true)
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+
+	code := RunWithOptions([]string{"run", "dnd-session", "--config", configPath, "--input", inputPath, "--output-dir", outputDir, "--diagnostics-dir", diagnosticsDir}, &stdout, &stderr, Options{
+		LLMClientFactory: fakeLLMFactory(client, nil),
+	})
+
 	if code != 0 {
 		t.Fatalf("RunWithOptions() code = %d, stderr=%q", code, stderr.String())
 	}
 	if !strings.Contains(stdout.String(), "outputs=1") || !strings.Contains(stdout.String(), "rejected=0") {
-		t.Fatalf("stdout = %q, want raw output count", stdout.String())
+		t.Fatalf("stdout = %q, want accepted output count", stdout.String())
+	}
+	var manifest artifacts.RunManifest
+	readJSONFile(t, filepath.Join(onlyChildDir(t, outputDir), "manifest.json"), &manifest)
+	gotChain := manifestValidatorChain(t, manifest, pipeline.StageExtract, "spells", spells.Key)
+	if len(gotChain.Validators) != 0 {
+		t.Fatalf("validator chain = %#v, want explicit empty chain", gotChain)
+	}
+}
+
+func TestRunPipelineExplicitValidatorOverrideReplacesDNDSpellDefaults(t *testing.T) {
+	configPath := writeTestConfig(t, mvpConfigYAMLWithExtractValidators("dnd-session", "\n            - "+alwaysaccept.Key+"\n"))
+	inputPath := writeSeriatimInput(t)
+	outputDir := t.TempDir()
+	diagnosticsDir := t.TempDir()
+	client := newFakeRunLLMClient(true)
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+
+	code := RunWithOptions([]string{"run", "dnd-session", "--config", configPath, "--input", inputPath, "--output-dir", outputDir, "--diagnostics-dir", diagnosticsDir}, &stdout, &stderr, Options{
+		LLMClientFactory: fakeLLMFactory(client, nil),
+	})
+
+	if code != 0 {
+		t.Fatalf("RunWithOptions() code = %d, stderr=%q", code, stderr.String())
+	}
+	if !strings.Contains(stdout.String(), "outputs=1") || !strings.Contains(stdout.String(), "rejected=0") {
+		t.Fatalf("stdout = %q, want accepted output count", stdout.String())
+	}
+	var manifest artifacts.RunManifest
+	readJSONFile(t, filepath.Join(onlyChildDir(t, outputDir), "manifest.json"), &manifest)
+	gotChain := manifestValidatorChain(t, manifest, pipeline.StageExtract, "spells", spells.Key)
+	if got := manifestValidatorKeys(gotChain); !reflect.DeepEqual(got, []string{alwaysaccept.Key}) {
+		t.Fatalf("validator chain keys = %#v, want explicit override", got)
+	}
+}
+
+func TestRunPipelineConfiguredValidatorOrderIsPreserved(t *testing.T) {
+	configPath := writeTestConfig(t, mvpConfigYAMLWithExtractValidators("dnd-session", "\n            - "+alwaysaccept.Key+"\n            - "+validjson.Key+"\n"))
+	inputPath := writeSeriatimInput(t)
+	outputDir := t.TempDir()
+	diagnosticsDir := t.TempDir()
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+
+	code := RunWithOptions([]string{"run", "dnd-session", "--config", configPath, "--input", inputPath, "--output-dir", outputDir, "--diagnostics-dir", diagnosticsDir}, &stdout, &stderr, Options{
+		LLMClientFactory: fakeLLMFactory(newFakeRunLLMClient(false), nil),
+	})
+
+	if code != 0 {
+		t.Fatalf("RunWithOptions() code = %d, stderr=%q", code, stderr.String())
+	}
+	var manifest artifacts.RunManifest
+	readJSONFile(t, filepath.Join(onlyChildDir(t, outputDir), "manifest.json"), &manifest)
+	gotChain := manifestValidatorChain(t, manifest, pipeline.StageExtract, "spells", spells.Key)
+	wantKeys := []string{alwaysaccept.Key, validjson.Key}
+	if got := manifestValidatorKeys(gotChain); !reflect.DeepEqual(got, wantKeys) {
+		t.Fatalf("validator chain keys = %#v, want %#v", got, wantKeys)
 	}
 }
 
@@ -816,6 +999,42 @@ pipelines:
 	}
 }
 
+func TestRunConfigValidateChecksExplicitLLMValidatorProfileIDs(t *testing.T) {
+	profilePath := writeScriptoriumProfileFile(t, "known", "http://profile.test/v1", "test-model")
+	configPath := writeTestConfig(t, `version: 2
+scriptorium:
+  profile_file: `+profilePath+`
+pipelines:
+  example:
+    input: fake/input
+    artifacts:
+      events:
+        extract:
+          module: fake/extract
+          validators:
+            - module: fake/llm-validator
+              llm_profile: missing
+`)
+	catalog := fakeCatalog(t)
+	mustRegisterValidator(t, catalog.Validators, pipeline.ValidatorSpec{
+		Key:            "fake/llm-validator",
+		ExecutionClass: contracts.ExecutionClassLLMBacked,
+	})
+	var stdout bytes.Buffer
+	var stderr bytes.Buffer
+
+	code := RunWithOptions([]string{"config", "validate", "--config", configPath, "--pipeline", "example"}, &stdout, &stderr, Options{
+		Catalog: catalog,
+	})
+
+	if code != 1 {
+		t.Fatalf("RunWithOptions() code = %d, want 1", code)
+	}
+	if !strings.Contains(stderr.String(), "Scriptorium profile") || !strings.Contains(stderr.String(), "missing") {
+		t.Fatalf("stderr = %q, want unknown validator Scriptorium profile", stderr.String())
+	}
+}
+
 func TestRunConfigValidateIncludesMergeAndIgnoresNonLLMStageScriptoriumProfiles(t *testing.T) {
 	profilePath := writeScriptoriumProfileFile(t, "known", "http://profile.test/v1", "test-model")
 	configPath := writeTestConfig(t, `version: 2
@@ -861,13 +1080,29 @@ func TestEffectiveLLMProfileIDsUsesLLMCapableStagesOnly(t *testing.T) {
 				Extract:   pipeline.ModuleBinding{LLMProfile: "extract-profile"},
 				Merge:     pipeline.ModuleBinding{LLMProfile: "merge-profile"},
 				Normalize: pipeline.ModuleBinding{LLMProfile: "normalize-profile"},
-				Validators: []pipeline.ModuleBinding{{LLMProfile: "validator-profile"}},
+			},
+		},
+		ValidatorChains: []pipeline.ResolvedValidatorChain{
+			{
+				Stage:     pipeline.StageExtract,
+				LaneID:    "events",
+				ModuleKey: "extract",
+				Validators: []pipeline.ResolvedValidator{
+					{
+						Binding:        pipeline.ModuleBinding{Module: "deterministic-validator", LLMProfile: "ignored-validator-profile"},
+						ExecutionClass: contracts.ExecutionClassDeterministic,
+					},
+					{
+						Binding:        pipeline.ModuleBinding{Module: "llm-validator", LLMProfile: "validator-profile"},
+						ExecutionClass: contracts.ExecutionClassLLMBacked,
+					},
+				},
 			},
 		},
 	}
 
 	got := effectiveLLMProfileIDs(resolved)
-	want := []string{"chunk-profile", "extract-profile", "merge-profile", "normalize-profile"}
+	want := []string{"chunk-profile", "extract-profile", "merge-profile", "normalize-profile", "validator-profile"}
 	if !reflect.DeepEqual(got, want) {
 		t.Fatalf("effectiveLLMProfileIDs() = %#v, want %#v", got, want)
 	}
@@ -2254,18 +2489,18 @@ func TestExampleFixtureFailureCoverage(t *testing.T) {
 			wantStderr: "completion unavailable",
 		},
 		{
-			name:             "malformed LLM response carried as raw output",
+			name:             "malformed LLM response rejected by validation",
 			args:             []string{"run", "dnd-session", "--config", configPath, "--input", inputPath},
 			factory:          fakeLLMFactory(newMalformedRunLLMClient(), nil),
 			wantCode:         0,
-			wantOutputStatus: "approved",
+			wantOutputStatus: "rejected",
 		},
 		{
-			name:             "invalid source reference raw output",
+			name:             "invalid source reference rejected by validation",
 			args:             []string{"run", "dnd-session", "--config", configPath, "--input", inputPath},
 			factory:          fakeLLMFactory(newFakeRunLLMClient(true), nil),
 			wantCode:         0,
-			wantOutputStatus: "approved",
+			wantOutputStatus: "rejected",
 		},
 	}
 
@@ -2424,6 +2659,18 @@ pipelines:
 `
 }
 
+func mvpConfigYAMLWithExtractValidators(pipelineID string, validators string) string {
+	return `version: 2
+pipelines:
+  ` + pipelineID + `:
+    input: seriatim
+    artifacts:
+      spells:
+        extract:
+          module: dnd/spells
+          validators:` + validators
+}
+
 func mvpConfigYAMLWithChunk(pipelineID string, chunker string, extractor string) string {
 	return `version: 2
 pipelines:
@@ -2886,6 +3133,25 @@ func resolvedArtifactLane(t *testing.T, resolved pipeline.ResolvedPipeline, lane
 	return pipeline.ResolvedArtifactLane{}
 }
 
+func manifestValidatorChain(t *testing.T, manifest artifacts.RunManifest, stage pipeline.ModuleStage, laneID string, module string) artifacts.ValidatorChainManifest {
+	t.Helper()
+	for _, chain := range manifest.ValidatorChains {
+		if chain.Stage == string(stage) && chain.LaneID == laneID && chain.ModuleKey == module {
+			return chain
+		}
+	}
+	t.Fatalf("validator chain %s/%s/%s not found in %#v", stage, laneID, module, manifest.ValidatorChains)
+	return artifacts.ValidatorChainManifest{}
+}
+
+func manifestValidatorKeys(chain artifacts.ValidatorChainManifest) []string {
+	keys := make([]string, 0, len(chain.Validators))
+	for _, validator := range chain.Validators {
+		keys = append(keys, validator.Key)
+	}
+	return keys
+}
+
 func assertNoTemporaryFiles(t *testing.T, root string) {
 	t.Helper()
 	if err := filepath.WalkDir(root, func(path string, entry os.DirEntry, err error) error {
@@ -2937,6 +3203,7 @@ func fakeCatalog(t *testing.T, overrides ...pipeline.ModuleSpec) pipeline.Module
 		Mergers:         mergers,
 		Normalizers:     normalizers,
 		Validators:      validators,
+		ValidatorChains: pipeline.NewValidatorChainRegistry(),
 		Outputs:         outputs,
 	}
 }
@@ -2983,6 +3250,32 @@ func mustRegisterOutput(t *testing.T, registry *pipeline.OutputEncoderRegistry,
 	}
 }
 
+func mustRegisterValidator(t *testing.T, registry *pipeline.ValidatorRegistry, spec pipeline.ValidatorSpec) {
+	t.Helper()
+	if err := registry.RegisterWithSpec(spec, func() (contracts.Validator, error) {
+		return fakeConfigValidator{name: spec.Key, executionClass: spec.ExecutionClass}, nil
+	}); err != nil {
+		t.Fatalf("register validator: %v", err)
+	}
+}
+
+type fakeConfigValidator struct {
+	name           string
+	executionClass contracts.ExecutionClass
+}
+
+func (validator fakeConfigValidator) Name() string {
+	return validator.name
+}
+
+func (validator fakeConfigValidator) ExecutionClass() contracts.ExecutionClass {
+	return validator.executionClass
+}
+
+func (validator fakeConfigValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	return contracts.ValidationResult{Approved: true}, nil
+}
+
 func mapLookup(values map[string]string) func(string) (string, bool) {
 	return func(key string) (string, bool) {
 		value, ok := values[key]

internal/core/artifacts/artifacts.go

@@ -1,47 +1,29 @@
 package artifacts
 
 import (
-	"encoding/json"
 	"time"
-
-	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
 )
 
-type ArtifactCandidate struct {
-	Index         int                `json:"index"`
-	ExtractorKey  string             `json:"extractor_key"`
-	ArtifactType  string             `json:"artifact_type"`
-	SchemaVersion string             `json:"schema_version"`
-	Payload       json.RawMessage    `json:"payload"`
-	SourceRefs    []source.SourceRef `json:"source_refs,omitempty"`
-	Metadata      map[string]any     `json:"metadata,omitempty"`
-}
-
-type Artifact struct {
-	ExtractorKey  string             `json:"extractor_key"`
-	ArtifactType  string             `json:"artifact_type"`
-	SchemaVersion string             `json:"schema_version"`
-	Payload       json.RawMessage    `json:"payload"`
-	SourceRefs    []source.SourceRef `json:"source_refs,omitempty"`
-	Metadata      map[string]any     `json:"metadata,omitempty"`
-}
-
-type RejectedArtifact struct {
-	Candidate     ArtifactCandidate `json:"candidate"`
-	ValidatorName string            `json:"validator_name"`
-	ReasonCode    string            `json:"reason_code"`
-	Message       string            `json:"message"`
-}
-
 type ArtifactLaneManifest struct {
 	ID         string         `json:"id"`
 	Extractor  string         `json:"extractor"`
 	Merger     string         `json:"merger"`
 	Normalizer string         `json:"normalizer"`
-	Validators []string       `json:"validators,omitempty"`
 	Metadata   map[string]any `json:"metadata,omitempty"`
 }
 
+type ValidatorChainManifest struct {
+	Stage      string              `json:"stage"`
+	LaneID     string              `json:"lane_id,omitempty"`
+	ModuleKey  string              `json:"module_key"`
+	Validators []ValidatorManifest `json:"validators"`
+}
+
+type ValidatorManifest struct {
+	Key            string `json:"key"`
+	ExecutionClass string `json:"execution_class"`
+}
+
 type LLMProfileManifest struct {
 	ID       string `json:"id"`
 	Provider string `json:"provider,omitempty"`
@@ -100,6 +82,7 @@ type RunManifest struct {
 	OutputEncoder     string                     `json:"output_encoder,omitempty"`
 	ModuleMetadata    map[string]map[string]any  `json:"module_metadata,omitempty"`
 	ArtifactLanes     []ArtifactLaneManifest     `json:"artifact_lanes,omitempty"`
+	ValidatorChains   []ValidatorChainManifest   `json:"validator_chains,omitempty"`
 	References        []ReferenceProvenance      `json:"references,omitempty"`
 	NormalizedOutputs []NormalizedOutputManifest `json:"normalized_outputs,omitempty"`
 	RejectedOutputs   []RejectedOutputManifest   `json:"rejected_outputs,omitempty"`
@@ -110,26 +93,3 @@ type RunManifest struct {
 	StartedAt         *time.Time                 `json:"started_at,omitempty"`
 	CompletedAt       *time.Time                 `json:"completed_at,omitempty"`
 }
-
-func ArtifactFromCandidate(candidate ArtifactCandidate) Artifact {
-	return Artifact{
-		ExtractorKey:  candidate.ExtractorKey,
-		ArtifactType:  candidate.ArtifactType,
-		SchemaVersion: candidate.SchemaVersion,
-		Payload:       append(json.RawMessage(nil), candidate.Payload...),
-		SourceRefs:    append([]source.SourceRef(nil), candidate.SourceRefs...),
-		Metadata:      copyMetadata(candidate.Metadata),
-	}
-}
-
-func copyMetadata(metadata map[string]any) map[string]any {
-	if len(metadata) == 0 {
-		return nil
-	}
-
-	copied := make(map[string]any, len(metadata))
-	for key, value := range metadata {
-		copied[key] = value
-	}
-	return copied
-}

internal/core/artifacts/artifacts_test.go

@@ -2,116 +2,9 @@ package artifacts
 
 import (
 	"encoding/json"
-	"reflect"
 	"testing"
-
-	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
 )
 
-func TestArtifactFromCandidatePreservesCandidateFields(t *testing.T) {
-	candidate := ArtifactCandidate{
-		Index:         7,
-		ExtractorKey:  "generic-extractor",
-		ArtifactType:  "generic-artifact",
-		SchemaVersion: "v1",
-		Payload:       json.RawMessage(`{"name":"example"}`),
-		SourceRefs: []source.SourceRef{
-			{SourceID: "source-1", StartUnitID: 1, EndUnitID: 2},
-		},
-		Metadata: map[string]any{
-			"confidence": 0.75,
-		},
-	}
-
-	artifact := ArtifactFromCandidate(candidate)
-
-	if artifact.ExtractorKey != candidate.ExtractorKey {
-		t.Fatalf("ExtractorKey = %q, want %q", artifact.ExtractorKey, candidate.ExtractorKey)
-	}
-	if artifact.ArtifactType != candidate.ArtifactType {
-		t.Fatalf("ArtifactType = %q, want %q", artifact.ArtifactType, candidate.ArtifactType)
-	}
-	if artifact.SchemaVersion != candidate.SchemaVersion {
-		t.Fatalf("SchemaVersion = %q, want %q", artifact.SchemaVersion, candidate.SchemaVersion)
-	}
-	if string(artifact.Payload) != string(candidate.Payload) {
-		t.Fatalf("Payload = %s, want %s", artifact.Payload, candidate.Payload)
-	}
-	if !reflect.DeepEqual(artifact.SourceRefs, candidate.SourceRefs) {
-		t.Fatalf("SourceRefs = %#v, want %#v", artifact.SourceRefs, candidate.SourceRefs)
-	}
-	if !reflect.DeepEqual(artifact.Metadata, candidate.Metadata) {
-		t.Fatalf("Metadata = %#v, want %#v", artifact.Metadata, candidate.Metadata)
-	}
-
-	candidate.Payload[0] = '['
-	candidate.SourceRefs[0].StartUnitID = 99
-	candidate.Metadata["confidence"] = 0.5
-
-	if string(artifact.Payload) != `{"name":"example"}` {
-		t.Fatalf("Payload changed after candidate mutation: %s", artifact.Payload)
-	}
-	if artifact.SourceRefs[0].StartUnitID != 1 {
-		t.Fatalf("SourceRefs changed after candidate mutation: %#v", artifact.SourceRefs)
-	}
-	if artifact.Metadata["confidence"] != 0.75 {
-		t.Fatalf("Metadata changed after candidate mutation: %#v", artifact.Metadata)
-	}
-}
-
-func TestJSONMarshalUsesExpectedFieldNames(t *testing.T) {
-	candidate := ArtifactCandidate{
-		Index:         1,
-		ExtractorKey:  "generic-extractor",
-		ArtifactType:  "generic-artifact",
-		SchemaVersion: "v1",
-		Payload:       json.RawMessage(`{"value":true}`),
-		SourceRefs: []source.SourceRef{
-			{SourceID: "source-1", StartUnitID: 1, EndUnitID: 1},
-		},
-		Metadata: map[string]any{
-			"reviewed": true,
-		},
-	}
-	rejected := RejectedArtifact{
-		Candidate:     candidate,
-		ValidatorName: "generic-validator",
-		ReasonCode:    "invalid",
-		Message:       "candidate was not accepted",
-	}
-
-	gotJSON, err := json.Marshal(rejected)
-	if err != nil {
-		t.Fatalf("json.Marshal() error = %v", err)
-	}
-
-	var got map[string]any
-	if err := json.Unmarshal(gotJSON, &got); err != nil {
-		t.Fatalf("json.Unmarshal() error = %v", err)
-	}
-
-	assertHasKeys(t, got, "candidate", "validator_name", "reason_code", "message")
-
-	gotCandidate, ok := got["candidate"].(map[string]any)
-	if !ok {
-		t.Fatalf("candidate = %#v, want object", got["candidate"])
-	}
-	assertHasKeys(t, gotCandidate, "index", "extractor_key", "artifact_type", "schema_version", "payload", "source_refs", "metadata")
-
-	gotRefs, ok := gotCandidate["source_refs"].([]any)
-	if !ok {
-		t.Fatalf("source_refs = %#v, want array", gotCandidate["source_refs"])
-	}
-	if len(gotRefs) != 1 {
-		t.Fatalf("len(source_refs) = %d, want 1", len(gotRefs))
-	}
-	gotRef, ok := gotRefs[0].(map[string]any)
-	if !ok {
-		t.Fatalf("source_refs[0] = %#v, want object", gotRefs[0])
-	}
-	assertHasKeys(t, gotRef, "source_id", "start_unit_id", "end_unit_id")
-}
-
 func TestRunManifestOmitsEmptyOptionalFields(t *testing.T) {
 	gotJSON, err := json.Marshal(RunManifest{})
 	if err != nil {
@@ -136,12 +29,21 @@ func TestRunManifestIncludesPipelineAndArtifactLaneFields(t *testing.T) {
 				Extractor:  "event-extractor",
 				Merger:     "appendorder",
 				Normalizer: "noop",
-				Validators: []string{"grounded"},
 				Metadata: map[string]any{
 					"extractor": map[string]any{"prompt_id": "test.prompt"},
 				},
 			},
 		},
+		ValidatorChains: []ValidatorChainManifest{
+			{
+				Stage:     "extract",
+				LaneID:    "events",
+				ModuleKey: "event-extractor",
+				Validators: []ValidatorManifest{
+					{Key: "grounded", ExecutionClass: "deterministic"},
+				},
+			},
+		},
 	}
 
 	gotJSON, err := json.Marshal(manifest)
@@ -154,7 +56,7 @@ func TestRunManifestIncludesPipelineAndArtifactLaneFields(t *testing.T) {
 		t.Fatalf("json.Unmarshal() error = %v", err)
 	}
 
-	assertHasKeys(t, got, "pipeline_id", "pipeline_digest", "artifact_lanes", "llm_profiles")
+	assertHasKeys(t, got, "pipeline_id", "pipeline_digest", "artifact_lanes", "validator_chains", "llm_profiles")
 
 	profiles, ok := got["llm_profiles"].([]any)
 	if !ok {
@@ -180,7 +82,20 @@ func TestRunManifestIncludesPipelineAndArtifactLaneFields(t *testing.T) {
 	if !ok {
 		t.Fatalf("artifact_lanes[0] = %#v, want object", lanes[0])
 	}
-	assertHasKeys(t, lane, "id", "extractor", "merger", "normalizer", "validators", "metadata")
+	assertHasKeys(t, lane, "id", "extractor", "merger", "normalizer", "metadata")
+
+	chains, ok := got["validator_chains"].([]any)
+	if !ok {
+		t.Fatalf("validator_chains = %#v, want array", got["validator_chains"])
+	}
+	if len(chains) != 1 {
+		t.Fatalf("len(validator_chains) = %d, want 1", len(chains))
+	}
+	chain, ok := chains[0].(map[string]any)
+	if !ok {
+		t.Fatalf("validator_chains[0] = %#v, want object", chains[0])
+	}
+	assertHasKeys(t, chain, "stage", "lane_id", "module_key", "validators")
 }
 
 func TestRunManifestIncludesReferenceProvenance(t *testing.T) {

internal/core/config/config.go

@@ -97,6 +97,18 @@ func cloneModuleBinding(in pipeline.ModuleBinding) pipeline.ModuleBinding {
 		out.Options = cloneOptions(in.Options)
 	}
 	out.References = cloneStringMap(in.References)
+	out.Validators = cloneValidatorOverride(in.Validators)
+	return out
+}
+
+func cloneValidatorOverride(in pipeline.ValidatorOverride) pipeline.ValidatorOverride {
+	out := pipeline.ValidatorOverride{Set: in.Set}
+	if len(in.Validators) > 0 {
+		out.Validators = make([]pipeline.ModuleBinding, len(in.Validators))
+		for i, binding := range in.Validators {
+			out.Validators[i] = cloneModuleBinding(binding)
+		}
+	}
 	return out
 }
 

internal/core/config/effective_config_test.go

@@ -161,6 +161,12 @@ func TestResolveLLMProfileOverrideAppliesBeforeDigest(t *testing.T) {
 	profile.Output.LLMProfile = "output-profile"
 	lane := profile.Artifacts["events"]
 	lane.Merge.LLMProfile = "merge-profile"
+	lane.Extract.Validators = pipeline.ValidatorOverride{
+		Set: true,
+		Validators: []pipeline.ModuleBinding{
+			{Module: "fake/llm-validator", LLMProfile: "validator-profile"},
+		},
+	}
 	profile.Artifacts["events"] = lane
 	cfg.Pipelines["example"] = profile
 
@@ -195,9 +201,22 @@ func TestResolveLLMProfileOverrideAppliesBeforeDigest(t *testing.T) {
 	if eventLane.Merge.LLMProfile != "runtime" {
 		t.Fatalf("merge profile = %q, want runtime", eventLane.Merge.LLMProfile)
 	}
-	if len(eventLane.Validators) != 0 {
+	validatorChain := findEffectiveValidatorChain(effective.ResolvedPipeline.ValidatorChains, pipeline.StageExtract, "events", "fake/extract")
-		t.Fatalf("validator profiles = %#v, want none", eventLane.Validators)
+	if validatorChain == nil || len(validatorChain.Validators) != 1 {
+		t.Fatalf("validator chain = %#v, want one extract validator", effective.ResolvedPipeline.ValidatorChains)
 	}
+	if validatorChain.Validators[0].Binding.LLMProfile != "validator-profile" {
+		t.Fatalf("validator profile = %q, want original validator-profile", validatorChain.Validators[0].Binding.LLMProfile)
+	}
+}
+
+func findEffectiveValidatorChain(chains []pipeline.ResolvedValidatorChain, stage pipeline.ModuleStage, laneID string, module string) *pipeline.ResolvedValidatorChain {
+	for i := range chains {
+		if chains[i].Stage == stage && chains[i].LaneID == laneID && chains[i].ModuleKey == module {
+			return &chains[i]
+		}
+	}
+	return nil
 }
 
 func llmCapableBindings(resolved pipeline.ResolvedPipeline) []pipeline.ModuleBinding {

internal/core/config/file_config.go

@@ -56,6 +56,7 @@ type fileModuleBinding struct {
 	Retries    int
 	Options    map[string]any
 	References map[string]string
+	Validators pipeline.ValidatorOverride
 }
 
 func (b *fileModuleBinding) UnmarshalYAML(node *yaml.Node) error {
@@ -102,6 +103,16 @@ func (b *fileModuleBinding) UnmarshalYAML(node *yaml.Node) error {
 					return err
 				}
 				b.References = references
+			case "validators":
+				b.Validators.Set = true
+				var validators []fileModuleBinding
+				if err := valueNode.Decode(&validators); err != nil {
+					return err
+				}
+				b.Validators.Validators = make([]pipeline.ModuleBinding, len(validators))
+				for i, validator := range validators {
+					b.Validators.Validators[i] = validator.toPipelineBinding()
+				}
 			default:
 				return fmt.Errorf("field %s not found in module binding", keyNode.Value)
 			}
@@ -119,6 +130,7 @@ func (b fileModuleBinding) toPipelineBinding() pipeline.ModuleBinding {
 		Retries:    b.Retries,
 		Options:    cloneOptions(b.Options),
 		References: normalizedStringMap(b.References),
+		Validators: b.Validators,
 	}
 }
 

internal/core/config/file_config_test.go

@@ -300,6 +300,61 @@ pipelines:
 	}
 }
 
+func TestParseFileConfigStageLocalValidatorOverrides(t *testing.T) {
+	cfg := parseAndApplyConfig(t, `
+version: 2
+pipelines:
+  example:
+    input: fake/input
+    chunk:
+      module: generic
+      validators: []
+    artifacts:
+      events:
+        extract:
+          module: fake/extract
+          validators:
+            - fake/validator
+            - module: fake/llm-validator
+              llm_profile: careful
+              options:
+                threshold: 0.7
+        merge:
+          module: appendorder
+          validators: []
+        normalize:
+          module: noop
+`)
+
+	profile := cfg.Pipelines["example"]
+	if !profile.Chunk.Validators.Set || len(profile.Chunk.Validators.Validators) != 0 {
+		t.Fatalf("chunk validator override = %#v, want explicit empty", profile.Chunk.Validators)
+	}
+	lane := profile.Artifacts["events"]
+	if !lane.Extract.Validators.Set {
+		t.Fatalf("extract validator override Set = false, want true")
+	}
+	validators := lane.Extract.Validators.Validators
+	if len(validators) != 2 {
+		t.Fatalf("extract validators = %#v, want two validators", validators)
+	}
+	if validators[0].Module != "fake/validator" {
+		t.Fatalf("first validator = %#v, want fake/validator", validators[0])
+	}
+	if validators[1].Module != "fake/llm-validator" || validators[1].LLMProfile != "careful" {
+		t.Fatalf("second validator = %#v, want LLM validator with profile", validators[1])
+	}
+	if validators[1].Options["threshold"] != 0.7 {
+		t.Fatalf("second validator options = %#v, want threshold", validators[1].Options)
+	}
+	if !lane.Merge.Validators.Set || len(lane.Merge.Validators.Validators) != 0 {
+		t.Fatalf("merge validator override = %#v, want explicit empty", lane.Merge.Validators)
+	}
+	if lane.Normalize.Validators.Set {
+		t.Fatalf("normalize validator override Set = true, want omitted")
+	}
+}
+
 func TestApplyFileConfigRejectsDuplicateTrimmedPipelineIDs(t *testing.T) {
 	fileCfg, err := ParseFileConfigYAML([]byte(`
 version: 2

internal/core/config/redaction.go

@@ -27,6 +27,12 @@ func cloneResolvedPipeline(in pipeline.ResolvedPipeline) pipeline.ResolvedPipeli
 	out.Chunk = cloneModuleBinding(in.Chunk)
 	out.ChunkReferences = pipeline.CloneReferenceTarget(in.ChunkReferences)
 	out.Output = cloneModuleBinding(in.Output)
+	if len(in.ValidatorChains) > 0 {
+		out.ValidatorChains = make([]pipeline.ResolvedValidatorChain, len(in.ValidatorChains))
+		for i, chain := range in.ValidatorChains {
+			out.ValidatorChains[i] = cloneResolvedValidatorChain(chain)
+		}
+	}
 	if len(in.ArtifactLanes) > 0 {
 		out.ArtifactLanes = make([]pipeline.ResolvedArtifactLane, len(in.ArtifactLanes))
 		for i, lane := range in.ArtifactLanes {
@@ -36,6 +42,20 @@ func cloneResolvedPipeline(in pipeline.ResolvedPipeline) pipeline.ResolvedPipeli
 	return out
 }
 
+func cloneResolvedValidatorChain(in pipeline.ResolvedValidatorChain) pipeline.ResolvedValidatorChain {
+	out := in
+	if len(in.Validators) > 0 {
+		out.Validators = make([]pipeline.ResolvedValidator, len(in.Validators))
+		for i, validator := range in.Validators {
+			out.Validators[i] = pipeline.ResolvedValidator{
+				Binding:        cloneModuleBinding(validator.Binding),
+				ExecutionClass: validator.ExecutionClass,
+			}
+		}
+	}
+	return out
+}
+
 func cloneResolvedArtifactLane(in pipeline.ResolvedArtifactLane) pipeline.ResolvedArtifactLane {
 	out := in
 	out.Extract = cloneModuleBinding(in.Extract)

internal/core/config/validation.go

@@ -85,7 +85,7 @@ func validatePipelineProfiles(profiles map[string]pipeline.PipelineProfile) erro
 				return err
 			}
 			if len(lane.Validators) > 0 {
-				return fmt.Errorf("pipeline %q lane %q validators are not supported by the current raw validation runner", id, laneID)
+				return fmt.Errorf("pipeline %q lane %q validators are not supported at artifact lane level; use extract.validators, merge.validators, or normalize.validators", id, laneID)
 			}
 		}
 	}
@@ -108,6 +108,9 @@ func validateBinding(
 		}
 		return fmt.Errorf("pipeline %q %s retries must be greater than or equal to zero", pipelineID, slot)
 	}
+	if err := validateValidatorOverride(pipelineID, laneID, slot, binding.Validators); err != nil {
+		return err
+	}
 	if len(binding.References) == 0 {
 		return nil
 	}
@@ -120,6 +123,36 @@ func validateBinding(
 	return validateReferenceMapForContext(pipelineID, laneID, slot, binding.References)
 }
 
+func validateValidatorOverride(pipelineID string, laneID string, slot string, override pipeline.ValidatorOverride) error {
+	if !override.Set {
+		return nil
+	}
+	switch slot {
+	case "chunk", "extract", "merge", "normalize":
+	default:
+		return fmt.Errorf("%s validators are not supported", referenceContext(pipelineID, laneID, slot))
+	}
+	for i, validator := range override.Validators {
+		context := fmt.Sprintf("%s validators[%d]", referenceContext(pipelineID, laneID, slot), i)
+		if strings.TrimSpace(validator.Module) == "" {
+			return fmt.Errorf("%s module must not be empty", context)
+		}
+		if len(validator.References) > 0 {
+			return fmt.Errorf("%s references are not supported", context)
+		}
+		if validator.Validators.Set {
+			return fmt.Errorf("%s nested validators are not supported", context)
+		}
+		if validator.Retries != 0 {
+			return fmt.Errorf("%s retries are not supported", context)
+		}
+		if validator.LLMProfile != "" && strings.TrimSpace(validator.LLMProfile) == "" {
+			return fmt.Errorf("%s llm_profile must not be empty when set", context)
+		}
+	}
+	return nil
+}
+
 func validateReferenceMap(pipelineID string, laneID string, references map[string]string) error {
 	return validateReferenceMapForContext(pipelineID, laneID, "", references)
 }

internal/core/config/validation_test.go

@@ -341,13 +341,82 @@ func TestValidateRejectsConfiguredValidators(t *testing.T) {
 	if err == nil {
 		t.Fatal("Validate() error = nil, want configured validators error")
 	}
-	for _, want := range []string{"example", "events", "validators", "not supported"} {
+	for _, want := range []string{"example", "events", "validators", "extract.validators", "merge.validators", "normalize.validators"} {
 		if !strings.Contains(err.Error(), want) {
 			t.Fatalf("Validate() error = %q, want substring %q", err.Error(), want)
 		}
 	}
 }
 
+func TestValidateAcceptsStageLocalValidatorOverrides(t *testing.T) {
+	cfg := validConfig()
+	profile := cfg.Pipelines["example"]
+	profile.Chunk.Validators = pipeline.ValidatorOverride{Set: true}
+	lane := profile.Artifacts["events"]
+	lane.Extract.Validators = pipeline.ValidatorOverride{
+		Set: true,
+		Validators: []pipeline.ModuleBinding{
+			pipeline.Binding("fake/validator"),
+			{Module: "fake/llm-validator", LLMProfile: "careful", Options: map[string]any{"threshold": 0.7}},
+		},
+	}
+	lane.Merge.Validators = pipeline.ValidatorOverride{Set: true}
+	profile.Artifacts["events"] = lane
+	cfg.Pipelines["example"] = profile
+
+	if err := cfg.Validate(); err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+}
+
+func TestValidateRejectsInvalidValidatorBindings(t *testing.T) {
+	tests := []struct {
+		name    string
+		binding pipeline.ModuleBinding
+		want    string
+	}{
+		{
+			name:    "empty module",
+			binding: pipeline.ModuleBinding{},
+			want:    "module must not be empty",
+		},
+		{
+			name:    "references",
+			binding: pipeline.ModuleBinding{Module: "fake/validator", References: map[string]string{"roster": "./roster.txt"}},
+			want:    "references are not supported",
+		},
+		{
+			name:    "nested validators",
+			binding: pipeline.ModuleBinding{Module: "fake/validator", Validators: pipeline.ValidatorOverride{Set: true}},
+			want:    "nested validators are not supported",
+		},
+		{
+			name:    "retries",
+			binding: pipeline.ModuleBinding{Module: "fake/validator", Retries: 1},
+			want:    "retries are not supported",
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			cfg := validConfig()
+			profile := cfg.Pipelines["example"]
+			lane := profile.Artifacts["events"]
+			lane.Extract.Validators = pipeline.ValidatorOverride{
+				Set:        true,
+				Validators: []pipeline.ModuleBinding{test.binding},
+			}
+			profile.Artifacts["events"] = lane
+			cfg.Pipelines["example"] = profile
+
+			err := cfg.Validate()
+			if err == nil || !strings.Contains(err.Error(), test.want) {
+				t.Fatalf("Validate() error = %v, want %q", err, test.want)
+			}
+		})
+	}
+}
+
 func validConfig() Config {
 	cfg := Default()
 	cfg.Pipelines["example"] = pipeline.PipelineProfile{
@@ -402,6 +471,12 @@ func fakeCatalog(t *testing.T, overrides ...pipeline.ModuleSpec) pipeline.Module
 			Requires: []string{"normalized"},
 			Provides: []string{"validated"},
 		},
+		"fake/llm-validator": {
+			Key:      "fake/llm-validator",
+			Stage:    pipeline.StageValidate,
+			Requires: []string{"normalized"},
+			Provides: []string{"validated"},
+		},
 		"json": {
 			Key:      "json",
 			Stage:    pipeline.StageOutput,
@@ -426,6 +501,7 @@ func fakeCatalog(t *testing.T, overrides ...pipeline.ModuleSpec) pipeline.Module
 	mustRegisterMerger(t, mergers, specs["appendorder"])
 	mustRegisterNormalizer(t, normalizers, specs["noop"])
 	mustRegisterValidator(t, validators, specs["fake/validator"])
+	mustRegisterValidator(t, validators, specs["fake/llm-validator"])
 	mustRegisterOutput(t, outputs, specs["json"])
 
 	return pipeline.ModuleCatalog{
@@ -435,6 +511,7 @@ func fakeCatalog(t *testing.T, overrides ...pipeline.ModuleSpec) pipeline.Module
 		Mergers:         mergers,
 		Normalizers:     normalizers,
 		Validators:      validators,
+		ValidatorChains: pipeline.NewValidatorChainRegistry(),
 		Outputs:         outputs,
 	}
 }
@@ -476,7 +553,12 @@ func mustRegisterNormalizer(t *testing.T, registry *pipeline.NormalizerRegistry,
 
 func mustRegisterValidator(t *testing.T, registry *pipeline.ValidatorRegistry, spec pipeline.ModuleSpec) {
 	t.Helper()
-	if err := registry.RegisterWithSpec(spec, func() (contracts.Validator, error) { return nil, nil }); err != nil {
+	executionClass := contracts.ExecutionClassDeterministic
+	if spec.Key == "fake/llm-validator" {
+		executionClass = contracts.ExecutionClassLLMBacked
+	}
+	validatorSpec := pipeline.ValidatorSpec{Key: spec.Key, ExecutionClass: executionClass}
+	if err := registry.RegisterWithSpec(validatorSpec, func() (contracts.Validator, error) { return nil, nil }); err != nil {
 		t.Fatalf("register validator: %v", err)
 	}
 }

internal/framework/contracts/composition_test.go

@@ -255,20 +255,15 @@ func (validator compositionValidator) Name() string {
 	return "generic-validator"
 }
 
-func (validator compositionValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+func (validator compositionValidator) ExecutionClass() contracts.ExecutionClass {
-	decisions := make([]contracts.ValidationDecision, 0, len(req.Candidates))
+	return contracts.ExecutionClassDeterministic
-	for _, candidate := range req.Candidates {
-		decisions = append(decisions, contracts.ValidationDecision{
-			CandidateIndex: candidate.Index,
-			Approved:       true,
-			ReasonCode:     "accepted",
-			Message:        "candidate accepted",
-		})
 }
 
+func (validator compositionValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
 	return contracts.ValidationResult{
-		ValidatorName: validator.Name(),
+		Approved:   true,
-		Decisions:     decisions,
+		ReasonCode: "accepted",
+		Message:    "output accepted",
 	}, nil
 }
 

internal/framework/contracts/contracts.go

@@ -203,20 +203,37 @@ type RawPayload struct {
 	Warnings  []Warning      `json:"warnings,omitempty"`
 }
 
-type RawValidationRequest struct {
+type ExecutionClass string
+
+const (
+	ExecutionClassDeterministic ExecutionClass = "deterministic"
+	ExecutionClassLLMBacked     ExecutionClass = "llm_backed"
+)
+
+type ValidationRequest struct {
 	Stage          string                 `json:"stage"`
 	LaneID         string                 `json:"lane_id,omitempty"`
 	ModuleKey      string                 `json:"module_key"`
 	Source         *source.SourceDocument `json:"-"`
 	SourceID       string                 `json:"source_id,omitempty"`
-	ChunkID    string                 `json:"chunk_id,omitempty"`
+	SourceInput    LLMInputMaterial       `json:"source_input,omitempty"`
-	ChunkIndex int                    `json:"chunk_index,omitempty"`
+	SessionID      string                 `json:"session_id,omitempty"`
+	References     ReferenceSet           `json:"references,omitempty"`
+	LLMClient      StructuredLLMClient    `json:"-"`
+	LLMProfile     string                 `json:"llm_profile,omitempty"`
+	Options        map[string]any         `json:"options,omitempty"`
+	Metadata       map[string]any         `json:"metadata,omitempty"`
 	Schema         ResponseSchema         `json:"schema,omitempty"`
 	Payload        RawPayload             `json:"payload"`
-	Metadata   map[string]any         `json:"metadata,omitempty"`
+	ChunkID        string                 `json:"chunk_id,omitempty"`
+	ChunkIndex     int                    `json:"chunk_index,omitempty"`
+	Chunk          *SourceChunk           `json:"chunk,omitempty"`
+	Chunks         []SourceChunk          `json:"chunks,omitempty"`
+	ExtractOutputs []ExtractOutput        `json:"extract_outputs,omitempty"`
+	MergeOutput    MergeOutput            `json:"merge_output,omitempty"`
 }
 
-type RawValidationResult struct {
+type ValidationResult struct {
 	Approved               bool      `json:"approved"`
 	ReasonCode             string    `json:"reason_code,omitempty"`
 	Message                string    `json:"message,omitempty"`
@@ -224,15 +241,17 @@ type RawValidationResult struct {
 	Warnings               []Warning `json:"warnings,omitempty"`
 }
 
-type RawValidator interface {
+type Validator interface {
 	Name() string
-	ValidateRaw(ctx context.Context, req RawValidationRequest) (RawValidationResult, error)
+	ExecutionClass() ExecutionClass
+	Validate(ctx context.Context, req ValidationRequest) (ValidationResult, error)
 }
 
 type ResponseSchema struct {
 	ID         string `json:"id,omitempty"`
 	Name       string `json:"name,omitempty"`
 	Version    string `json:"version,omitempty"`
+	JSONSchema []byte `json:"-"`
 }
 
 type ExtractOutput struct {
@@ -308,33 +327,6 @@ type Normalizer interface {
 	Normalize(ctx context.Context, req NormalizeRequest) (NormalizeResult, error)
 }
 
-type ValidationRequest struct {
-	Source     *source.SourceDocument        `json:"-"`
-	Candidates []artifacts.ArtifactCandidate `json:"candidates"`
-	LLMProfile string                        `json:"llm_profile,omitempty"`
-	Options    map[string]any                `json:"options,omitempty"`
-	Metadata   map[string]any                `json:"metadata,omitempty"`
-}
-
-type ValidationDecision struct {
-	CandidateIndex         int    `json:"candidate_index"`
-	Approved               bool   `json:"approved"`
-	ReasonCode             string `json:"reason_code"`
-	Message                string `json:"message"`
-	DiagnosticArtifactPath string `json:"diagnostic_artifact_path,omitempty"`
-}
-
-type ValidationResult struct {
-	ValidatorName string               `json:"validator_name"`
-	Decisions     []ValidationDecision `json:"decisions"`
-	Warnings      []Warning            `json:"warnings,omitempty"`
-}
-
-type Validator interface {
-	Name() string
-	Validate(ctx context.Context, req ValidationRequest) (ValidationResult, error)
-}
-
 type Warning struct {
 	Scope      string `json:"scope,omitempty"`
 	ReasonCode string `json:"reason_code"`

internal/framework/contracts/contracts_test.go

@@ -322,6 +322,32 @@ func TestLLMInputSetCloneCopiesContent(t *testing.T) {
 	}
 }
 
+func TestResponseSchemaJSONOmitRawSchemaContent(t *testing.T) {
+	schema := ResponseSchema{
+		ID:         "schema-id",
+		Name:       "schema-name",
+		Version:    "v1",
+		JSONSchema: []byte(`{"type":"object"}`),
+	}
+	encoded, err := json.Marshal(schema)
+	if err != nil {
+		t.Fatalf("json.Marshal() error = %v, want nil", err)
+	}
+	var got map[string]any
+	if err := json.Unmarshal(encoded, &got); err != nil {
+		t.Fatalf("json.Unmarshal() error = %v, want nil", err)
+	}
+	if got["id"] != "schema-id" || got["name"] != "schema-name" || got["version"] != "v1" {
+		t.Fatalf("encoded schema = %#v, want schema provenance", got)
+	}
+	if _, ok := got["json_schema"]; ok {
+		t.Fatalf("encoded schema leaked raw schema content: %s", encoded)
+	}
+	if _, ok := got["JSONSchema"]; ok {
+		t.Fatalf("encoded schema leaked raw schema content: %s", encoded)
+	}
+}
+
 func TestFakeMergeNormalizeAndOutputContracts(t *testing.T) {
 	extractOutput := ExtractOutput{
 		LaneID:       "generic-lane",
@@ -577,20 +603,15 @@ func (validator fakeValidator) Name() string {
 	return validator.name
 }
 
-func (validator fakeValidator) Validate(ctx context.Context, req ValidationRequest) (ValidationResult, error) {
+func (validator fakeValidator) ExecutionClass() ExecutionClass {
-	decisions := make([]ValidationDecision, 0, len(req.Candidates))
+	return ExecutionClassDeterministic
-	for _, candidate := range req.Candidates {
-		decisions = append(decisions, ValidationDecision{
-			CandidateIndex: candidate.Index,
-			Approved:       true,
-			ReasonCode:     "accepted",
-			Message:        "candidate accepted",
-		})
 }
 
+func (validator fakeValidator) Validate(ctx context.Context, req ValidationRequest) (ValidationResult, error) {
 	return ValidationResult{
-		ValidatorName: validator.name,
+		Approved:   true,
-		Decisions:     decisions,
+		ReasonCode: "accepted",
+		Message:    "output accepted",
 	}, nil
 }
 

internal/framework/pipeline/chunker_registry_test.go

@@ -386,6 +386,10 @@ func (validator registryValidator) Name() string {
 	return validator.name
 }
 
-func (validator registryValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+func (validator registryValidator) ExecutionClass() contracts.ExecutionClass {
-	return contracts.ValidationResult{}, nil
+	return contracts.ExecutionClassDeterministic
+}
+
+func (validator registryValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	return contracts.ValidationResult{Approved: true}, nil
 }

internal/framework/pipeline/default_modules_test.go

@@ -97,6 +97,7 @@ func defaultModuleCatalog(t *testing.T) pipeline.ModuleCatalog {
 		Extractors:      extractors,
 		Mergers:         mergers,
 		Normalizers:     normalizers,
+		ValidatorChains: pipeline.NewValidatorChainRegistry(),
 		Outputs:         outputs,
 	}
 }

internal/framework/pipeline/profile.go

@@ -25,6 +25,35 @@ type ModuleBinding struct {
 	Retries    int               `json:"retries,omitempty"`
 	Options    map[string]any    `json:"options,omitempty"`
 	References map[string]string `json:"references,omitempty"`
+	Validators ValidatorOverride `json:"validators,omitempty"`
+}
+
+type ValidatorOverride struct {
+	Set        bool            `json:"set,omitempty"`
+	Validators []ModuleBinding `json:"validators,omitempty"`
+}
+
+func (binding ModuleBinding) MarshalJSON() ([]byte, error) {
+	type moduleBindingJSON struct {
+		Module     string            `json:"module"`
+		LLMProfile string            `json:"llm_profile,omitempty"`
+		Retries    int               `json:"retries,omitempty"`
+		Options    map[string]any    `json:"options,omitempty"`
+		References map[string]string `json:"references,omitempty"`
+		Validators *[]ModuleBinding  `json:"validators,omitempty"`
+	}
+	out := moduleBindingJSON{
+		Module:     binding.Module,
+		LLMProfile: binding.LLMProfile,
+		Retries:    binding.Retries,
+		Options:    binding.Options,
+		References: binding.References,
+	}
+	if binding.Validators.Set {
+		validators := cloneModuleBindings(binding.Validators.Validators)
+		out.Validators = &validators
+	}
+	return json.Marshal(out)
 }
 
 type ArtifactLaneProfile struct {
@@ -83,6 +112,18 @@ type ResolvedArtifactLane struct {
 	NormalizeReferences ResolvedReferenceTarget `json:"normalize_references"`
 }
 
+type ResolvedValidatorChain struct {
+	Stage      ModuleStage         `json:"stage"`
+	LaneID     string              `json:"lane_id,omitempty"`
+	ModuleKey  string              `json:"module_key"`
+	Validators []ResolvedValidator `json:"validators"`
+}
+
+type ResolvedValidator struct {
+	Binding        ModuleBinding            `json:"binding"`
+	ExecutionClass contracts.ExecutionClass `json:"execution_class"`
+}
+
 type ResolvedPipeline struct {
 	ID              string
 	Digest          string
@@ -90,6 +131,7 @@ type ResolvedPipeline struct {
 	Chunk           ModuleBinding
 	ChunkReferences ResolvedReferenceTarget `json:"chunk_references"`
 	ArtifactLanes   []ResolvedArtifactLane
+	ValidatorChains []ResolvedValidatorChain `json:"validator_chains"`
 	Output          ModuleBinding
 }
 
@@ -100,6 +142,7 @@ type ModuleCatalog struct {
 	Mergers         *MergerRegistry
 	Normalizers     *NormalizerRegistry
 	Validators      *ValidatorRegistry
+	ValidatorChains *ValidatorChainRegistry
 	Outputs         *OutputEncoderRegistry
 }
 
@@ -172,15 +215,21 @@ func ResolvePipeline(profile PipelineProfile, options ResolveOptions, catalog Mo
 		ChunkReferences: referenceTarget(StageChunk, "", chunk.Module, chunkReferences),
 		Output:          resolveBinding(profile.Output, DefaultOutputModule),
 	}
+	chunkValidatorChain, err := resolveValidatorChain(pipelineID, "", StageChunk, chunk.Module, chunk.Validators, catalog)
+	if err != nil {
+		return ResolvedPipeline{}, err
+	}
+	resolved.ValidatorChains = append(resolved.ValidatorChains, chunkValidatorChain)
 	outputCapabilities := capabilities.clone()
 
 	for _, laneID := range selectedLaneIDs {
 		laneProfile := lanesByID[laneID]
-		lane, laneCapabilities, err := resolveArtifactLane(pipelineID, laneID, laneProfile, profile.References, options, capabilities, catalog)
+		lane, validatorChains, laneCapabilities, err := resolveArtifactLane(pipelineID, laneID, laneProfile, profile.References, options, capabilities, catalog)
 		if err != nil {
 			return ResolvedPipeline{}, err
 		}
 		resolved.ArtifactLanes = append(resolved.ArtifactLanes, lane)
+		resolved.ValidatorChains = append(resolved.ValidatorChains, validatorChains...)
 		outputCapabilities.addSet(laneCapabilities)
 	}
 
@@ -208,7 +257,7 @@ func resolveArtifactLane(
 	options ResolveOptions,
 	inherited capabilitySet,
 	catalog ModuleCatalog,
-) (ResolvedArtifactLane, capabilitySet, error) {
+) (ResolvedArtifactLane, []ResolvedValidatorChain, capabilitySet, error) {
 	lane := ResolvedArtifactLane{
 		ID:         laneID,
 		Extract:    resolveBinding(profile.Extract, ""),
@@ -217,17 +266,17 @@ func resolveArtifactLane(
 		Validators: resolveBindings(profile.Validators, ""),
 	}
 	if lane.Extract.Module == "" {
-		return ResolvedArtifactLane{}, nil, fmt.Errorf("pipeline %q lane %q extract module must not be empty", pipelineID, laneID)
+		return ResolvedArtifactLane{}, nil, nil, fmt.Errorf("pipeline %q lane %q extract module must not be empty", pipelineID, laneID)
 	}
 
 	capabilities := inherited.clone()
 
 	extractSpec, err := extractorSpec(catalog, lane.Extract.Module)
 	if err != nil {
-		return ResolvedArtifactLane{}, nil, moduleLookupError(pipelineID, laneID, StageExtract, lane.Extract.Module, err)
+		return ResolvedArtifactLane{}, nil, nil, moduleLookupError(pipelineID, laneID, StageExtract, lane.Extract.Module, err)
 	}
 	if missing, ok := capabilities.missing(extractSpec.Requires); ok {
-		return ResolvedArtifactLane{}, nil, capabilityError(pipelineID, laneID, StageExtract, lane.Extract.Module, missing)
+		return ResolvedArtifactLane{}, nil, nil, capabilityError(pipelineID, laneID, StageExtract, lane.Extract.Module, missing)
 	}
 	extractReferences := mergeReferenceMaps(profile.References, lane.Extract.References)
 	references, err := resolveReferenceTargetBindings(referenceResolutionTarget{
@@ -241,17 +290,17 @@ func resolveArtifactLane(
 		Options:            options,
 	})
 	if err != nil {
-		return ResolvedArtifactLane{}, nil, err
+		return ResolvedArtifactLane{}, nil, nil, err
 	}
 	lane.ExtractReferences = referenceTarget(StageExtract, laneID, lane.Extract.Module, references)
 	capabilities.add(extractSpec.Provides...)
 
 	mergeSpec, err := mergerSpec(catalog, lane.Merge.Module)
 	if err != nil {
-		return ResolvedArtifactLane{}, nil, moduleLookupError(pipelineID, laneID, StageMerge, lane.Merge.Module, err)
+		return ResolvedArtifactLane{}, nil, nil, moduleLookupError(pipelineID, laneID, StageMerge, lane.Merge.Module, err)
 	}
 	if missing, ok := capabilities.missing(mergeSpec.Requires); ok {
-		return ResolvedArtifactLane{}, nil, capabilityError(pipelineID, laneID, StageMerge, lane.Merge.Module, missing)
+		return ResolvedArtifactLane{}, nil, nil, capabilityError(pipelineID, laneID, StageMerge, lane.Merge.Module, missing)
 	}
 	mergeReferences, err := resolveReferenceTargetBindings(referenceResolutionTarget{
 		PipelineID:         pipelineID,
@@ -264,17 +313,17 @@ func resolveArtifactLane(
 		Options:            options,
 	})
 	if err != nil {
-		return ResolvedArtifactLane{}, nil, err
+		return ResolvedArtifactLane{}, nil, nil, err
 	}
 	lane.MergeReferences = referenceTarget(StageMerge, laneID, lane.Merge.Module, mergeReferences)
 	capabilities.add(mergeSpec.Provides...)
 
 	normalizeSpec, err := normalizerSpec(catalog, lane.Normalize.Module)
 	if err != nil {
-		return ResolvedArtifactLane{}, nil, moduleLookupError(pipelineID, laneID, StageNormalize, lane.Normalize.Module, err)
+		return ResolvedArtifactLane{}, nil, nil, moduleLookupError(pipelineID, laneID, StageNormalize, lane.Normalize.Module, err)
 	}
 	if missing, ok := capabilities.missing(normalizeSpec.Requires); ok {
-		return ResolvedArtifactLane{}, nil, capabilityError(pipelineID, laneID, StageNormalize, lane.Normalize.Module, missing)
+		return ResolvedArtifactLane{}, nil, nil, capabilityError(pipelineID, laneID, StageNormalize, lane.Normalize.Module, missing)
 	}
 	normalizeReferences, err := resolveReferenceTargetBindings(referenceResolutionTarget{
 		PipelineID:         pipelineID,
@@ -287,20 +336,108 @@ func resolveArtifactLane(
 		Options:            options,
 	})
 	if err != nil {
-		return ResolvedArtifactLane{}, nil, err
+		return ResolvedArtifactLane{}, nil, nil, err
 	}
 	lane.NormalizeReferences = referenceTarget(StageNormalize, laneID, lane.Normalize.Module, normalizeReferences)
 	capabilities.add(normalizeSpec.Provides...)
 
 	if len(lane.Validators) > 0 {
-		return ResolvedArtifactLane{}, nil, configuredValidatorsError(pipelineID, laneID)
+		return ResolvedArtifactLane{}, nil, nil, configuredValidatorsError(pipelineID, laneID)
 	}
 
-	return lane, capabilities, nil
+	extractValidatorChain, err := resolveValidatorChain(pipelineID, laneID, StageExtract, lane.Extract.Module, lane.Extract.Validators, catalog)
+	if err != nil {
+		return ResolvedArtifactLane{}, nil, nil, err
+	}
+	mergeValidatorChain, err := resolveValidatorChain(pipelineID, laneID, StageMerge, lane.Merge.Module, lane.Merge.Validators, catalog)
+	if err != nil {
+		return ResolvedArtifactLane{}, nil, nil, err
+	}
+	normalizeValidatorChain, err := resolveValidatorChain(pipelineID, laneID, StageNormalize, lane.Normalize.Module, lane.Normalize.Validators, catalog)
+	if err != nil {
+		return ResolvedArtifactLane{}, nil, nil, err
+	}
+	validatorChains := []ResolvedValidatorChain{extractValidatorChain, mergeValidatorChain, normalizeValidatorChain}
+
+	return lane, validatorChains, capabilities, nil
 }
 
 func configuredValidatorsError(pipelineID string, laneID string) error {
-	return fmt.Errorf("pipeline %q lane %q configured validators are not supported by the current raw validation runner", pipelineID, laneID)
+	return fmt.Errorf("pipeline %q lane %q validators are not supported at artifact lane level; use extract.validators, merge.validators, or normalize.validators", pipelineID, laneID)
+}
+
+func resolveValidatorChain(pipelineID string, laneID string, stage ModuleStage, module string, override ValidatorOverride, catalog ModuleCatalog) (ResolvedValidatorChain, error) {
+	chain := ResolvedValidatorChain{
+		Stage:     stage,
+		LaneID:    strings.TrimSpace(laneID),
+		ModuleKey: strings.TrimSpace(module),
+	}
+	if chain.ModuleKey == "" {
+		return ResolvedValidatorChain{}, fmt.Errorf("pipeline %q validator chain %q module key must not be empty", pipelineID, stage)
+	}
+	switch stage {
+	case StageChunk, StageExtract, StageMerge, StageNormalize:
+	default:
+		return ResolvedValidatorChain{}, fmt.Errorf("pipeline %q validator chain stage %q is not supported", pipelineID, stage)
+	}
+
+	var bindings []ModuleBinding
+	if override.Set {
+		bindings = cloneModuleBindings(override.Validators)
+	} else if catalog.ValidatorChains != nil {
+		bindings = catalog.ValidatorChains.Validators(stage, chain.ModuleKey)
+	}
+	if len(bindings) == 0 {
+		return chain, nil
+	}
+	if catalog.Validators == nil {
+		return ResolvedValidatorChain{}, fmt.Errorf("pipeline %q validator registry must not be nil for %s validator chain on module %q", pipelineID, stage, chain.ModuleKey)
+	}
+	chain.Validators = make([]ResolvedValidator, 0, len(bindings))
+	for _, validator := range bindings {
+		spec, ok := catalog.Validators.Spec(validator.Module)
+		if !ok {
+			return ResolvedValidatorChain{}, fmt.Errorf("pipeline %q %s validator chain for module %q references unknown validator %q", pipelineID, stage, chain.ModuleKey, validator.Module)
+		}
+		if strings.TrimSpace(validator.LLMProfile) != "" && spec.ExecutionClass != contracts.ExecutionClassLLMBacked {
+			return ResolvedValidatorChain{}, fmt.Errorf("pipeline %q %s validator chain for module %q assigns llm_profile to deterministic validator %q", pipelineID, stage, chain.ModuleKey, validator.Module)
+		}
+		chain.Validators = append(chain.Validators, ResolvedValidator{
+			Binding:        cloneModuleBinding(validator),
+			ExecutionClass: spec.ExecutionClass,
+		})
+	}
+	return chain, nil
+}
+
+func cloneResolvedValidatorChains(chains []ResolvedValidatorChain) []ResolvedValidatorChain {
+	if len(chains) == 0 {
+		return nil
+	}
+	out := make([]ResolvedValidatorChain, len(chains))
+	for i, chain := range chains {
+		out[i] = ResolvedValidatorChain{
+			Stage:      chain.Stage,
+			LaneID:     strings.TrimSpace(chain.LaneID),
+			ModuleKey:  strings.TrimSpace(chain.ModuleKey),
+			Validators: cloneResolvedValidators(chain.Validators),
+		}
+	}
+	return out
+}
+
+func cloneResolvedValidators(validators []ResolvedValidator) []ResolvedValidator {
+	if len(validators) == 0 {
+		return nil
+	}
+	out := make([]ResolvedValidator, len(validators))
+	for i, validator := range validators {
+		out[i] = ResolvedValidator{
+			Binding:        cloneModuleBinding(validator.Binding),
+			ExecutionClass: validator.ExecutionClass,
+		}
+	}
+	return out
 }
 
 func referenceTarget(stage ModuleStage, laneID string, module string, bindings []ReferenceBinding) ResolvedReferenceTarget {
@@ -625,6 +762,7 @@ func resolveBinding(binding ModuleBinding, defaultModule string) ModuleBinding {
 		Retries:    binding.Retries,
 		Options:    cloneOptions(binding.Options),
 		References: normalizeReferenceMap(binding.References),
+		Validators: cloneValidatorOverride(binding.Validators),
 	}
 }
 

internal/framework/pipeline/profile_test.go

@@ -110,6 +110,215 @@ func TestResolvePipelineAppliesDefaults(t *testing.T) {
 	}
 }
 
+func TestResolvePipelineRecordsValidatorChains(t *testing.T) {
+	catalog := newProfileCatalog(t)
+	if err := catalog.ValidatorChains.Register(ValidatorChainMapping{
+		Stage:      StageExtract,
+		Module:     "event-extractor",
+		Validators: []ModuleBinding{Binding("grounded")},
+	}); err != nil {
+		t.Fatalf("register validator chain: %v", err)
+	}
+
+	resolved, err := ResolvePipeline(PipelineProfile{
+		ID:    "validated",
+		Input: Binding("text"),
+		Artifacts: map[string]ArtifactLaneProfile{
+			"events": {Extract: Binding("event-extractor")},
+		},
+	}, ResolveOptions{}, catalog)
+	if err != nil {
+		t.Fatalf("ResolvePipeline() error = %v, want nil", err)
+	}
+
+	if len(resolved.ValidatorChains) != 4 {
+		t.Fatalf("len(ValidatorChains) = %d, want chunk plus lane extract/merge/normalize", len(resolved.ValidatorChains))
+	}
+	extractChain := findResolvedValidatorChain(resolved.ValidatorChains, StageExtract, "events", "event-extractor")
+	if extractChain == nil {
+		t.Fatal("extract validator chain not found")
+	}
+	if len(extractChain.Validators) != 1 {
+		t.Fatalf("extract validators = %#v, want one validator", extractChain.Validators)
+	}
+	if extractChain.Validators[0].Binding.Module != "grounded" {
+		t.Fatalf("extract validator key = %q, want grounded", extractChain.Validators[0].Binding.Module)
+	}
+	if extractChain.Validators[0].ExecutionClass != contracts.ExecutionClassDeterministic {
+		t.Fatalf("extract validator execution class = %q, want deterministic", extractChain.Validators[0].ExecutionClass)
+	}
+
+	chunkChain := findResolvedValidatorChain(resolved.ValidatorChains, StageChunk, "", DefaultChunkModule)
+	if chunkChain == nil {
+		t.Fatal("chunk validator chain not found")
+	}
+	if len(chunkChain.Validators) != 0 {
+		t.Fatalf("chunk validators = %#v, want explicit empty chain", chunkChain.Validators)
+	}
+}
+
+func TestResolvePipelineRejectsUnknownDefaultValidator(t *testing.T) {
+	catalog := newProfileCatalog(t)
+	if err := catalog.ValidatorChains.Register(ValidatorChainMapping{
+		Stage:      StageNormalize,
+		Module:     DefaultNormalizeModule,
+		Validators: []ModuleBinding{Binding("missing-validator")},
+	}); err != nil {
+		t.Fatalf("register validator chain: %v", err)
+	}
+
+	_, err := ResolvePipeline(PipelineProfile{
+		ID:    "invalid-chain",
+		Input: Binding("text"),
+		Artifacts: map[string]ArtifactLaneProfile{
+			"events": {Extract: Binding("event-extractor")},
+		},
+	}, ResolveOptions{}, catalog)
+	if err == nil {
+		t.Fatal("ResolvePipeline() error = nil, want unknown validator error")
+	}
+	if !strings.Contains(err.Error(), "missing-validator") {
+		t.Fatalf("ResolvePipeline() error = %q, want missing validator context", err.Error())
+	}
+}
+
+func TestResolvePipelineValidatorOverrideReplacesDefaultChain(t *testing.T) {
+	catalog := newProfileCatalog(t)
+	registerProfileValidatorSpec(t, catalog, ValidatorSpec{Key: "second-validator", ExecutionClass: contracts.ExecutionClassLLMBacked})
+	if err := catalog.ValidatorChains.Register(ValidatorChainMapping{
+		Stage:      StageExtract,
+		Module:     "event-extractor",
+		Validators: []ModuleBinding{Binding("grounded")},
+	}); err != nil {
+		t.Fatalf("register validator chain: %v", err)
+	}
+
+	profile := PipelineProfile{
+		ID:    "validated",
+		Input: Binding("text"),
+		Artifacts: map[string]ArtifactLaneProfile{
+			"events": {
+				Extract: ModuleBinding{
+					Module: "event-extractor",
+					Validators: ValidatorOverride{
+						Set: true,
+						Validators: []ModuleBinding{
+							{Module: "second-validator", LLMProfile: "careful"},
+							Binding("grounded"),
+						},
+					},
+				},
+			},
+		},
+	}
+
+	resolved, err := ResolvePipeline(profile, ResolveOptions{}, catalog)
+	if err != nil {
+		t.Fatalf("ResolvePipeline() error = %v, want nil", err)
+	}
+
+	extractChain := findResolvedValidatorChain(resolved.ValidatorChains, StageExtract, "events", "event-extractor")
+	if extractChain == nil {
+		t.Fatal("extract validator chain not found")
+	}
+	if len(extractChain.Validators) != 2 {
+		t.Fatalf("extract validators = %#v, want explicit two-validator override", extractChain.Validators)
+	}
+	if extractChain.Validators[0].Binding.Module != "second-validator" || extractChain.Validators[0].Binding.LLMProfile != "careful" {
+		t.Fatalf("first validator = %#v, want explicit LLM-backed validator first", extractChain.Validators[0])
+	}
+	if extractChain.Validators[1].Binding.Module != "grounded" {
+		t.Fatalf("second validator = %#v, want grounded second", extractChain.Validators[1])
+	}
+}
+
+func TestResolvePipelineExplicitEmptyValidatorOverrideSuppressesDefaultChain(t *testing.T) {
+	catalog := newProfileCatalog(t)
+	if err := catalog.ValidatorChains.Register(ValidatorChainMapping{
+		Stage:      StageExtract,
+		Module:     "event-extractor",
+		Validators: []ModuleBinding{Binding("grounded")},
+	}); err != nil {
+		t.Fatalf("register validator chain: %v", err)
+	}
+
+	profile := PipelineProfile{
+		ID:    "validated",
+		Input: Binding("text"),
+		Artifacts: map[string]ArtifactLaneProfile{
+			"events": {
+				Extract: ModuleBinding{
+					Module:     "event-extractor",
+					Validators: ValidatorOverride{Set: true},
+				},
+			},
+		},
+	}
+
+	resolved, err := ResolvePipeline(profile, ResolveOptions{}, catalog)
+	if err != nil {
+		t.Fatalf("ResolvePipeline() error = %v, want nil", err)
+	}
+
+	extractChain := findResolvedValidatorChain(resolved.ValidatorChains, StageExtract, "events", "event-extractor")
+	if extractChain == nil {
+		t.Fatal("extract validator chain not found")
+	}
+	if len(extractChain.Validators) != 0 {
+		t.Fatalf("extract validators = %#v, want explicit empty override", extractChain.Validators)
+	}
+}
+
+func TestResolvePipelineRejectsUnknownOverrideValidator(t *testing.T) {
+	_, err := ResolvePipeline(PipelineProfile{
+		ID:    "validated",
+		Input: Binding("text"),
+		Artifacts: map[string]ArtifactLaneProfile{
+			"events": {
+				Extract: ModuleBinding{
+					Module: "event-extractor",
+					Validators: ValidatorOverride{
+						Set:        true,
+						Validators: []ModuleBinding{Binding("missing-validator")},
+					},
+				},
+			},
+		},
+	}, ResolveOptions{}, newProfileCatalog(t))
+	if err == nil {
+		t.Fatal("ResolvePipeline() error = nil, want unknown validator error")
+	}
+	if !strings.Contains(err.Error(), "missing-validator") {
+		t.Fatalf("ResolvePipeline() error = %q, want missing validator context", err.Error())
+	}
+}
+
+func TestResolvePipelineRejectsLLMProfileForDeterministicValidator(t *testing.T) {
+	_, err := ResolvePipeline(PipelineProfile{
+		ID:    "validated",
+		Input: Binding("text"),
+		Artifacts: map[string]ArtifactLaneProfile{
+			"events": {
+				Extract: ModuleBinding{
+					Module: "event-extractor",
+					Validators: ValidatorOverride{
+						Set: true,
+						Validators: []ModuleBinding{
+							{Module: "grounded", LLMProfile: "careful"},
+						},
+					},
+				},
+			},
+		},
+	}, ResolveOptions{}, newProfileCatalog(t))
+	if err == nil {
+		t.Fatal("ResolvePipeline() error = nil, want deterministic validator profile error")
+	}
+	if !strings.Contains(err.Error(), "grounded") || !strings.Contains(err.Error(), "llm_profile") {
+		t.Fatalf("ResolvePipeline() error = %q, want validator profile context", err.Error())
+	}
+}
+
 func TestResolvePipelineSelectsOnlyRequestedLanes(t *testing.T) {
 	profile := multiLaneProfile()
 	resolved, err := ResolvePipeline(profile, ResolveOptions{Only: []string{" summaries ", "events", "summaries"}}, newProfileCatalog(t))
@@ -778,7 +987,7 @@ func TestResolvePipelineRejectsConfiguredValidators(t *testing.T) {
 	if err == nil {
 		t.Fatal("ResolvePipeline() error = nil, want error")
 	}
-	assertErrorContains(t, err, "baseline", "events", "configured validators", "not supported")
+	assertErrorContains(t, err, "baseline", "events", "validators", "extract.validators")
 }
 
 func TestResolvePipelineOrdersLanesDeterministically(t *testing.T) {
@@ -955,6 +1164,15 @@ func assertBindingSource(t *testing.T, bindings []ReferenceBinding, slotName str
 	t.Fatalf("binding %q not found in %#v", slotName, bindings)
 }
 
+func findResolvedValidatorChain(chains []ResolvedValidatorChain, stage ModuleStage, laneID string, module string) *ResolvedValidatorChain {
+	for i := range chains {
+		if chains[i].Stage == stage && chains[i].LaneID == laneID && chains[i].ModuleKey == module {
+			return &chains[i]
+		}
+	}
+	return nil
+}
+
 func newProfileCatalog(t *testing.T) ModuleCatalog {
 	t.Helper()
 
@@ -1000,6 +1218,7 @@ func emptyProfileCatalog() ModuleCatalog {
 		Mergers:         NewMergerRegistry(),
 		Normalizers:     NewNormalizerRegistry(),
 		Validators:      NewValidatorRegistry(),
+		ValidatorChains: NewValidatorChainRegistry(),
 		Outputs:         NewOutputEncoderRegistry(),
 	}
 }
@@ -1043,7 +1262,8 @@ func registerProfileSpecs(t *testing.T, catalog ModuleCatalog, specs ...ModuleSp
 				t.Fatalf("register normalizer spec %#v: %v", spec, err)
 			}
 		case StageValidate:
-			if err := catalog.Validators.RegisterWithSpec(spec, profileValidatorConstructor(spec.Key)); err != nil {
+			validatorSpec := ValidatorSpec{Key: spec.Key, ExecutionClass: contracts.ExecutionClassDeterministic}
+			if err := catalog.Validators.RegisterWithSpec(validatorSpec, profileValidatorConstructor(spec.Key)); err != nil {
 				t.Fatalf("register validator spec %#v: %v", spec, err)
 			}
 		case StageOutput:
@@ -1056,6 +1276,13 @@ func registerProfileSpecs(t *testing.T, catalog ModuleCatalog, specs ...ModuleSp
 	}
 }
 
+func registerProfileValidatorSpec(t *testing.T, catalog ModuleCatalog, spec ValidatorSpec) {
+	t.Helper()
+	if err := catalog.Validators.RegisterWithSpec(spec, profileValidatorConstructor(spec.Key)); err != nil {
+		t.Fatalf("register validator spec %#v: %v", spec, err)
+	}
+}
+
 func profileInputConstructor(key string) InputAdapterConstructor {
 	return func() (contracts.InputAdapter, error) {
 		return profileInputAdapter{key: key}, nil

internal/framework/pipeline/raw_validation_registry.go

@@ -1,73 +0,0 @@
-package pipeline
-
-import (
-	"fmt"
-	"strings"
-
-	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
-)
-
-type rawValidationKey struct {
-	stage  ModuleStage
-	module string
-}
-
-type RawValidationRegistry struct {
-	chains map[rawValidationKey][]contracts.RawValidator
-}
-
-func NewRawValidationRegistry() *RawValidationRegistry {
-	return &RawValidationRegistry{
-		chains: make(map[rawValidationKey][]contracts.RawValidator),
-	}
-}
-
-func (r *RawValidationRegistry) Register(stage ModuleStage, module string, validators ...contracts.RawValidator) error {
-	if r == nil {
-		return fmt.Errorf("raw validation registry must not be nil")
-	}
-	normalizedModule := strings.TrimSpace(module)
-	if normalizedModule == "" {
-		return fmt.Errorf("raw validation module key must not be empty")
-	}
-	switch stage {
-	case StageChunk, StageExtract, StageMerge, StageNormalize:
-	default:
-		return fmt.Errorf("raw validation stage %q is not supported", stage)
-	}
-	if len(validators) == 0 {
-		return fmt.Errorf("raw validation chain for %q %q must not be empty", stage, normalizedModule)
-	}
-
-	chain := make([]contracts.RawValidator, 0, len(validators))
-	for i, validator := range validators {
-		if validator == nil {
-			return fmt.Errorf("raw validator %d for %q %q must not be nil", i, stage, normalizedModule)
-		}
-		if strings.TrimSpace(validator.Name()) == "" {
-			return fmt.Errorf("raw validator %d for %q %q must not have an empty name", i, stage, normalizedModule)
-		}
-		chain = append(chain, validator)
-	}
-
-	if r.chains == nil {
-		r.chains = make(map[rawValidationKey][]contracts.RawValidator)
-	}
-	key := rawValidationKey{stage: stage, module: normalizedModule}
-	if _, exists := r.chains[key]; exists {
-		return fmt.Errorf("raw validation chain for %q %q is already registered", stage, normalizedModule)
-	}
-	r.chains[key] = append([]contracts.RawValidator(nil), chain...)
-	return nil
-}
-
-func (r *RawValidationRegistry) Validators(stage ModuleStage, module string) []contracts.RawValidator {
-	if r == nil {
-		return nil
-	}
-	chain := r.chains[rawValidationKey{stage: stage, module: strings.TrimSpace(module)}]
-	if len(chain) == 0 {
-		return nil
-	}
-	return append([]contracts.RawValidator(nil), chain...)
-}

internal/framework/pipeline/runner.go

@@ -24,7 +24,7 @@ type Registries struct {
 	Mergers         *MergerRegistry
 	Normalizers     *NormalizerRegistry
 	Validators      *ValidatorRegistry
-	RawValidators *RawValidationRegistry
+	ValidatorChains *ValidatorChainRegistry
 	Outputs         *OutputEncoderRegistry
 }
 
@@ -127,7 +127,7 @@ func (r *Runner) Run(ctx context.Context, input RunInput) (output RunOutput, err
 		if err != nil {
 			return false, nil, fmt.Errorf("validate chunks from chunker %q: %w", chunker.Key(), err)
 		}
-		rejection, err := r.validateChunksRaw(ctx, doc, chunker.Key(), chunks, input.Metadata, attempt)
+		rejection, err := r.validateChunksRaw(ctx, doc, chunker.Key(), chunks, sourceInput, sessionID, input.Pipeline.ChunkReferences.ReferenceSet, input.LLMClient, input.Metadata, input.Pipeline.ValidatorChains, attempt)
 		if err != nil || rejection != nil {
 			return false, rejection, err
 		}
@@ -237,9 +237,15 @@ func (r *Runner) runLane(ctx context.Context, input RunInput, doc *source.Source
 				sourceID:    doc.ID,
 				chunkID:     chunk.ID,
 				chunkIndex:  chunk.Index,
+				chunk:       &chunk,
+				sourceInput: chunkInputMaterial(sourceInput, chunk),
+				sessionID:   sessionID,
+				references:  lane.ExtractReferences.ReferenceSet,
+				llmClient:   input.LLMClient,
 				schema:      extractOutput.Schema,
 				payload:     extractOutput.Payload,
 				metadata:    input.Metadata,
+				chains:      input.Pipeline.ValidatorChains,
 				attempt:     attempt,
 			})
 			if err != nil || rejection != nil {
@@ -293,9 +299,15 @@ func (r *Runner) runLane(ctx context.Context, input RunInput, doc *source.Source
 			moduleKey:      merger.Key(),
 			source:         doc,
 			sourceID:       doc.ID,
+			sourceInput:    sourceInput.Clone(),
+			sessionID:      sessionID,
+			references:     lane.MergeReferences.ReferenceSet,
+			llmClient:      input.LLMClient,
 			schema:         mergeOutput.Schema,
 			payload:        mergeOutput.Payload,
+			extractOutputs: extractOutputs,
 			metadata:       input.Metadata,
+			chains:         input.Pipeline.ValidatorChains,
 			attempt:        attempt,
 		})
 		if err != nil || rejection != nil {
@@ -343,9 +355,15 @@ func (r *Runner) runLane(ctx context.Context, input RunInput, doc *source.Source
 			moduleKey:   normalizer.Key(),
 			source:      doc,
 			sourceID:    doc.ID,
+			sourceInput: sourceInput.Clone(),
+			sessionID:   sessionID,
+			references:  lane.NormalizeReferences.ReferenceSet,
+			llmClient:   input.LLMClient,
 			schema:      normalizeOutput.Schema,
 			payload:     normalizeOutput.Payload,
+			mergeOutput: acceptedMerge,
 			metadata:    input.Metadata,
+			chains:      input.Pipeline.ValidatorChains,
 			attempt:     attempt,
 		})
 		if err != nil || rejection != nil {
@@ -373,11 +391,20 @@ type rawValidationTarget struct {
 	moduleKey      string
 	source         *source.SourceDocument
 	sourceID       string
+	sourceInput    contracts.LLMInputMaterial
+	sessionID      string
+	references     contracts.ReferenceSet
+	llmClient      contracts.StructuredLLMClient
 	chunkID        string
 	chunkIndex     int
+	chunk          *contracts.SourceChunk
+	chunks         []contracts.SourceChunk
 	schema         contracts.ResponseSchema
 	payload        contracts.RawPayload
+	extractOutputs []contracts.ExtractOutput
+	mergeOutput    contracts.MergeOutput
 	metadata       map[string]any
+	chains         []ResolvedValidatorChain
 	attempt        int
 }
 
@@ -428,21 +455,29 @@ func runWithRetry(ctx context.Context, retries int, run func(attempt int) (bool,
 	return false, lastRejection, nil
 }
 
-func (r *Runner) validateChunksRaw(ctx context.Context, doc *source.SourceDocument, moduleKey string, chunks []contracts.SourceChunk, metadata map[string]any, attempt int) (*contracts.RejectedOutput, error) {
+func (r *Runner) validateChunksRaw(ctx context.Context, doc *source.SourceDocument, moduleKey string, chunks []contracts.SourceChunk, sourceInput contracts.LLMInputMaterial, sessionID string, references contracts.ReferenceSet, llmClient contracts.StructuredLLMClient, metadata map[string]any, chains []ResolvedValidatorChain, attempt int) (*contracts.RejectedOutput, error) {
-	for _, chunk := range chunks {
+	for index := range chunks {
+		chunk := chunks[index]
 		_, rejection, err := r.validateRaw(ctx, rawValidationTarget{
 			stage:       StageChunk,
 			moduleKey:   moduleKey,
 			source:      doc,
 			sourceID:    doc.ID,
+			sourceInput: sourceInput.Clone(),
+			sessionID:   sessionID,
+			references:  references,
+			llmClient:   llmClient,
 			chunkID:     chunk.ID,
 			chunkIndex:  chunk.Index,
+			chunk:       &chunk,
+			chunks:      chunks,
 			payload: contracts.RawPayload{
 				Content:   append([]byte(nil), chunk.Content...),
 				MediaType: chunk.MediaType,
 				Metadata:  cloneMetadata(chunk.Metadata),
 			},
 			metadata: metadata,
+			chains:   chains,
 			attempt:  attempt,
 		})
 		if err != nil {
@@ -456,27 +491,22 @@ func (r *Runner) validateChunksRaw(ctx context.Context, doc *source.SourceDocume
 }
 
 func (r *Runner) validateRaw(ctx context.Context, target rawValidationTarget) ([]contracts.Warning, *contracts.RejectedOutput, error) {
-	validators := r.registries.RawValidators.Validators(target.stage, target.moduleKey)
+	chain := resolvedValidatorChain(target.stage, target.laneID, target.moduleKey, target.chains)
-	if len(validators) == 0 {
+	if len(chain.Validators) == 0 {
 		return nil, nil, nil
 	}
-
+	if r.registries.Validators == nil {
-	request := contracts.RawValidationRequest{
+		return nil, nil, fmt.Errorf("validator registry must not be nil")
-		Stage:      string(target.stage),
-		LaneID:     target.laneID,
-		ModuleKey:  target.moduleKey,
-		Source:     target.source,
-		SourceID:   target.sourceID,
-		ChunkID:    target.chunkID,
-		ChunkIndex: target.chunkIndex,
-		Schema:     target.schema,
-		Payload:    cloneRawPayload(target.payload),
-		Metadata:   cloneMetadata(target.metadata),
 	}
 
 	var warnings []contracts.Warning
-	for _, validator := range validators {
+	for _, validatorBinding := range chain.Validators {
-		result, err := validator.ValidateRaw(ctx, request)
+		validator, err := r.registries.Validators.Build(validatorBinding.Binding.Module)
+		if err != nil {
+			return nil, nil, fmt.Errorf("build validator %q: %w", validatorBinding.Binding.Module, err)
+		}
+		request := target.validationRequest(validatorBinding.Binding)
+		result, err := validator.Validate(ctx, request)
 		if err != nil {
 			return nil, nil, fmt.Errorf("validate raw %s output with validator %q: %w", target.stage, validator.Name(), err)
 		}
@@ -507,6 +537,56 @@ func (r *Runner) validateRaw(ctx context.Context, target rawValidationTarget) ([
 	return warnings, nil, nil
 }
 
+func (target rawValidationTarget) validationRequest(binding ModuleBinding) contracts.ValidationRequest {
+	return contracts.ValidationRequest{
+		Stage:          string(target.stage),
+		LaneID:         target.laneID,
+		ModuleKey:      target.moduleKey,
+		Source:         target.source,
+		SourceID:       target.sourceID,
+		SourceInput:    target.sourceInput.Clone(),
+		SessionID:      target.sessionID,
+		References:     CloneReferenceSet(target.references),
+		LLMClient:      target.llmClient,
+		LLMProfile:     binding.LLMProfile,
+		Options:        cloneOptions(binding.Options),
+		Metadata:       cloneMetadata(target.metadata),
+		Schema:         cloneResponseSchema(target.schema),
+		Payload:        cloneRawPayload(target.payload),
+		ChunkID:        target.chunkID,
+		ChunkIndex:     target.chunkIndex,
+		Chunk:          cloneSourceChunkPtr(target.chunk),
+		Chunks:         cloneSourceChunks(target.chunks),
+		ExtractOutputs: cloneExtractOutputs(target.extractOutputs),
+		MergeOutput:    cloneMergeOutput(target.mergeOutput),
+	}
+}
+
+func resolvedValidatorChain(stage ModuleStage, laneID string, moduleKey string, chains []ResolvedValidatorChain) ResolvedValidatorChain {
+	for _, chain := range chains {
+		if chain.Stage != stage {
+			continue
+		}
+		if chain.ModuleKey != moduleKey {
+			continue
+		}
+		if strings.TrimSpace(chain.LaneID) != strings.TrimSpace(laneID) {
+			continue
+		}
+		return ResolvedValidatorChain{
+			Stage:      chain.Stage,
+			LaneID:     chain.LaneID,
+			ModuleKey:  chain.ModuleKey,
+			Validators: cloneResolvedValidators(chain.Validators),
+		}
+	}
+	return ResolvedValidatorChain{
+		Stage:     stage,
+		LaneID:    strings.TrimSpace(laneID),
+		ModuleKey: strings.TrimSpace(moduleKey),
+	}
+}
+
 func (r *Runner) validateRegistries(pipeline ResolvedPipeline) error {
 	if r.registries.Inputs == nil {
 		return fmt.Errorf("input registry must not be nil")
@@ -562,7 +642,7 @@ func validateRunInput(input RunInput) error {
 			return fmt.Errorf("resolved pipeline lane %q normalize module must not be empty", lane.ID)
 		}
 		if len(lane.Validators) > 0 {
-			return fmt.Errorf("resolved pipeline lane %q configured validators are not supported by the current raw validation runner", lane.ID)
+			return fmt.Errorf("resolved pipeline lane %q validators are not supported at artifact lane level; use extract.validators, merge.validators, or normalize.validators", lane.ID)
 		}
 	}
 	return nil
@@ -586,6 +666,7 @@ func manifestFromPipeline(input RunInput) artifacts.RunManifest {
 		Chunker:         pipeline.Chunk.Module,
 		OutputEncoder:   pipeline.Output.Module,
 		ArtifactLanes:   make([]artifacts.ArtifactLaneManifest, 0, len(pipeline.ArtifactLanes)),
+		ValidatorChains: validatorChainManifests(pipeline.ValidatorChains),
 		RunID:           runID,
 		StartedAt:       timePtr(startedAt),
 		References:      ReferenceProvenance(pipeline),
@@ -607,6 +688,29 @@ func manifestFromPipeline(input RunInput) artifacts.RunManifest {
 	return manifest
 }
 
+func validatorChainManifests(chains []ResolvedValidatorChain) []artifacts.ValidatorChainManifest {
+	if len(chains) == 0 {
+		return nil
+	}
+	manifests := make([]artifacts.ValidatorChainManifest, 0, len(chains))
+	for _, chain := range chains {
+		manifest := artifacts.ValidatorChainManifest{
+			Stage:      string(chain.Stage),
+			LaneID:     chain.LaneID,
+			ModuleKey:  chain.ModuleKey,
+			Validators: make([]artifacts.ValidatorManifest, 0, len(chain.Validators)),
+		}
+		for _, validator := range chain.Validators {
+			manifest.Validators = append(manifest.Validators, artifacts.ValidatorManifest{
+				Key:            validator.Binding.Module,
+				ExecutionClass: string(validator.ExecutionClass),
+			})
+		}
+		manifests = append(manifests, manifest)
+	}
+	return manifests
+}
+
 func failOutput(output RunOutput) RunOutput {
 	if output.Manifest.PipelineID != "" {
 		populateRawOutputManifest(&output)
@@ -908,7 +1012,50 @@ func cloneRawPayload(payload contracts.RawPayload) contracts.RawPayload {
 	}
 }
 
+func cloneResponseSchema(schema contracts.ResponseSchema) contracts.ResponseSchema {
+	schema.JSONSchema = append([]byte(nil), schema.JSONSchema...)
+	return schema
+}
+
+func cloneSourceChunkPtr(chunk *contracts.SourceChunk) *contracts.SourceChunk {
+	if chunk == nil {
+		return nil
+	}
+	cloned := cloneSourceChunk(*chunk)
+	return &cloned
+}
+
+func cloneSourceChunk(chunk contracts.SourceChunk) contracts.SourceChunk {
+	chunk.Content = append([]byte(nil), chunk.Content...)
+	chunk.Units = cloneSourceUnits(chunk.Units)
+	chunk.Metadata = cloneMetadata(chunk.Metadata)
+	return chunk
+}
+
+func cloneSourceChunks(chunks []contracts.SourceChunk) []contracts.SourceChunk {
+	if len(chunks) == 0 {
+		return nil
+	}
+	out := make([]contracts.SourceChunk, 0, len(chunks))
+	for _, chunk := range chunks {
+		out = append(out, cloneSourceChunk(chunk))
+	}
+	return out
+}
+
+func cloneSourceUnits(units []source.SourceUnit) []source.SourceUnit {
+	if len(units) == 0 {
+		return nil
+	}
+	out := make([]source.SourceUnit, 0, len(units))
+	for _, unit := range units {
+		out = append(out, cloneSourceUnit(unit))
+	}
+	return out
+}
+
 func cloneExtractOutput(output contracts.ExtractOutput) contracts.ExtractOutput {
+	output.Schema = cloneResponseSchema(output.Schema)
 	output.Payload = cloneRawPayload(output.Payload)
 	return output
 }
@@ -925,11 +1072,13 @@ func cloneExtractOutputs(outputs []contracts.ExtractOutput) []contracts.ExtractO
 }
 
 func cloneMergeOutput(output contracts.MergeOutput) contracts.MergeOutput {
+	output.Schema = cloneResponseSchema(output.Schema)
 	output.Payload = cloneRawPayload(output.Payload)
 	return output
 }
 
 func cloneNormalizeOutput(output contracts.NormalizeOutput) contracts.NormalizeOutput {
+	output.Schema = cloneResponseSchema(output.Schema)
 	output.Payload = cloneRawPayload(output.Payload)
 	return output
 }

internal/framework/pipeline/runner_test.go

@@ -763,6 +763,110 @@ func TestRunPassesNormalizeReferencesToNormalizerRequest(t *testing.T) {
 	}
 }
 
+func TestRunPassesValidationRequestContextToValidators(t *testing.T) {
+	modules := defaultRunnerModules()
+	chunkValidator := &runnerChainValidator{name: "chain-chunk"}
+	extractValidator := &runnerChainValidator{name: "chain-extract", executionClass: contracts.ExecutionClassLLMBacked}
+	mergeValidator := &runnerChainValidator{name: "chain-merge"}
+	normalizeValidator := &runnerChainValidator{name: "chain-normalize"}
+	modules.validators[chunkValidator.name] = chunkValidator
+	modules.validators[extractValidator.name] = extractValidator
+	modules.validators[mergeValidator.name] = mergeValidator
+	modules.validators[normalizeValidator.name] = normalizeValidator
+
+	pipeline := resolvedPipeline()
+	pipeline.ChunkReferences.ReferenceSet = testReferenceSet("scene_guide", "chunk reference text")
+	pipeline.ArtifactLanes[0].ExtractReferences.ReferenceSet = testReferenceSet("roster", "extract reference text")
+	pipeline.ArtifactLanes[0].MergeReferences.ReferenceSet = testReferenceSet("merge_notes", "merge reference text")
+	pipeline.ArtifactLanes[0].NormalizeReferences.ReferenceSet = testReferenceSet("normalization_notes", "normalize reference text")
+	setResolvedValidatorChain(t, &pipeline, StageChunk, "", "chunk", resolvedValidatorForTest(chunkValidator))
+	setResolvedValidatorChain(t, &pipeline, StageExtract, "alpha", "extract-alpha", ResolvedValidator{
+		Binding:        ModuleBinding{Module: extractValidator.name, LLMProfile: "validator-profile", Options: map[string]any{"strict": true}},
+		ExecutionClass: extractValidator.ExecutionClass(),
+	})
+	setResolvedValidatorChain(t, &pipeline, StageMerge, "alpha", "merge", resolvedValidatorForTest(mergeValidator))
+	setResolvedValidatorChain(t, &pipeline, StageNormalize, "alpha", "normalize", resolvedValidatorForTest(normalizeValidator))
+
+	rawInput := []byte("{\"source\":\"exact bytes\"}")
+	llmClient := fakeLLMClient{}
+	_, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{
+		Pipeline:  pipeline,
+		Path:      "session.json",
+		RawInput:  rawInput,
+		LLMClient: llmClient,
+		SessionID: "session-123",
+		Metadata:  map[string]any{"request": "test"},
+	})
+	if err != nil {
+		t.Fatalf("Run() error = %v, want nil", err)
+	}
+
+	if len(chunkValidator.requests) != 2 {
+		t.Fatalf("chunk validator requests = %d, want one per chunk", len(chunkValidator.requests))
+	}
+	chunkReq := chunkValidator.requests[0]
+	if chunkReq.Stage != string(StageChunk) || chunkReq.ModuleKey != "chunk" || chunkReq.SourceID != "source-1" || chunkReq.SessionID != "session-123" {
+		t.Fatalf("chunk validation request = %#v, want stage/module/source/session provenance", chunkReq)
+	}
+	if chunkReq.LLMClient == nil || string(chunkReq.SourceInput.Content) != string(rawInput) {
+		t.Fatalf("chunk validation source/client = %#v, want full source input and LLM client", chunkReq.SourceInput)
+	}
+	if chunkReq.Chunk == nil || chunkReq.Chunk.ID != "chunk-0" || len(chunkReq.Chunks) != 2 || string(chunkReq.Payload.Content) != string(chunkReq.Chunk.Content) {
+		t.Fatalf("chunk validation chunk fields = %#v chunks=%#v payload=%s, want chunk payload and all chunks", chunkReq.Chunk, chunkReq.Chunks, chunkReq.Payload.Content)
+	}
+	if item := chunkReq.References.Slots["scene_guide"].Items[0]; string(item.Content) != "chunk reference text" {
+		t.Fatalf("chunk validation references = %#v, want chunk references", chunkReq.References)
+	}
+
+	if len(extractValidator.requests) != 2 {
+		t.Fatalf("extract validator requests = %d, want one per chunk", len(extractValidator.requests))
+	}
+	extractReq := extractValidator.requests[0]
+	if extractReq.Stage != string(StageExtract) || extractReq.LaneID != "alpha" || extractReq.ModuleKey != "extract-alpha" || extractReq.ChunkID != "chunk-0" || extractReq.ChunkIndex != 0 {
+		t.Fatalf("extract validation request = %#v, want extract provenance", extractReq)
+	}
+	if extractReq.LLMProfile != "validator-profile" || extractReq.Options["strict"] != true || extractReq.Metadata["request"] != "test" {
+		t.Fatalf("extract validator binding fields = profile %q options %#v metadata %#v", extractReq.LLMProfile, extractReq.Options, extractReq.Metadata)
+	}
+	if extractReq.Chunk == nil || string(extractReq.SourceInput.Content) != string(extractReq.Chunk.Content) {
+		t.Fatalf("extract source input = %#v chunk=%#v, want chunk material", extractReq.SourceInput, extractReq.Chunk)
+	}
+	if item := extractReq.References.Slots["roster"].Items[0]; string(item.Content) != "extract reference text" {
+		t.Fatalf("extract validation references = %#v, want extract references", extractReq.References)
+	}
+
+	if len(mergeValidator.requests) != 1 {
+		t.Fatalf("merge validator requests = %d, want one", len(mergeValidator.requests))
+	}
+	mergeReq := mergeValidator.requests[0]
+	if mergeReq.Stage != string(StageMerge) || mergeReq.LaneID != "alpha" || len(mergeReq.ExtractOutputs) != 2 {
+		t.Fatalf("merge validation request = %#v, want lane and extract outputs", mergeReq)
+	}
+	if mergeReq.ExtractOutputs[0].ChunkID != "chunk-0" || string(mergeReq.SourceInput.Content) != string(rawInput) {
+		t.Fatalf("merge validation upstream/source = %#v source=%#v, want ordered extracts and source input", mergeReq.ExtractOutputs, mergeReq.SourceInput)
+	}
+	if item := mergeReq.References.Slots["merge_notes"].Items[0]; string(item.Content) != "merge reference text" {
+		t.Fatalf("merge validation references = %#v, want merge references", mergeReq.References)
+	}
+
+	if len(normalizeValidator.requests) != 1 {
+		t.Fatalf("normalize validator requests = %d, want one", len(normalizeValidator.requests))
+	}
+	normalizeReq := normalizeValidator.requests[0]
+	if normalizeReq.Stage != string(StageNormalize) || normalizeReq.LaneID != "alpha" || string(normalizeReq.MergeOutput.Payload.Content) != `{"merged":true}` {
+		t.Fatalf("normalize validation request = %#v, want merge output context", normalizeReq)
+	}
+	if item := normalizeReq.References.Slots["normalization_notes"].Items[0]; string(item.Content) != "normalize reference text" {
+		t.Fatalf("normalize validation references = %#v, want normalize references", normalizeReq.References)
+	}
+
+	chunkReq.Payload.Content[0] = 'X'
+	chunkReq.Chunks[0].Content[0] = 'Y'
+	if got := string(modules.chunker.chunks[0].Content); got != `{"units":[{"id":1,"kind":"unit","text":"Source unit."}]}` {
+		t.Fatalf("validator request mutated original chunk content: %q", got)
+	}
+}
+
 func TestRunAllowsNilLLMClientWhenModulesDoNotUseIt(t *testing.T) {
 	_, err := New(newRunnerRegistries(t, defaultRunnerModules())).Run(context.Background(), RunInput{Pipeline: resolvedPipeline()})
 	if err != nil {
@@ -887,10 +991,12 @@ func TestRunPassesChunkContentAndMediaTypeToExtractors(t *testing.T) {
 
 func TestRunOmitsRejectedExtractOutputsFromMerge(t *testing.T) {
 	modules := defaultRunnerModules()
-	validator := &runnerRawValidator{name: "raw-extract", approved: []bool{false, true}, reason: "bad_extract", message: "extract rejected"}
+	validator := &runnerChainValidator{name: "chain-extract", approved: []bool{false, true}, reason: "bad_extract", message: "extract rejected"}
-	modules.rawValidators = rawValidationRegistry(t, StageExtract, "extract-alpha", validator)
+	modules.validators[validator.name] = validator
+	pipeline := resolvedPipeline()
+	setResolvedValidatorChain(t, &pipeline, StageExtract, "alpha", "extract-alpha", resolvedValidatorForTest(validator))
 
-	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: resolvedPipeline()})
+	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: pipeline})
 	if err != nil {
 		t.Fatalf("Run() error = %v, want nil", err)
 	}
@@ -909,10 +1015,12 @@ func TestRunOmitsRejectedExtractOutputsFromMerge(t *testing.T) {
 
 func TestRunOmitsLaneWithNoAcceptedExtractOutputs(t *testing.T) {
 	modules := defaultRunnerModules()
-	validator := &runnerRawValidator{name: "raw-extract", approved: []bool{false}, reason: "bad_extract", message: "extract rejected"}
+	validator := &runnerChainValidator{name: "chain-extract", approved: []bool{false}, reason: "bad_extract", message: "extract rejected"}
-	modules.rawValidators = rawValidationRegistry(t, StageExtract, "extract-alpha", validator)
+	modules.validators[validator.name] = validator
+	pipeline := resolvedPipeline()
+	setResolvedValidatorChain(t, &pipeline, StageExtract, "alpha", "extract-alpha", resolvedValidatorForTest(validator))
 
-	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: resolvedPipeline()})
+	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: pipeline})
 	if err != nil {
 		t.Fatalf("Run() error = %v, want nil", err)
 	}
@@ -933,10 +1041,12 @@ func TestRunOmitsLaneWithNoAcceptedExtractOutputs(t *testing.T) {
 
 func TestRunRejectedMergePreventsNormalizeForLane(t *testing.T) {
 	modules := defaultRunnerModules()
-	validator := &runnerRawValidator{name: "raw-merge", approved: []bool{false}, reason: "bad_merge", message: "merge rejected"}
+	validator := &runnerChainValidator{name: "chain-merge", approved: []bool{false}, reason: "bad_merge", message: "merge rejected"}
-	modules.rawValidators = rawValidationRegistry(t, StageMerge, "merge", validator)
+	modules.validators[validator.name] = validator
+	pipeline := resolvedPipeline()
+	setResolvedValidatorChain(t, &pipeline, StageMerge, "alpha", "merge", resolvedValidatorForTest(validator))
 
-	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: resolvedPipeline()})
+	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: pipeline})
 	if err != nil {
 		t.Fatalf("Run() error = %v, want nil", err)
 	}
@@ -954,10 +1064,12 @@ func TestRunRejectedMergePreventsNormalizeForLane(t *testing.T) {
 
 func TestRunRejectedNormalizePreventsOutputForLane(t *testing.T) {
 	modules := defaultRunnerModules()
-	validator := &runnerRawValidator{name: "raw-normalize", approved: []bool{false}, reason: "bad_normalize", message: "normalize rejected"}
+	validator := &runnerChainValidator{name: "chain-normalize", approved: []bool{false}, reason: "bad_normalize", message: "normalize rejected"}
-	modules.rawValidators = rawValidationRegistry(t, StageNormalize, "normalize", validator)
+	modules.validators[validator.name] = validator
+	pipeline := resolvedPipeline()
+	setResolvedValidatorChain(t, &pipeline, StageNormalize, "alpha", "normalize", resolvedValidatorForTest(validator))
 
-	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: resolvedPipeline()})
+	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: pipeline})
 	if err != nil {
 		t.Fatalf("Run() error = %v, want nil", err)
 	}
@@ -999,9 +1111,10 @@ func TestRunRetriesSameModuleInputAfterFrameworkError(t *testing.T) {
 
 func TestRunRetriesSameModuleInputAfterValidatorRejection(t *testing.T) {
 	modules := defaultRunnerModules()
-	validator := &runnerRawValidator{name: "raw-extract", approved: []bool{false, true, true}, reason: "bad_extract", message: "extract rejected"}
+	validator := &runnerChainValidator{name: "chain-extract", approved: []bool{false, true, true}, reason: "bad_extract", message: "extract rejected"}
-	modules.rawValidators = rawValidationRegistry(t, StageExtract, "extract-alpha", validator)
+	modules.validators[validator.name] = validator
 	pipeline := resolvedPipeline()
+	setResolvedValidatorChain(t, &pipeline, StageExtract, "alpha", "extract-alpha", resolvedValidatorForTest(validator))
 	pipeline.ArtifactLanes[0].Extract.Retries = 1
 
 	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: pipeline})
@@ -1023,9 +1136,10 @@ func TestRunRetriesSameModuleInputAfterValidatorRejection(t *testing.T) {
 
 func TestRunStopsRetryAfterConfiguredAttemptsAndRecordsAttemptCount(t *testing.T) {
 	modules := defaultRunnerModules()
-	validator := &runnerRawValidator{name: "raw-extract", approved: []bool{false}, reason: "bad_extract", message: "extract rejected"}
+	validator := &runnerChainValidator{name: "chain-extract", approved: []bool{false}, reason: "bad_extract", message: "extract rejected"}
-	modules.rawValidators = rawValidationRegistry(t, StageExtract, "extract-alpha", validator)
+	modules.validators[validator.name] = validator
 	pipeline := resolvedPipeline()
+	setResolvedValidatorChain(t, &pipeline, StageExtract, "alpha", "extract-alpha", resolvedValidatorForTest(validator))
 	pipeline.ArtifactLanes[0].Extract.Retries = 1
 
 	output, err := New(newRunnerRegistries(t, modules)).Run(context.Background(), RunInput{Pipeline: pipeline})
@@ -1075,7 +1189,7 @@ func TestRunRejectsConfiguredValidators(t *testing.T) {
 	_, err := New(newRunnerRegistries(t, defaultRunnerModules())).Run(context.Background(), RunInput{
 		Pipeline: resolvedPipelineWithValidators("configured", "second-validator"),
 	})
-	assertRunError(t, err, "configured validators")
+	assertRunError(t, err, "extract.validators")
 }
 
 func TestRunCollectsStageWarnings(t *testing.T) {
@@ -1312,8 +1426,11 @@ func TestRunManifestIncludesPipelineAndLaneDetails(t *testing.T) {
 	if lane.ID != "alpha" || lane.Extractor != "extract-alpha" || lane.Merger != "merge" || lane.Normalizer != "normalize" {
 		t.Fatalf("ArtifactLanes[0] = %#v, want lane details", lane)
 	}
-	if len(lane.Validators) != 0 {
+	if len(manifest.ValidatorChains) != 4 {
-		t.Fatalf("lane validators = %#v, want none", lane.Validators)
+		t.Fatalf("ValidatorChains = %#v, want four validation points", manifest.ValidatorChains)
+	}
+	if manifest.ValidatorChains[0].Stage != string(StageChunk) || manifest.ValidatorChains[0].ModuleKey != "chunk" || len(manifest.ValidatorChains[0].Validators) != 0 {
+		t.Fatalf("chunk validator chain = %#v, want explicit empty chunk chain", manifest.ValidatorChains[0])
 	}
 }
 
@@ -1454,6 +1571,12 @@ func resolvedPipeline() ResolvedPipeline {
 				NormalizeReferences: referenceTarget(StageNormalize, "alpha", "normalize", nil),
 			},
 		},
+		ValidatorChains: []ResolvedValidatorChain{
+			{Stage: StageChunk, ModuleKey: "chunk"},
+			{Stage: StageExtract, LaneID: "alpha", ModuleKey: "extract-alpha"},
+			{Stage: StageMerge, LaneID: "alpha", ModuleKey: "merge"},
+			{Stage: StageNormalize, LaneID: "alpha", ModuleKey: "normalize"},
+		},
 		Output: Binding("output"),
 	}
 }
@@ -1493,8 +1616,7 @@ type runnerModules struct {
 	extractors      map[string]*runnerExtractor
 	mergers         map[string]*runnerMerger
 	normalizers     map[string]*runnerNormalizer
-	validators      map[string]*runnerValidator
+	validators      map[string]contracts.Validator
-	rawValidators   *RawValidationRegistry
 	output          *runnerOutputEncoder
 	inputBuildErr   error
 	chunkerBuildErr error
@@ -1513,9 +1635,9 @@ func defaultRunnerModules() *runnerModules {
 		normalizers: map[string]*runnerNormalizer{
 			"normalize": {key: "normalize"},
 		},
-		validators: map[string]*runnerValidator{
+		validators: map[string]contracts.Validator{
-			"configured":       {name: "configured"},
+			"configured":       &runnerValidator{name: "configured"},
-			"second-validator": {name: "second-validator"},
+			"second-validator": &runnerValidator{name: "second-validator"},
 		},
 		output: &runnerOutputEncoder{
 			key: "output",
@@ -1539,7 +1661,7 @@ func newRunnerRegistries(t *testing.T, modules *runnerModules) Registries {
 		Mergers:         NewMergerRegistry(),
 		Normalizers:     NewNormalizerRegistry(),
 		Validators:      NewValidatorRegistry(),
-		RawValidators: modules.rawValidators,
+		ValidatorChains: NewValidatorChainRegistry(),
 		Outputs:         NewOutputEncoderRegistry(),
 	}
 	if err := registries.Inputs.Register("input", func() (contracts.InputAdapter, error) {
@@ -1578,7 +1700,8 @@ func newRunnerRegistries(t *testing.T, modules *runnerModules) Registries {
 	}
 	for key, validator := range modules.validators {
 		validator := validator
-		if err := registries.Validators.Register(key, func() (contracts.Validator, error) { return validator, nil }); err != nil {
+		spec := ValidatorSpec{Key: key, ExecutionClass: validator.ExecutionClass()}
+		if err := registries.Validators.RegisterWithSpec(spec, func() (contracts.Validator, error) { return validator, nil }); err != nil {
 			t.Fatalf("register validator %q: %v", key, err)
 		}
 	}
@@ -1797,8 +1920,10 @@ func (normalizer *runnerNormalizer) Normalize(ctx context.Context, req contracts
 
 type runnerValidator struct {
 	name           string
-	resultName string
+	executionClass contracts.ExecutionClass
-	decisions  func([]artifacts.ArtifactCandidate) []contracts.ValidationDecision
+	approved       []bool
+	reason         string
+	message        string
 	warnings       []contracts.Warning
 	err            error
 	order          *[]string
@@ -1806,26 +1931,34 @@ type runnerValidator struct {
 	requests       []contracts.ValidationRequest
 }
 
-type runnerRawValidator struct {
+type runnerChainValidator struct {
 	name           string
+	executionClass contracts.ExecutionClass
 	approved       []bool
 	reason         string
 	message        string
 	warnings       []contracts.Warning
 	err            error
 	calls          int
-	requests []contracts.RawValidationRequest
+	requests       []contracts.ValidationRequest
 }
 
-func (validator *runnerRawValidator) Name() string {
+func (validator *runnerChainValidator) Name() string {
 	return validator.name
 }
 
-func (validator *runnerRawValidator) ValidateRaw(ctx context.Context, req contracts.RawValidationRequest) (contracts.RawValidationResult, error) {
+func (validator *runnerChainValidator) ExecutionClass() contracts.ExecutionClass {
+	if validator.executionClass != "" {
+		return validator.executionClass
+	}
+	return contracts.ExecutionClassDeterministic
+}
+
+func (validator *runnerChainValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
 	validator.calls++
 	validator.requests = append(validator.requests, req)
 	if validator.err != nil {
-		return contracts.RawValidationResult{}, validator.err
+		return contracts.ValidationResult{}, validator.err
 	}
 	approved := true
 	if len(validator.approved) > 0 {
@@ -1835,7 +1968,7 @@ func (validator *runnerRawValidator) ValidateRaw(ctx context.Context, req contra
 		}
 		approved = validator.approved[index]
 	}
-	return contracts.RawValidationResult{
+	return contracts.ValidationResult{
 		Approved:   approved,
 		ReasonCode: validator.reason,
 		Message:    validator.message,
@@ -1847,23 +1980,31 @@ func (validator *runnerValidator) Name() string {
 	return validator.name
 }
 
+func (validator *runnerValidator) ExecutionClass() contracts.ExecutionClass {
+	if validator.executionClass != "" {
+		return validator.executionClass
+	}
+	return contracts.ExecutionClassDeterministic
+}
+
 func (validator *runnerValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
 	validator.calls++
 	validator.requests = append(validator.requests, req)
 	if validator.order != nil {
 		*validator.order = append(*validator.order, validator.name)
 	}
-	resultName := validator.resultName
+	approved := true
-	if resultName == "" {
+	if len(validator.approved) > 0 {
-		resultName = validator.name
+		index := validator.calls - 1
+		if index >= len(validator.approved) {
+			index = len(validator.approved) - 1
 		}
-	var decisions []contracts.ValidationDecision
+		approved = validator.approved[index]
-	if validator.decisions != nil {
-		decisions = validator.decisions(req.Candidates)
 	}
 	return contracts.ValidationResult{
-		ValidatorName: resultName,
+		Approved:   approved,
-		Decisions:     decisions,
+		ReasonCode: validator.reason,
+		Message:    validator.message,
 		Warnings:   validator.warnings,
 	}, validator.err
 }
@@ -2027,12 +2168,31 @@ func assertRunError(t *testing.T, err error, want string) {
 	}
 }
 
-func rawValidationRegistry(t *testing.T, stage ModuleStage, module string, validators ...contracts.RawValidator) *RawValidationRegistry {
+func resolvedValidatorForTest(validator contracts.Validator) ResolvedValidator {
+	return ResolvedValidator{
+		Binding:        Binding(validator.Name()),
+		ExecutionClass: validator.ExecutionClass(),
+	}
+}
+
+func setResolvedValidatorChain(t *testing.T, resolved *ResolvedPipeline, stage ModuleStage, laneID string, module string, validators ...ResolvedValidator) {
 	t.Helper()
 
-	registry := NewRawValidationRegistry()
+	if resolved == nil {
-	if err := registry.Register(stage, module, validators...); err != nil {
+		t.Fatal("resolved pipeline must not be nil")
-		t.Fatalf("register raw validators: %v", err)
 	}
-	return registry
+	chain := ResolvedValidatorChain{
+		Stage:      stage,
+		LaneID:     laneID,
+		ModuleKey:  module,
+		Validators: append([]ResolvedValidator(nil), validators...),
+	}
+	for index := range resolved.ValidatorChains {
+		existing := resolved.ValidatorChains[index]
+		if existing.Stage == stage && existing.LaneID == laneID && existing.ModuleKey == module {
+			resolved.ValidatorChains[index] = chain
+			return
+		}
+	}
+	resolved.ValidatorChains = append(resolved.ValidatorChains, chain)
 }

internal/framework/pipeline/validator_chain_registry.go

@@ -0,0 +1,110 @@
+package pipeline
+
+import (
+	"fmt"
+	"strings"
+)
+
+type validatorChainKey struct {
+	stage  ModuleStage
+	module string
+}
+
+type ValidatorChainMapping struct {
+	Stage      ModuleStage     `json:"stage"`
+	Module     string          `json:"module"`
+	Validators []ModuleBinding `json:"validators,omitempty"`
+}
+
+type ValidatorChainRegistry struct {
+	chains map[validatorChainKey][]ModuleBinding
+}
+
+func NewValidatorChainRegistry() *ValidatorChainRegistry {
+	return &ValidatorChainRegistry{
+		chains: make(map[validatorChainKey][]ModuleBinding),
+	}
+}
+
+func (r *ValidatorChainRegistry) Register(mapping ValidatorChainMapping) error {
+	if r == nil {
+		return fmt.Errorf("validator chain registry must not be nil")
+	}
+	normalized, err := normalizeValidatorChainMapping(mapping)
+	if err != nil {
+		return err
+	}
+	if r.chains == nil {
+		r.chains = make(map[validatorChainKey][]ModuleBinding)
+	}
+	key := validatorChainKey{stage: normalized.Stage, module: normalized.Module}
+	if _, exists := r.chains[key]; exists {
+		return fmt.Errorf("validator chain for %q %q is already registered", normalized.Stage, normalized.Module)
+	}
+	r.chains[key] = cloneModuleBindings(normalized.Validators)
+	return nil
+}
+
+func (r *ValidatorChainRegistry) Validators(stage ModuleStage, module string) []ModuleBinding {
+	if r == nil {
+		return nil
+	}
+	chain := r.chains[validatorChainKey{stage: stage, module: strings.TrimSpace(module)}]
+	return cloneModuleBindings(chain)
+}
+
+func normalizeValidatorChainMapping(mapping ValidatorChainMapping) (ValidatorChainMapping, error) {
+	normalized := ValidatorChainMapping{
+		Stage:      mapping.Stage,
+		Module:     strings.TrimSpace(mapping.Module),
+		Validators: cloneModuleBindings(mapping.Validators),
+	}
+	switch normalized.Stage {
+	case StageChunk, StageExtract, StageMerge, StageNormalize:
+	default:
+		return ValidatorChainMapping{}, fmt.Errorf("validator chain stage %q is not supported", normalized.Stage)
+	}
+	if normalized.Module == "" {
+		return ValidatorChainMapping{}, fmt.Errorf("validator chain module key must not be empty")
+	}
+	for i, validator := range normalized.Validators {
+		if strings.TrimSpace(validator.Module) == "" {
+			return ValidatorChainMapping{}, fmt.Errorf("validator chain for %q %q has empty validator key at index %d", normalized.Stage, normalized.Module, i)
+		}
+		normalized.Validators[i] = resolveBinding(validator, "")
+	}
+	return normalized, nil
+}
+
+func cloneModuleBindings(bindings []ModuleBinding) []ModuleBinding {
+	if len(bindings) == 0 {
+		return nil
+	}
+	out := make([]ModuleBinding, len(bindings))
+	for i, binding := range bindings {
+		out[i] = cloneModuleBinding(binding)
+	}
+	return out
+}
+
+func cloneModuleBinding(binding ModuleBinding) ModuleBinding {
+	binding.Module = strings.TrimSpace(binding.Module)
+	binding.LLMProfile = strings.TrimSpace(binding.LLMProfile)
+	binding.Options = cloneOptions(binding.Options)
+	if len(binding.References) > 0 {
+		references := make(map[string]string, len(binding.References))
+		for key, value := range binding.References {
+			references[key] = value
+		}
+		binding.References = references
+	}
+	binding.Validators = cloneValidatorOverride(binding.Validators)
+	return binding
+}
+
+func cloneValidatorOverride(override ValidatorOverride) ValidatorOverride {
+	return ValidatorOverride{
+		Set:        override.Set,
+		Validators: cloneModuleBindings(override.Validators),
+	}
+}

internal/framework/pipeline/validator_chain_registry_test.go

@@ -0,0 +1,74 @@
+package pipeline
+
+import "testing"
+
+func TestValidatorChainRegistryRegistersAndLooksUpChains(t *testing.T) {
+	registry := NewValidatorChainRegistry()
+	err := registry.Register(ValidatorChainMapping{
+		Stage:      StageExtract,
+		Module:     " extractor ",
+		Validators: []ModuleBinding{{Module: " first "}, {Module: "second"}},
+	})
+	if err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+
+	got := registry.Validators(StageExtract, " extractor ")
+	if len(got) != 2 || got[0].Module != "first" || got[1].Module != "second" {
+		t.Fatalf("Validators() = %#v, want trimmed chain", got)
+	}
+}
+
+func TestValidatorChainRegistryRejectsDuplicateMappings(t *testing.T) {
+	registry := NewValidatorChainRegistry()
+	mapping := ValidatorChainMapping{Stage: StageMerge, Module: "merge", Validators: []ModuleBinding{Binding("validator")}}
+	if err := registry.Register(mapping); err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+
+	if err := registry.Register(mapping); err == nil {
+		t.Fatal("Register() error = nil, want duplicate mapping error")
+	}
+}
+
+func TestValidatorChainRegistryRejectsUnsupportedStage(t *testing.T) {
+	registry := NewValidatorChainRegistry()
+	err := registry.Register(ValidatorChainMapping{Stage: StageInput, Module: "input"})
+	if err == nil {
+		t.Fatal("Register() error = nil, want unsupported stage error")
+	}
+}
+
+func TestValidatorChainRegistryAllowsAbsentAndEmptyChains(t *testing.T) {
+	registry := NewValidatorChainRegistry()
+	if got := registry.Validators(StageNormalize, "normalize"); got != nil {
+		t.Fatalf("absent chain = %#v, want nil", got)
+	}
+	if err := registry.Register(ValidatorChainMapping{Stage: StageNormalize, Module: "normalize"}); err != nil {
+		t.Fatalf("Register(empty) error = %v, want nil", err)
+	}
+	if got := registry.Validators(StageNormalize, "normalize"); got != nil {
+		t.Fatalf("empty chain = %#v, want nil", got)
+	}
+}
+
+func TestValidatorChainRegistryReturnsDefensiveCopies(t *testing.T) {
+	registry := NewValidatorChainRegistry()
+	err := registry.Register(ValidatorChainMapping{
+		Stage:      StageChunk,
+		Module:     "chunk",
+		Validators: []ModuleBinding{{Module: "validator", Options: map[string]any{"level": "strict"}}},
+	})
+	if err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+
+	got := registry.Validators(StageChunk, "chunk")
+	got[0].Module = "changed"
+	got[0].Options["level"] = "changed"
+
+	again := registry.Validators(StageChunk, "chunk")
+	if again[0].Module != "validator" || again[0].Options["level"] != "strict" {
+		t.Fatalf("Validators() after caller mutation = %#v, want original chain", again)
+	}
+}

internal/framework/pipeline/validator_registry.go

@@ -2,6 +2,7 @@ package pipeline
 
 import (
 	"fmt"
+	"sort"
 	"strings"
 
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
@@ -9,29 +10,34 @@ import (
 
 type ValidatorConstructor func() (contracts.Validator, error)
 
+type ValidatorSpec struct {
+	Key            string                   `json:"key"`
+	ExecutionClass contracts.ExecutionClass `json:"execution_class"`
+}
+
 type ValidatorRegistry struct {
 	constructors map[string]ValidatorConstructor
-	specs        map[string]ModuleSpec
+	specs        map[string]ValidatorSpec
 }
 
 func NewValidatorRegistry() *ValidatorRegistry {
 	return &ValidatorRegistry{
 		constructors: make(map[string]ValidatorConstructor),
-		specs:        make(map[string]ModuleSpec),
+		specs:        make(map[string]ValidatorSpec),
 	}
 }
 
 func (r *ValidatorRegistry) Register(key string, constructor ValidatorConstructor) error {
-	return r.RegisterWithSpec(defaultModuleSpec(key, StageValidate), constructor)
+	return r.RegisterWithSpec(ValidatorSpec{Key: key, ExecutionClass: contracts.ExecutionClassDeterministic}, constructor)
 }
 
-func (r *ValidatorRegistry) RegisterWithSpec(spec ModuleSpec, constructor ValidatorConstructor) error {
+func (r *ValidatorRegistry) RegisterWithSpec(spec ValidatorSpec, constructor ValidatorConstructor) error {
 	if r == nil {
 		return fmt.Errorf("validator registry must not be nil")
 	}
 
-	normalizedSpec := normalizeModuleSpec(spec)
+	normalizedSpec, err := normalizeValidatorSpec(spec)
-	if err := validateModuleSpec("validator", StageValidate, normalizedSpec); err != nil {
+	if err != nil {
 		return err
 	}
 	if constructor == nil {
@@ -45,10 +51,10 @@ func (r *ValidatorRegistry) RegisterWithSpec(spec ModuleSpec, constructor Valida
 		r.constructors = make(map[string]ValidatorConstructor)
 	}
 	if r.specs == nil {
-		r.specs = make(map[string]ModuleSpec)
+		r.specs = make(map[string]ValidatorSpec)
 	}
 	r.constructors[normalizedSpec.Key] = constructor
-	r.specs[normalizedSpec.Key] = cloneModuleSpec(normalizedSpec)
+	r.specs[normalizedSpec.Key] = normalizedSpec
 	return nil
 }
 
@@ -77,20 +83,45 @@ func (r *ValidatorRegistry) Build(key string) (contracts.Validator, error) {
 	if validator.Name() != normalizedKey {
 		return nil, fmt.Errorf("validator %q returned name %q", normalizedKey, validator.Name())
 	}
+	spec, ok := r.specs[normalizedKey]
+	if !ok {
+		return nil, fmt.Errorf("validator %q spec is not registered", normalizedKey)
+	}
+	if validator.ExecutionClass() != spec.ExecutionClass {
+		return nil, fmt.Errorf("validator %q returned execution class %q, want %q", normalizedKey, validator.ExecutionClass(), spec.ExecutionClass)
+	}
 
 	return validator, nil
 }
 
-func (r *ValidatorRegistry) Spec(key string) (ModuleSpec, bool) {
+func (r *ValidatorRegistry) Spec(key string) (ValidatorSpec, bool) {
 	if r == nil {
-		return ModuleSpec{}, false
+		return ValidatorSpec{}, false
 	}
 
 	spec, ok := r.specs[strings.TrimSpace(key)]
 	if !ok {
-		return ModuleSpec{}, false
+		return ValidatorSpec{}, false
 	}
-	return cloneModuleSpec(spec), true
+	return spec, true
+}
+
+func (r *ValidatorRegistry) RegisteredSpecs() []ValidatorSpec {
+	if r == nil || len(r.specs) == 0 {
+		return nil
+	}
+
+	keys := make([]string, 0, len(r.specs))
+	for key := range r.specs {
+		keys = append(keys, key)
+	}
+	sort.Strings(keys)
+
+	specs := make([]ValidatorSpec, 0, len(keys))
+	for _, key := range keys {
+		specs = append(specs, r.specs[key])
+	}
+	return specs
 }
 
 func (r *ValidatorRegistry) RegisteredKeys() []string {
@@ -100,3 +131,19 @@ func (r *ValidatorRegistry) RegisteredKeys() []string {
 
 	return sortedRegistryKeys(r.constructors)
 }
+
+func normalizeValidatorSpec(spec ValidatorSpec) (ValidatorSpec, error) {
+	normalized := ValidatorSpec{
+		Key:            strings.TrimSpace(spec.Key),
+		ExecutionClass: spec.ExecutionClass,
+	}
+	if normalized.Key == "" {
+		return ValidatorSpec{}, fmt.Errorf("validator key must not be empty")
+	}
+	switch normalized.ExecutionClass {
+	case contracts.ExecutionClassDeterministic, contracts.ExecutionClassLLMBacked:
+	default:
+		return ValidatorSpec{}, fmt.Errorf("validator %q execution class %q is not supported", normalized.Key, normalized.ExecutionClass)
+	}
+	return normalized, nil
+}

internal/framework/pipeline/validator_registry_test.go

@@ -1,58 +1,117 @@
 package pipeline
 
 import (
+	"context"
+	"reflect"
+	"strings"
 	"testing"
 
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
 )
 
 func TestValidatorRegistryBehavior(t *testing.T) {
-	runRegistryBehaviorTests(t, registryBehaviorCase[contracts.Validator]{
+	registry := NewValidatorRegistry()
-		name:       "ValidatorRegistry",
+	if err := registry.Register(" generic-validator ", validatorConstructor("generic-validator", contracts.ExecutionClassDeterministic)); err != nil {
-		key:        "generic-validator",
+		t.Fatalf("Register() error = %v, want nil", err)
-		stage:      StageValidate,
+	}
-		wrongStage: StageExtract,
+
-		newRegistry: func() any {
+	validator, err := registry.Build("generic-validator")
-			return NewValidatorRegistry()
+	if err != nil {
-		},
+		t.Fatalf("Build() error = %v, want nil", err)
-		register: func(registry any, key string, constructor func() (contracts.Validator, error)) error {
+	}
-			return registry.(*ValidatorRegistry).Register(key, constructor)
+	if validator.Name() != "generic-validator" {
-		},
+		t.Fatalf("validator name = %q, want generic-validator", validator.Name())
-		registerWithSpec: func(registry any, spec ModuleSpec, constructor func() (contracts.Validator, error)) error {
+	}
-			return registry.(*ValidatorRegistry).RegisterWithSpec(spec, constructor)
+
-		},
+	spec, ok := registry.Spec(" generic-validator ")
-		build: func(registry any, key string) (contracts.Validator, error) {
+	if !ok {
-			return registry.(*ValidatorRegistry).Build(key)
+		t.Fatal("Spec() ok = false, want true")
-		},
+	}
-		spec: func(registry any, key string) (ModuleSpec, bool) {
+	want := ValidatorSpec{Key: "generic-validator", ExecutionClass: contracts.ExecutionClassDeterministic}
-			return registry.(*ValidatorRegistry).Spec(key)
+	if !reflect.DeepEqual(spec, want) {
-		},
+		t.Fatalf("Spec() = %#v, want %#v", spec, want)
-		registeredKeys: func(registry any) []string {
+	}
-			return registry.(*ValidatorRegistry).RegisteredKeys()
+}
-		},
+
-		nilRegister: func(key string, constructor func() (contracts.Validator, error)) error {
+func TestValidatorRegistryRegistersSpecs(t *testing.T) {
-			var registry *ValidatorRegistry
+	registry := NewValidatorRegistry()
-			return registry.Register(key, constructor)
+	spec := ValidatorSpec{Key: " llm-validator ", ExecutionClass: contracts.ExecutionClassLLMBacked}
-		},
+	if err := registry.RegisterWithSpec(spec, validatorConstructor("llm-validator", contracts.ExecutionClassLLMBacked)); err != nil {
-		nilBuild: func(key string) (contracts.Validator, error) {
+		t.Fatalf("RegisterWithSpec() error = %v, want nil", err)
-			var registry *ValidatorRegistry
+	}
-			return registry.Build(key)
+
-		},
+	got, ok := registry.Spec("llm-validator")
-		nilSpec: func(key string) (ModuleSpec, bool) {
+	if !ok {
-			var registry *ValidatorRegistry
+		t.Fatal("Spec() ok = false, want true")
-			return registry.Spec(key)
+	}
-		},
+	want := ValidatorSpec{Key: "llm-validator", ExecutionClass: contracts.ExecutionClassLLMBacked}
-		nilRegisteredKey: func() []string {
+	if !reflect.DeepEqual(got, want) {
-			var registry *ValidatorRegistry
+		t.Fatalf("Spec() = %#v, want %#v", got, want)
-			return registry.RegisteredKeys()
+	}
-		},
+}
-		constructor: func(key string) func() (contracts.Validator, error) {
+
+func TestValidatorRegistryRegisteredSpecsAreSorted(t *testing.T) {
+	registry := NewValidatorRegistry()
+	for _, key := range []string{"zeta", "alpha"} {
+		if err := registry.Register(key, validatorConstructor(key, contracts.ExecutionClassDeterministic)); err != nil {
+			t.Fatalf("Register(%q) error = %v", key, err)
+		}
+	}
+
+	specs := registry.RegisteredSpecs()
+	if len(specs) != 2 || specs[0].Key != "alpha" || specs[1].Key != "zeta" {
+		t.Fatalf("RegisteredSpecs() = %#v, want sorted specs", specs)
+	}
+}
+
+func TestValidatorRegistryRejectsUnsupportedExecutionClass(t *testing.T) {
+	registry := NewValidatorRegistry()
+	err := registry.RegisterWithSpec(
+		ValidatorSpec{Key: "invalid-validator", ExecutionClass: contracts.ExecutionClass("unsupported")},
+		validatorConstructor("invalid-validator", contracts.ExecutionClass("unsupported")),
+	)
+	if err == nil {
+		t.Fatal("RegisterWithSpec() error = nil, want unsupported execution class error")
+	}
+}
+
+func TestValidatorRegistryRejectsConstructorExecutionClassMismatch(t *testing.T) {
+	registry := NewValidatorRegistry()
+	if err := registry.RegisterWithSpec(
+		ValidatorSpec{Key: "validator", ExecutionClass: contracts.ExecutionClassDeterministic},
+		validatorConstructor("validator", contracts.ExecutionClassLLMBacked),
+	); err != nil {
+		t.Fatalf("RegisterWithSpec() error = %v, want nil", err)
+	}
+
+	_, err := registry.Build("validator")
+	if err == nil {
+		t.Fatal("Build() error = nil, want execution class mismatch")
+	}
+	if !strings.Contains(err.Error(), "execution class") {
+		t.Fatalf("Build() error = %q, want execution class context", err.Error())
+	}
+}
+
+type testValidator struct {
+	name           string
+	executionClass contracts.ExecutionClass
+}
+
+func validatorConstructor(name string, executionClass contracts.ExecutionClass) ValidatorConstructor {
 	return func() (contracts.Validator, error) {
-				return registryValidator{name: key}, nil
+		return testValidator{name: name, executionClass: executionClass}, nil
 	}
-		},
+}
-		moduleKey: func(module contracts.Validator) string {
+
-			return module.Name()
+func (validator testValidator) Name() string {
-		},
+	return validator.name
-	})
+}
+
+func (validator testValidator) ExecutionClass() contracts.ExecutionClass {
+	return validator.executionClass
+}
+
+func (validator testValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	return contracts.ValidationResult{Approved: true}, nil
 }

internal/framework/pipeline/walking_skeleton_test.go

@@ -102,6 +102,7 @@ func walkingSkeletonCatalog(t *testing.T) ModuleCatalog {
 		Extractors:      NewExtractorRegistry(),
 		Mergers:         NewMergerRegistry(),
 		Normalizers:     NewNormalizerRegistry(),
+		ValidatorChains: NewValidatorChainRegistry(),
 		Outputs:         NewOutputEncoderRegistry(),
 	}
 	if err := catalog.Inputs.RegisterWithSpec(ModuleSpec{

internal/framework/validate/validate.go

@@ -1,10 +1,8 @@
 package validate
 
 import (
-	"fmt"
 	"strings"
 
-	"gitea.maximumdirect.net/eric/notarius/internal/core/artifacts"
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
 )
 
@@ -12,53 +10,18 @@ const (
 	ReasonApproved = "approved"
 )
 
-func Approved(candidateIndex int) contracts.ValidationDecision {
+func Approved() contracts.ValidationResult {
-	return contracts.ValidationDecision{
+	return contracts.ValidationResult{
-		CandidateIndex: candidateIndex,
 		Approved:   true,
 		ReasonCode: ReasonApproved,
 		Message:    ReasonApproved,
 	}
 }
 
-func Rejected(candidateIndex int, reasonCode string, message string) contracts.ValidationDecision {
+func Rejected(reasonCode string, message string) contracts.ValidationResult {
-	return contracts.ValidationDecision{
+	return contracts.ValidationResult{
-		CandidateIndex: candidateIndex,
 		Approved:   false,
 		ReasonCode: strings.TrimSpace(reasonCode),
 		Message:    strings.TrimSpace(message),
 	}
 }
-
-func EnforceDecisionCardinality(candidates []artifacts.ArtifactCandidate, decisions []contracts.ValidationDecision) error {
-	if len(candidates) != len(decisions) {
-		return fmt.Errorf("validator returned %d decisions for %d candidates", len(decisions), len(candidates))
-	}
-
-	expected := make(map[int]struct{}, len(candidates))
-	for _, candidate := range candidates {
-		if _, ok := expected[candidate.Index]; ok {
-			return fmt.Errorf("candidate index %d is duplicated", candidate.Index)
-		}
-		expected[candidate.Index] = struct{}{}
-	}
-
-	seen := make(map[int]struct{}, len(decisions))
-	for _, decision := range decisions {
-		if _, ok := expected[decision.CandidateIndex]; !ok {
-			return fmt.Errorf("validator returned decision for unknown candidate index %d", decision.CandidateIndex)
-		}
-		if _, ok := seen[decision.CandidateIndex]; ok {
-			return fmt.Errorf("validator returned duplicate decision for candidate index %d", decision.CandidateIndex)
-		}
-		seen[decision.CandidateIndex] = struct{}{}
-	}
-
-	for candidateIndex := range expected {
-		if _, ok := seen[candidateIndex]; !ok {
-			return fmt.Errorf("validator did not return decision for candidate index %d", candidateIndex)
-		}
-	}
-
-	return nil
-}

internal/framework/validate/validate_test.go

@@ -1,105 +1,40 @@
 package validate
 
 import (
-	"strings"
 	"testing"
 
-	"gitea.maximumdirect.net/eric/notarius/internal/core/artifacts"
 	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
 )
 
 func TestApproved(t *testing.T) {
-	decision := Approved(7)
+	result := Approved()
 
-	if decision.CandidateIndex != 7 {
+	if !result.Approved {
-		t.Fatalf("CandidateIndex = %d, want 7", decision.CandidateIndex)
-	}
-	if !decision.Approved {
 		t.Fatal("Approved = false, want true")
 	}
-	if decision.ReasonCode != ReasonApproved {
+	if result.ReasonCode != ReasonApproved {
-		t.Fatalf("ReasonCode = %q, want %q", decision.ReasonCode, ReasonApproved)
+		t.Fatalf("ReasonCode = %q, want %q", result.ReasonCode, ReasonApproved)
 	}
-	if decision.Message != "approved" {
+	if result.Message != "approved" {
-		t.Fatalf("Message = %q, want approved", decision.Message)
+		t.Fatalf("Message = %q, want approved", result.Message)
 	}
 }
 
 func TestRejectedTrimsReasonAndMessage(t *testing.T) {
-	decision := Rejected(3, " invalid ", "\tmessage\n")
+	result := Rejected(" invalid ", "\tmessage\n")
 
-	if decision.CandidateIndex != 3 {
+	if result.Approved {
-		t.Fatalf("CandidateIndex = %d, want 3", decision.CandidateIndex)
-	}
-	if decision.Approved {
 		t.Fatal("Approved = true, want false")
 	}
-	if decision.ReasonCode != "invalid" {
+	if result.ReasonCode != "invalid" {
-		t.Fatalf("ReasonCode = %q, want invalid", decision.ReasonCode)
+		t.Fatalf("ReasonCode = %q, want invalid", result.ReasonCode)
 	}
-	if decision.Message != "message" {
+	if result.Message != "message" {
-		t.Fatalf("Message = %q, want message", decision.Message)
+		t.Fatalf("Message = %q, want message", result.Message)
 	}
 }
 
-func TestEnforceDecisionCardinalityAllowsNonZeroCandidateIndices(t *testing.T) {
+func TestHelpersReturnValidationResults(t *testing.T) {
-	candidates := []artifacts.ArtifactCandidate{{Index: 4}, {Index: 8}}
+	var _ contracts.ValidationResult = Approved()
-	decisions := []contracts.ValidationDecision{Approved(8), Approved(4)}
+	var _ contracts.ValidationResult = Rejected("reason", "message")
-
-	if err := EnforceDecisionCardinality(candidates, decisions); err != nil {
-		t.Fatalf("EnforceDecisionCardinality() error = %v, want nil", err)
-	}
-}
-
-func TestEnforceDecisionCardinalityAllowsEmptyInputs(t *testing.T) {
-	if err := EnforceDecisionCardinality(nil, nil); err != nil {
-		t.Fatalf("EnforceDecisionCardinality() error = %v, want nil", err)
-	}
-}
-
-func TestEnforceDecisionCardinalityRejectsUnknownDecisionIndex(t *testing.T) {
-	err := EnforceDecisionCardinality(
-		[]artifacts.ArtifactCandidate{{Index: 1}},
-		[]contracts.ValidationDecision{Approved(2)},
-	)
-
-	assertCardinalityError(t, err, "unknown candidate index 2")
-}
-
-func TestEnforceDecisionCardinalityRejectsDuplicateDecisionIndex(t *testing.T) {
-	err := EnforceDecisionCardinality(
-		[]artifacts.ArtifactCandidate{{Index: 1}, {Index: 2}},
-		[]contracts.ValidationDecision{Approved(1), Approved(1)},
-	)
-
-	assertCardinalityError(t, err, "duplicate decision")
-}
-
-func TestEnforceDecisionCardinalityRejectsMissingDecisionIndex(t *testing.T) {
-	err := EnforceDecisionCardinality(
-		[]artifacts.ArtifactCandidate{{Index: 1}, {Index: 2}},
-		[]contracts.ValidationDecision{Approved(1)},
-	)
-
-	assertCardinalityError(t, err, "1 decisions for 2 candidates")
-}
-
-func TestEnforceDecisionCardinalityRejectsDuplicateCandidateIndex(t *testing.T) {
-	err := EnforceDecisionCardinality(
-		[]artifacts.ArtifactCandidate{{Index: 1}, {Index: 1}},
-		[]contracts.ValidationDecision{Approved(1), Approved(1)},
-	)
-
-	assertCardinalityError(t, err, "candidate index 1 is duplicated")
-}
-
-func assertCardinalityError(t *testing.T, err error, want string) {
-	t.Helper()
-
-	if err == nil {
-		t.Fatal("EnforceDecisionCardinality() error = nil, want error")
-	}
-	if !strings.Contains(err.Error(), want) {
-		t.Fatalf("EnforceDecisionCardinality() error = %q, want substring %q", err.Error(), want)
-	}
 }

internal/modules/extract/dnd/spells/config_test.go

@@ -217,6 +217,7 @@ func dndSpellsTestCatalog(t *testing.T, specs dndSpellsCatalogSpecs) pipeline.Mo
 		Extractors:      extractors,
 		Mergers:         mergers,
 		Normalizers:     normalizers,
+		ValidatorChains: pipeline.NewValidatorChainRegistry(),
 		Outputs:         outputs,
 	}
 }

internal/modules/extract/dnd/spells/extractor.go

@@ -115,12 +115,17 @@ func (e *Extractor) Extract(ctx context.Context, req contracts.ExtractionRequest
 			return contracts.ExtractionResult{}, extractorErrorf("marshal raw output: %w", err)
 		}
 	}
+	schema, err := loadResponseSchema()
+	if err != nil {
+		return contracts.ExtractionResult{}, extractorErrorf("load response schema: %w", err)
+	}
 	return contracts.ExtractionResult{
 		Output: contracts.ExtractOutput{
 			Schema: contracts.ResponseSchema{
 				ID:         ResponseSchemaID,
 				Name:       ResponseSchemaName,
 				Version:    SchemaVersion,
+				JSONSchema: append([]byte(nil), schema.JSONSchema...),
 			},
 			Payload: contracts.RawPayload{
 				Content:   content,

internal/modules/extract/dnd/spells/extractor_test.go

@@ -60,6 +60,9 @@ func TestExtractReturnsRawOutputFromStructuredResponse(t *testing.T) {
 	if result.Output.Schema.ID != ResponseSchemaID || result.Output.Schema.Name != ResponseSchemaName || result.Output.Schema.Version != SchemaVersion {
 		t.Fatalf("schema = %#v, want response schema provenance", result.Output.Schema)
 	}
+	if !json.Valid(result.Output.Schema.JSONSchema) {
+		t.Fatalf("schema JSON is invalid or missing: %s", result.Output.Schema.JSONSchema)
+	}
 	if got := string(result.Output.Payload.Content); got != string(client.content) {
 		t.Fatalf("content = %q, want exact raw completion content", got)
 	}

internal/modules/extract/dnd/spells/validator.go

@@ -1,168 +0,0 @@
-package spells
-
-import (
-	"context"
-	"encoding/json"
-	"fmt"
-	"strings"
-
-	"gitea.maximumdirect.net/eric/notarius/internal/core/artifacts"
-	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
-	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
-	"gitea.maximumdirect.net/eric/notarius/internal/framework/validate"
-)
-
-const (
-	shapeValidatorName     = "dnd/spells/shape"
-	sourceRefValidatorName = "dnd/spells/source_refs"
-
-	reasonInvalidPayload       = "invalid_payload"
-	reasonMissingRequiredField = "missing_required_field"
-	reasonMissingSourceRef     = "missing_source_ref"
-	reasonInvalidSourceRef     = "invalid_source_ref"
-	reasonSpellNotNearSource   = "spell_not_near_source"
-)
-
-var _ contracts.Validator = ShapeValidator{}
-var _ contracts.Validator = SourceRefValidator{}
-
-type ShapeValidator struct{}
-
-func (validator ShapeValidator) Name() string {
-	return shapeValidatorName
-}
-
-func (validator ShapeValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
-	decisions := make([]contracts.ValidationDecision, 0, len(req.Candidates))
-	for _, candidate := range req.Candidates {
-		decisions = append(decisions, validateShape(candidate))
-	}
-	return contracts.ValidationResult{
-		ValidatorName: validator.Name(),
-		Decisions:     decisions,
-	}, nil
-}
-
-type SourceRefValidator struct{}
-
-func (validator SourceRefValidator) Name() string {
-	return sourceRefValidatorName
-}
-
-func (validator SourceRefValidator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
-	if req.Source == nil {
-		return contracts.ValidationResult{}, fmt.Errorf("dnd spells source refs validator: source must not be nil")
-	}
-
-	decisions := make([]contracts.ValidationDecision, 0, len(req.Candidates))
-	var warnings []contracts.Warning
-	for _, candidate := range req.Candidates {
-		decisions = append(decisions, validateSourceRefs(req.Source, candidate))
-		warnings = append(warnings, sourceRelatednessWarnings(req.Source, candidate)...)
-	}
-	return contracts.ValidationResult{
-		ValidatorName: validator.Name(),
-		Decisions:     decisions,
-		Warnings:      warnings,
-	}, nil
-}
-
-func validateShape(candidate artifacts.ArtifactCandidate) contracts.ValidationDecision {
-	var payload SpellCast
-	if err := json.Unmarshal(candidate.Payload, &payload); err != nil {
-		return validate.Rejected(candidate.Index, reasonInvalidPayload, fmt.Sprintf("invalid spell cast payload: %v", err))
-	}
-	for _, field := range requiredSpellCastFields(payload) {
-		if strings.TrimSpace(field.value) == "" {
-			return validate.Rejected(candidate.Index, reasonMissingRequiredField, fmt.Sprintf("missing required field %q", field.name))
-		}
-	}
-	return validate.Approved(candidate.Index)
-}
-
-func validateSourceRefs(doc *source.SourceDocument, candidate artifacts.ArtifactCandidate) contracts.ValidationDecision {
-	if len(candidate.SourceRefs) == 0 {
-		return validate.Rejected(candidate.Index, reasonMissingSourceRef, "spell cast candidate must include at least one source ref")
-	}
-	for _, ref := range candidate.SourceRefs {
-		if err := source.ValidateRef(doc, ref); err != nil {
-			return validate.Rejected(candidate.Index, reasonInvalidSourceRef, err.Error())
-		}
-	}
-	return validate.Approved(candidate.Index)
-}
-
-func sourceRelatednessWarnings(doc *source.SourceDocument, candidate artifacts.ArtifactCandidate) []contracts.Warning {
-	if doc == nil || len(candidate.SourceRefs) == 0 {
-		return nil
-	}
-	var payload SpellCast
-	if err := json.Unmarshal(candidate.Payload, &payload); err != nil {
-		return nil
-	}
-	spell := strings.TrimSpace(payload.Spell)
-	if spell == "" {
-		return nil
-	}
-
-	needle := strings.ToLower(spell)
-	for _, ref := range candidate.SourceRefs {
-		text, ok := sourceRefText(doc, ref)
-		if !ok {
-			continue
-		}
-		if strings.Contains(strings.ToLower(text), needle) {
-			return nil
-		}
-	}
-
-	return []contracts.Warning{
-		{
-			Scope:      fmt.Sprintf("candidate.%d", candidate.Index),
-			ReasonCode: reasonSpellNotNearSource,
-			Message:    fmt.Sprintf("spell %q was not found in the cited source text", spell),
-		},
-	}
-}
-
-func sourceRefText(doc *source.SourceDocument, ref source.SourceRef) (string, bool) {
-	if err := source.ValidateRef(doc, ref); err != nil {
-		return "", false
-	}
-	start := -1
-	end := -1
-	for i, unit := range doc.Units {
-		if unit.ID == ref.StartUnitID {
-			start = i
-		}
-		if unit.ID == ref.EndUnitID {
-			end = i
-		}
-	}
-	if start < 0 || end < start {
-		return "", false
-	}
-	var b strings.Builder
-	for i := start; i <= end; i++ {
-		if b.Len() > 0 {
-			b.WriteString("\n")
-		}
-		b.WriteString(doc.Units[i].Text)
-	}
-	return b.String(), true
-}
-
-func requiredSpellCastFields(payload SpellCast) []struct {
-	name  string
-	value string
-} {
-	return []struct {
-		name  string
-		value string
-	}{
-		{name: "caster", value: payload.Caster},
-		{name: "spell", value: payload.Spell},
-		{name: "effect", value: payload.Effect},
-		{name: "narrative_description", value: payload.NarrativeDescription},
-	}
-}

internal/modules/extract/dnd/spells/validator_test.go

@@ -1,277 +0,0 @@
-package spells
-
-import (
-	"context"
-	"encoding/json"
-	"strings"
-	"testing"
-
-	"gitea.maximumdirect.net/eric/notarius/internal/core/artifacts"
-	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
-	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
-	"gitea.maximumdirect.net/eric/notarius/internal/framework/validate"
-)
-
-func TestValidatorsApproveValidCandidate(t *testing.T) {
-	candidate := validSpellCandidate(7)
-
-	shapeResult, err := ShapeValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-		Candidates: []artifacts.ArtifactCandidate{candidate},
-	})
-	if err != nil {
-		t.Fatalf("ShapeValidator.Validate() error = %v, want nil", err)
-	}
-	assertSingleDecision(t, shapeResult, shapeValidatorName, 7, true, validate.ReasonApproved)
-
-	sourceRefResult, err := SourceRefValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-		Source:     promptSourceDocument(),
-		Candidates: []artifacts.ArtifactCandidate{candidate},
-	})
-	if err != nil {
-		t.Fatalf("SourceRefValidator.Validate() error = %v, want nil", err)
-	}
-	assertSingleDecision(t, sourceRefResult, sourceRefValidatorName, 7, true, validate.ReasonApproved)
-	if len(sourceRefResult.Warnings) != 0 {
-		t.Fatalf("warnings = %#v, want none", sourceRefResult.Warnings)
-	}
-}
-
-func TestShapeValidatorRejectsMalformedPayload(t *testing.T) {
-	candidate := validSpellCandidate(3)
-	candidate.Payload = json.RawMessage(`{"caster":`)
-
-	result, err := ShapeValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-		Candidates: []artifacts.ArtifactCandidate{candidate},
-	})
-	if err != nil {
-		t.Fatalf("ShapeValidator.Validate() error = %v, want nil", err)
-	}
-	assertSingleDecision(t, result, shapeValidatorName, 3, false, reasonInvalidPayload)
-}
-
-func TestShapeValidatorRejectsBlankRequiredFields(t *testing.T) {
-	tests := []struct {
-		name   string
-		mutate func(*SpellCast)
-	}{
-		{name: "caster", mutate: func(payload *SpellCast) { payload.Caster = " \t" }},
-		{name: "spell", mutate: func(payload *SpellCast) { payload.Spell = "" }},
-		{name: "effect", mutate: func(payload *SpellCast) { payload.Effect = "\n" }},
-		{name: "narrative description", mutate: func(payload *SpellCast) { payload.NarrativeDescription = " " }},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			payload := validSpellPayload()
-			tt.mutate(&payload)
-			candidate := validSpellCandidate(5)
-			candidate.Payload = mustSpellPayload(t, payload)
-
-			result, err := ShapeValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-				Candidates: []artifacts.ArtifactCandidate{candidate},
-			})
-			if err != nil {
-				t.Fatalf("ShapeValidator.Validate() error = %v, want nil", err)
-			}
-			assertSingleDecision(t, result, shapeValidatorName, 5, false, reasonMissingRequiredField)
-		})
-	}
-}
-
-func TestShapeValidatorDoesNotRequireSourceDocument(t *testing.T) {
-	result, err := ShapeValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-		Candidates: []artifacts.ArtifactCandidate{validSpellCandidate(11)},
-	})
-	if err != nil {
-		t.Fatalf("ShapeValidator.Validate() error = %v, want nil", err)
-	}
-	assertSingleDecision(t, result, shapeValidatorName, 11, true, validate.ReasonApproved)
-}
-
-func TestSourceRefValidatorRejectsMissingRefs(t *testing.T) {
-	candidate := validSpellCandidate(13)
-	candidate.SourceRefs = nil
-
-	result, err := SourceRefValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-		Source:     promptSourceDocument(),
-		Candidates: []artifacts.ArtifactCandidate{candidate},
-	})
-	if err != nil {
-		t.Fatalf("SourceRefValidator.Validate() error = %v, want nil", err)
-	}
-	assertSingleDecision(t, result, sourceRefValidatorName, 13, false, reasonMissingSourceRef)
-}
-
-func TestSourceRefValidatorRejectsInvalidRefs(t *testing.T) {
-	tests := []struct {
-		name string
-		ref  source.SourceRef
-		want string
-	}{
-		{
-			name: "unknown source id",
-			ref:  source.SourceRef{SourceID: "session-beta", StartUnitID: 1, EndUnitID: 2},
-			want: "does not match",
-		},
-		{
-			name: "unknown start unit",
-			ref:  source.SourceRef{SourceID: "session-alpha", StartUnitID: 999, EndUnitID: 2},
-			want: "start_unit_id",
-		},
-		{
-			name: "unknown end unit",
-			ref:  source.SourceRef{SourceID: "session-alpha", StartUnitID: 1, EndUnitID: 999},
-			want: "end_unit_id",
-		},
-		{
-			name: "reversed unit range",
-			ref:  source.SourceRef{SourceID: "session-alpha", StartUnitID: 2, EndUnitID: 1},
-			want: "appears after",
-		},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			candidate := validSpellCandidate(17)
-			candidate.SourceRefs = []source.SourceRef{tt.ref}
-
-			result, err := SourceRefValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-				Source:     promptSourceDocument(),
-				Candidates: []artifacts.ArtifactCandidate{candidate},
-			})
-			if err != nil {
-				t.Fatalf("SourceRefValidator.Validate() error = %v, want nil", err)
-			}
-			assertSingleDecision(t, result, sourceRefValidatorName, 17, false, reasonInvalidSourceRef)
-			if !strings.Contains(result.Decisions[0].Message, tt.want) {
-				t.Fatalf("Message = %q, want substring %q", result.Decisions[0].Message, tt.want)
-			}
-		})
-	}
-}
-
-func TestSourceRefValidatorWarnsWhenSpellNameIsNotInCitedSource(t *testing.T) {
-	candidate := validSpellCandidate(31)
-	payload := validSpellPayload()
-	payload.Spell = "Shield"
-	candidate.Payload = mustSpellPayload(t, payload)
-
-	result, err := SourceRefValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-		Source:     promptSourceDocument(),
-		Candidates: []artifacts.ArtifactCandidate{candidate},
-	})
-	if err != nil {
-		t.Fatalf("SourceRefValidator.Validate() error = %v, want nil", err)
-	}
-	assertSingleDecision(t, result, sourceRefValidatorName, 31, true, validate.ReasonApproved)
-	if len(result.Warnings) != 1 {
-		t.Fatalf("warnings = %#v, want one relatedness warning", result.Warnings)
-	}
-	warning := result.Warnings[0]
-	if warning.ReasonCode != reasonSpellNotNearSource || !strings.Contains(warning.Message, "Shield") {
-		t.Fatalf("warning = %#v, want spell relatedness warning", warning)
-	}
-}
-
-func TestSourceRefValidatorRequiresSourceDocument(t *testing.T) {
-	_, err := SourceRefValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-		Candidates: []artifacts.ArtifactCandidate{validSpellCandidate(19)},
-	})
-	if err == nil {
-		t.Fatal("SourceRefValidator.Validate() error = nil, want source error")
-	}
-	if !strings.Contains(err.Error(), "dnd spells") || !strings.Contains(err.Error(), "source") {
-		t.Fatalf("SourceRefValidator.Validate() error = %q, want source context", err.Error())
-	}
-}
-
-func TestValidatorsPreserveCandidateIndexes(t *testing.T) {
-	candidates := []artifacts.ArtifactCandidate{
-		validSpellCandidate(23),
-		validSpellCandidate(29),
-	}
-
-	shapeResult, err := ShapeValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-		Candidates: candidates,
-	})
-	if err != nil {
-		t.Fatalf("ShapeValidator.Validate() error = %v, want nil", err)
-	}
-	assertDecisionIndexes(t, shapeResult.Decisions, []int{23, 29})
-
-	sourceRefResult, err := SourceRefValidator{}.Validate(context.Background(), contracts.ValidationRequest{
-		Source:     promptSourceDocument(),
-		Candidates: candidates,
-	})
-	if err != nil {
-		t.Fatalf("SourceRefValidator.Validate() error = %v, want nil", err)
-	}
-	assertDecisionIndexes(t, sourceRefResult.Decisions, []int{23, 29})
-}
-
-func validSpellCandidate(index int) artifacts.ArtifactCandidate {
-	return artifacts.ArtifactCandidate{
-		Index:   index,
-		Payload: spellPayload(validSpellPayload()),
-		SourceRefs: []source.SourceRef{
-			{SourceID: "session-alpha", StartUnitID: 1, EndUnitID: 2},
-		},
-	}
-}
-
-func validSpellPayload() SpellCast {
-	return SpellCast{
-		Caster:               "Aria",
-		Spell:                "Cure Wounds",
-		Effect:               "Heals an injured ally.",
-		NarrativeDescription: "Aria restores the fighter after the fight.",
-	}
-}
-
-func mustSpellPayload(t *testing.T, payload SpellCast) json.RawMessage {
-	t.Helper()
-
-	return spellPayload(payload)
-}
-
-func spellPayload(payload SpellCast) json.RawMessage {
-	encoded, err := json.Marshal(payload)
-	if err != nil {
-		panic(err)
-	}
-	return encoded
-}
-
-func assertSingleDecision(t *testing.T, result contracts.ValidationResult, wantName string, wantIndex int, wantApproved bool, wantReason string) {
-	t.Helper()
-
-	if result.ValidatorName != wantName {
-		t.Fatalf("ValidatorName = %q, want %q", result.ValidatorName, wantName)
-	}
-	if len(result.Decisions) != 1 {
-		t.Fatalf("len(Decisions) = %d, want 1", len(result.Decisions))
-	}
-	decision := result.Decisions[0]
-	if decision.CandidateIndex != wantIndex {
-		t.Fatalf("CandidateIndex = %d, want %d", decision.CandidateIndex, wantIndex)
-	}
-	if decision.Approved != wantApproved {
-		t.Fatalf("Approved = %t, want %t", decision.Approved, wantApproved)
-	}
-	if decision.ReasonCode != wantReason {
-		t.Fatalf("ReasonCode = %q, want %q", decision.ReasonCode, wantReason)
-	}
-}
-
-func assertDecisionIndexes(t *testing.T, decisions []contracts.ValidationDecision, want []int) {
-	t.Helper()
-
-	if len(decisions) != len(want) {
-		t.Fatalf("len(Decisions) = %d, want %d", len(decisions), len(want))
-	}
-	for i := range want {
-		if decisions[i].CandidateIndex != want[i] {
-			t.Fatalf("Decisions[%d].CandidateIndex = %d, want %d", i, decisions[i].CandidateIndex, want[i])
-		}
-	}
-}

internal/modules/input/seriatim/config_test.go

@@ -156,6 +156,7 @@ func seriatimTestCatalog(t *testing.T, inputSpec pipeline.ModuleSpec) pipeline.M
 		Extractors:      extractors,
 		Mergers:         mergers,
 		Normalizers:     normalizers,
+		ValidatorChains: pipeline.NewValidatorChainRegistry(),
 		Outputs:         outputs,
 	}
 }

internal/modules/merge/appendorder/merger.go

@@ -189,18 +189,28 @@ func commonSchema(outputs []contracts.ExtractOutput) contracts.ResponseSchema {
 	}
 	schema := outputs[0].Schema
 	for _, output := range outputs[1:] {
-		if output.Schema != schema {
+		if !sameResponseSchema(output.Schema, schema) {
 			return contracts.ResponseSchema{}
 		}
 	}
 	return schema
 }
 
+func sameResponseSchema(left contracts.ResponseSchema, right contracts.ResponseSchema) bool {
+	return left.ID == right.ID && left.Name == right.Name && left.Version == right.Version && string(left.JSONSchema) == string(right.JSONSchema)
+}
+
 func cloneExtractOutput(output contracts.ExtractOutput) contracts.ExtractOutput {
+	output.Schema = cloneResponseSchema(output.Schema)
 	output.Payload = cloneRawPayload(output.Payload)
 	return output
 }
 
+func cloneResponseSchema(schema contracts.ResponseSchema) contracts.ResponseSchema {
+	schema.JSONSchema = append([]byte(nil), schema.JSONSchema...)
+	return schema
+}
+
 func cloneRawPayload(payload contracts.RawPayload) contracts.RawPayload {
 	return contracts.RawPayload{
 		Content:   append([]byte(nil), payload.Content...),

internal/validators/extract/dnd/spells/shape/validator.go

@@ -0,0 +1,60 @@
+package shape
+
+import (
+	"context"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+	"gitea.maximumdirect.net/eric/notarius/internal/validators/extract/dnd/spells/spellpayload"
+)
+
+const Key = "extract/dnd/spells/shape"
+const ReasonCode = "invalid_spell_shape"
+
+var _ contracts.Validator = (*Validator)(nil)
+
+type Validator struct{}
+
+func New() *Validator {
+	return &Validator{}
+}
+
+func (v *Validator) Name() string {
+	return Key
+}
+
+func (v *Validator) ExecutionClass() contracts.ExecutionClass {
+	return contracts.ExecutionClassDeterministic
+}
+
+func (v *Validator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	payload, err := spellpayload.ValidationRequestPayload(req)
+	if err != nil {
+		return rejection(err.Error()), nil
+	}
+	if err := spellpayload.ValidateShape(payload); err != nil {
+		return rejection(err.Error()), nil
+	}
+	return contracts.ValidationResult{Approved: true}, nil
+}
+
+func Spec() pipeline.ValidatorSpec {
+	return pipeline.ValidatorSpec{
+		Key:            Key,
+		ExecutionClass: contracts.ExecutionClassDeterministic,
+	}
+}
+
+func Register(registry *pipeline.ValidatorRegistry) error {
+	return registry.RegisterWithSpec(Spec(), func() (contracts.Validator, error) {
+		return New(), nil
+	})
+}
+
+func rejection(message string) contracts.ValidationResult {
+	return contracts.ValidationResult{
+		Approved:   false,
+		ReasonCode: ReasonCode,
+		Message:    message,
+	}
+}

internal/validators/extract/dnd/spells/shape/validator_test.go

@@ -0,0 +1,68 @@
+package shape
+
+import (
+	"context"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+func TestValidatorApprovesWellFormedSpellPayload(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(`{"spell_casts":[{"caster":"Aria","spell":"Cure Wounds","effect":"heals","narrative_description":"Aria heals Borin.","source_refs":[{"source_id":"session","start_unit_id":1,"end_unit_id":1}]}]}`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if !result.Approved {
+		t.Fatalf("Validate() = %#v, want approved", result)
+	}
+}
+
+func TestValidatorRejectsMalformedPayload(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(`{"spell_casts":`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if result.Approved {
+		t.Fatalf("Approved = true, want false")
+	}
+	if result.ReasonCode != ReasonCode {
+		t.Fatalf("ReasonCode = %q, want %q", result.ReasonCode, ReasonCode)
+	}
+}
+
+func TestValidatorRejectsMissingRequiredSpellFields(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(`{"spell_casts":[{"caster":"Aria","effect":"heals","narrative_description":"Aria heals Borin.","source_refs":[{"source_id":"session","start_unit_id":1,"end_unit_id":1}]}]}`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if result.Approved {
+		t.Fatalf("Approved = true, want false")
+	}
+	if result.ReasonCode != ReasonCode {
+		t.Fatalf("ReasonCode = %q, want %q", result.ReasonCode, ReasonCode)
+	}
+}
+
+func TestSpecAndRegister(t *testing.T) {
+	registry := pipeline.NewValidatorRegistry()
+	if err := Register(registry); err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+	validator, err := registry.Build(Key)
+	if err != nil {
+		t.Fatalf("Build(%q) error = %v, want nil", Key, err)
+	}
+	if validator.Name() != Key || validator.ExecutionClass() != contracts.ExecutionClassDeterministic {
+		t.Fatalf("validator = %q/%q, want key and deterministic execution", validator.Name(), validator.ExecutionClass())
+	}
+}
+
+func requestWithPayload(payload string) contracts.ValidationRequest {
+	return contracts.ValidationRequest{
+		Payload: contracts.RawPayload{
+			Content:   []byte(payload),
+			MediaType: "application/json",
+		},
+	}
+}

internal/validators/extract/dnd/spells/source_refs/validator.go

@@ -0,0 +1,69 @@
+package sourcerefs
+
+import (
+	"context"
+	"fmt"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+	"gitea.maximumdirect.net/eric/notarius/internal/validators/extract/dnd/spells/spellpayload"
+)
+
+const Key = "extract/dnd/spells/source_refs"
+const ReasonCode = "invalid_source_refs"
+
+var _ contracts.Validator = (*Validator)(nil)
+
+type Validator struct{}
+
+func New() *Validator {
+	return &Validator{}
+}
+
+func (v *Validator) Name() string {
+	return Key
+}
+
+func (v *Validator) ExecutionClass() contracts.ExecutionClass {
+	return contracts.ExecutionClassDeterministic
+}
+
+func (v *Validator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	payload, err := spellpayload.ValidationRequestPayload(req)
+	if err != nil {
+		return rejection(err.Error()), nil
+	}
+	if err := spellpayload.ValidateShape(payload); err != nil {
+		return rejection(err.Error()), nil
+	}
+	for spellIndex, spell := range payload.SpellCasts {
+		for refIndex, ref := range spellpayload.SourceRefCandidates(req.Source, spell) {
+			if err := source.ValidateRef(req.Source, ref); err != nil {
+				return rejection(fmt.Sprintf("spell_casts[%d].source_refs[%d]: %v", spellIndex, refIndex, err)), nil
+			}
+		}
+	}
+	return contracts.ValidationResult{Approved: true}, nil
+}
+
+func Spec() pipeline.ValidatorSpec {
+	return pipeline.ValidatorSpec{
+		Key:            Key,
+		ExecutionClass: contracts.ExecutionClassDeterministic,
+	}
+}
+
+func Register(registry *pipeline.ValidatorRegistry) error {
+	return registry.RegisterWithSpec(Spec(), func() (contracts.Validator, error) {
+		return New(), nil
+	})
+}
+
+func rejection(message string) contracts.ValidationResult {
+	return contracts.ValidationResult{
+		Approved:   false,
+		ReasonCode: ReasonCode,
+		Message:    message,
+	}
+}

internal/validators/extract/dnd/spells/source_refs/validator_test.go

@@ -0,0 +1,83 @@
+package sourcerefs
+
+import (
+	"context"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+func TestValidatorApprovesValidSourceRefs(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(validDocument(), `{"spell_casts":[{"caster":"Aria","spell":"Cure Wounds","effect":"heals","narrative_description":"Aria casts Cure Wounds.","source_refs":[{"source_id":"session","start_unit_id":1,"end_unit_id":2}]}]}`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if !result.Approved {
+		t.Fatalf("Validate() = %#v, want approved", result)
+	}
+}
+
+func TestValidatorRejectsInvalidSourceRefs(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(validDocument(), `{"spell_casts":[{"caster":"Aria","spell":"Cure Wounds","effect":"heals","narrative_description":"Aria casts Cure Wounds.","source_refs":[{"source_id":"session","start_unit_id":99,"end_unit_id":99}]}]}`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if result.Approved {
+		t.Fatalf("Approved = true, want false")
+	}
+	if result.ReasonCode != ReasonCode {
+		t.Fatalf("ReasonCode = %q, want %q", result.ReasonCode, ReasonCode)
+	}
+}
+
+func TestValidatorRejectsMissingSourceDocument(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(nil, `{"spell_casts":[{"caster":"Aria","spell":"Cure Wounds","effect":"heals","narrative_description":"Aria casts Cure Wounds.","source_refs":[{"source_id":"session","start_unit_id":1,"end_unit_id":1}]}]}`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if result.Approved {
+		t.Fatalf("Approved = true, want false")
+	}
+	if result.ReasonCode != ReasonCode {
+		t.Fatalf("ReasonCode = %q, want %q", result.ReasonCode, ReasonCode)
+	}
+}
+
+func TestSpecAndRegister(t *testing.T) {
+	registry := pipeline.NewValidatorRegistry()
+	if err := Register(registry); err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+	validator, err := registry.Build(Key)
+	if err != nil {
+		t.Fatalf("Build(%q) error = %v, want nil", Key, err)
+	}
+	if validator.Name() != Key || validator.ExecutionClass() != contracts.ExecutionClassDeterministic {
+		t.Fatalf("validator = %q/%q, want key and deterministic execution", validator.Name(), validator.ExecutionClass())
+	}
+}
+
+func requestWithPayload(doc *source.SourceDocument, payload string) contracts.ValidationRequest {
+	return contracts.ValidationRequest{
+		Source: doc,
+		Payload: contracts.RawPayload{
+			Content:   []byte(payload),
+			MediaType: "application/json",
+		},
+	}
+}
+
+func validDocument() *source.SourceDocument {
+	return &source.SourceDocument{
+		ID:     "session",
+		Kind:   "transcript",
+		Format: "application/json",
+		Digest: "sha256:session",
+		Units: []source.SourceUnit{
+			{ID: 1, Kind: "message", Text: "Aria raises her holy symbol."},
+			{ID: 2, Kind: "message", Text: "Aria casts Cure Wounds on Borin."},
+		},
+	}
+}

internal/validators/extract/dnd/spells/source_relatedness/validator.go

@@ -0,0 +1,83 @@
+package sourcerelatedness
+
+import (
+	"context"
+	"fmt"
+	"strings"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+	"gitea.maximumdirect.net/eric/notarius/internal/validators/extract/dnd/spells/spellpayload"
+)
+
+const Key = "extract/dnd/spells/source_relatedness"
+const WarningReasonCode = "spell_not_near_source"
+
+var _ contracts.Validator = (*Validator)(nil)
+
+type Validator struct{}
+
+func New() *Validator {
+	return &Validator{}
+}
+
+func (v *Validator) Name() string {
+	return Key
+}
+
+func (v *Validator) ExecutionClass() contracts.ExecutionClass {
+	return contracts.ExecutionClassDeterministic
+}
+
+func (v *Validator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	payload, err := spellpayload.ValidationRequestPayload(req)
+	if err != nil {
+		return contracts.ValidationResult{Approved: true}, nil
+	}
+	if err := spellpayload.ValidateShape(payload); err != nil {
+		return contracts.ValidationResult{Approved: true}, nil
+	}
+
+	var warnings []contracts.Warning
+	for spellIndex, spell := range payload.SpellCasts {
+		if !spellAppearsInCitedText(req.Source, spell) {
+			warnings = append(warnings, contracts.Warning{
+				Scope:      fmt.Sprintf("spell_casts[%d]", spellIndex),
+				ReasonCode: WarningReasonCode,
+				Message:    fmt.Sprintf("spell %q was not found in cited source text", strings.TrimSpace(spell.Spell)),
+			})
+		}
+	}
+	return contracts.ValidationResult{Approved: true, Warnings: warnings}, nil
+}
+
+func Spec() pipeline.ValidatorSpec {
+	return pipeline.ValidatorSpec{
+		Key:            Key,
+		ExecutionClass: contracts.ExecutionClassDeterministic,
+	}
+}
+
+func Register(registry *pipeline.ValidatorRegistry) error {
+	return registry.RegisterWithSpec(Spec(), func() (contracts.Validator, error) {
+		return New(), nil
+	})
+}
+
+func spellAppearsInCitedText(doc *source.SourceDocument, spell spellpayload.SpellCast) bool {
+	name := strings.ToLower(strings.TrimSpace(spell.Spell))
+	if name == "" {
+		return true
+	}
+	for _, ref := range spellpayload.SourceRefCandidates(doc, spell) {
+		text, ok := spellpayload.CitedText(doc, ref)
+		if !ok {
+			continue
+		}
+		if strings.Contains(strings.ToLower(text), name) {
+			return true
+		}
+	}
+	return false
+}

internal/validators/extract/dnd/spells/source_relatedness/validator_test.go

@@ -0,0 +1,86 @@
+package sourcerelatedness
+
+import (
+	"context"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+func TestValidatorApprovesWithoutWarningWhenSpellAppearsInCitedText(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(validDocument(), `{"spell_casts":[{"caster":"Aria","spell":"Cure Wounds","effect":"heals","narrative_description":"Aria casts Cure Wounds.","source_refs":[{"source_id":"session","start_unit_id":2,"end_unit_id":2}]}]}`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if !result.Approved {
+		t.Fatalf("Approved = false, want true")
+	}
+	if len(result.Warnings) != 0 {
+		t.Fatalf("Warnings = %#v, want none", result.Warnings)
+	}
+}
+
+func TestValidatorWarnsWhenSpellDoesNotAppearInCitedText(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(validDocument(), `{"spell_casts":[{"caster":"Borin","spell":"Fire Bolt","effect":"scorches","narrative_description":"Borin casts Fire Bolt.","source_refs":[{"source_id":"session","start_unit_id":1,"end_unit_id":1}]}]}`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if !result.Approved {
+		t.Fatalf("Approved = false, want true")
+	}
+	if len(result.Warnings) != 1 {
+		t.Fatalf("Warnings = %#v, want one warning", result.Warnings)
+	}
+	if result.Warnings[0].ReasonCode != WarningReasonCode {
+		t.Fatalf("ReasonCode = %q, want %q", result.Warnings[0].ReasonCode, WarningReasonCode)
+	}
+}
+
+func TestValidatorApprovesMalformedPayloadWithoutWarning(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(validDocument(), `{"spell_casts":`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if !result.Approved {
+		t.Fatalf("Approved = false, want true")
+	}
+}
+
+func TestSpecAndRegister(t *testing.T) {
+	registry := pipeline.NewValidatorRegistry()
+	if err := Register(registry); err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+	validator, err := registry.Build(Key)
+	if err != nil {
+		t.Fatalf("Build(%q) error = %v, want nil", Key, err)
+	}
+	if validator.Name() != Key || validator.ExecutionClass() != contracts.ExecutionClassDeterministic {
+		t.Fatalf("validator = %q/%q, want key and deterministic execution", validator.Name(), validator.ExecutionClass())
+	}
+}
+
+func requestWithPayload(doc *source.SourceDocument, payload string) contracts.ValidationRequest {
+	return contracts.ValidationRequest{
+		Source: doc,
+		Payload: contracts.RawPayload{
+			Content:   []byte(payload),
+			MediaType: "application/json",
+		},
+	}
+}
+
+func validDocument() *source.SourceDocument {
+	return &source.SourceDocument{
+		ID:     "session",
+		Kind:   "transcript",
+		Format: "application/json",
+		Digest: "sha256:session",
+		Units: []source.SourceUnit{
+			{ID: 1, Kind: "message", Text: "Borin draws his dagger."},
+			{ID: 2, Kind: "message", Text: "Aria casts Cure Wounds on Borin."},
+		},
+	}
+}

internal/validators/extract/dnd/spells/spellpayload/payload.go

@@ -0,0 +1,98 @@
+package spellpayload
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"io"
+	"strings"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/core/source"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/modules/sharedassets/dnd"
+)
+
+type Payload struct {
+	SpellCasts []SpellCast `json:"spell_casts"`
+}
+
+type SpellCast struct {
+	Caster               string                  `json:"caster"`
+	Spell                string                  `json:"spell"`
+	Effect               string                  `json:"effect"`
+	NarrativeDescription string                  `json:"narrative_description"`
+	SourceRefs           []dnd.SourceRefResponse `json:"source_refs"`
+}
+
+func Parse(raw []byte) (Payload, error) {
+	decoder := json.NewDecoder(bytes.NewReader(raw))
+	decoder.DisallowUnknownFields()
+
+	var payload Payload
+	if err := decoder.Decode(&payload); err != nil {
+		return Payload{}, fmt.Errorf("parse spell payload: %w", err)
+	}
+	var extra any
+	if err := decoder.Decode(&extra); err != io.EOF {
+		return Payload{}, fmt.Errorf("parse spell payload: multiple JSON values")
+	}
+	return payload, nil
+}
+
+func ValidateShape(payload Payload) error {
+	if payload.SpellCasts == nil {
+		return fmt.Errorf("spell_casts must be present")
+	}
+	for index, spell := range payload.SpellCasts {
+		if strings.TrimSpace(spell.Caster) == "" {
+			return fmt.Errorf("spell_casts[%d].caster must not be empty", index)
+		}
+		if strings.TrimSpace(spell.Spell) == "" {
+			return fmt.Errorf("spell_casts[%d].spell must not be empty", index)
+		}
+		if strings.TrimSpace(spell.Effect) == "" {
+			return fmt.Errorf("spell_casts[%d].effect must not be empty", index)
+		}
+		if strings.TrimSpace(spell.NarrativeDescription) == "" {
+			return fmt.Errorf("spell_casts[%d].narrative_description must not be empty", index)
+		}
+		if len(spell.SourceRefs) == 0 {
+			return fmt.Errorf("spell_casts[%d].source_refs must not be empty", index)
+		}
+	}
+	return nil
+}
+
+func SourceRefCandidates(doc *source.SourceDocument, spell SpellCast) []source.SourceRef {
+	refs := make([]source.SourceRef, 0, len(spell.SourceRefs))
+	for _, ref := range spell.SourceRefs {
+		refs = append(refs, dnd.SourceRefCandidate(doc, ref))
+	}
+	return refs
+}
+
+func CitedText(doc *source.SourceDocument, ref source.SourceRef) (string, bool) {
+	if doc == nil {
+		return "", false
+	}
+	startIndex, ok := source.UnitIndex(doc, ref.StartUnitID)
+	if !ok {
+		return "", false
+	}
+	endIndex, ok := source.UnitIndex(doc, ref.EndUnitID)
+	if !ok || startIndex > endIndex {
+		return "", false
+	}
+	var b strings.Builder
+	for i := startIndex; i <= endIndex; i++ {
+		if b.Len() > 0 {
+			b.WriteByte('\n')
+		}
+		b.WriteString(doc.Units[i].Text)
+	}
+	return b.String(), true
+}
+
+func ValidationRequestPayload(req contracts.ValidationRequest) (Payload, error) {
+	return Parse(req.Payload.Content)
+}

internal/validators/generic/always_accept/validator.go

@@ -0,0 +1,43 @@
+package alwaysaccept
+
+import (
+	"context"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+const Key = "generic/always_accept"
+
+var _ contracts.Validator = (*Validator)(nil)
+
+type Validator struct{}
+
+func New() *Validator {
+	return &Validator{}
+}
+
+func (v *Validator) Name() string {
+	return Key
+}
+
+func (v *Validator) ExecutionClass() contracts.ExecutionClass {
+	return contracts.ExecutionClassDeterministic
+}
+
+func (v *Validator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	return contracts.ValidationResult{Approved: true}, nil
+}
+
+func Spec() pipeline.ValidatorSpec {
+	return pipeline.ValidatorSpec{
+		Key:            Key,
+		ExecutionClass: contracts.ExecutionClassDeterministic,
+	}
+}
+
+func Register(registry *pipeline.ValidatorRegistry) error {
+	return registry.RegisterWithSpec(Spec(), func() (contracts.Validator, error) {
+		return New(), nil
+	})
+}

internal/validators/generic/always_accept/validator_test.go

@@ -0,0 +1,40 @@
+package alwaysaccept
+
+import (
+	"context"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+func TestValidatorApproves(t *testing.T) {
+	result, err := New().Validate(context.Background(), contracts.ValidationRequest{})
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if !result.Approved {
+		t.Fatalf("Approved = false, want true")
+	}
+	if result.ReasonCode != "" || result.Message != "" {
+		t.Fatalf("result = %#v, want approval without rejection details", result)
+	}
+}
+
+func TestSpecAndRegister(t *testing.T) {
+	if Spec().Key != Key || Spec().ExecutionClass != contracts.ExecutionClassDeterministic {
+		t.Fatalf("Spec() = %#v, want key and deterministic execution", Spec())
+	}
+
+	registry := pipeline.NewValidatorRegistry()
+	if err := Register(registry); err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+	validator, err := registry.Build(Key)
+	if err != nil {
+		t.Fatalf("Build(%q) error = %v, want nil", Key, err)
+	}
+	if validator.Name() != Key {
+		t.Fatalf("Name() = %q, want %q", validator.Name(), Key)
+	}
+}

internal/validators/generic/always_reject/validator.go

@@ -0,0 +1,48 @@
+package alwaysreject
+
+import (
+	"context"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+const Key = "generic/always_reject"
+const ReasonCode = "always_reject"
+
+var _ contracts.Validator = (*Validator)(nil)
+
+type Validator struct{}
+
+func New() *Validator {
+	return &Validator{}
+}
+
+func (v *Validator) Name() string {
+	return Key
+}
+
+func (v *Validator) ExecutionClass() contracts.ExecutionClass {
+	return contracts.ExecutionClassDeterministic
+}
+
+func (v *Validator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	return contracts.ValidationResult{
+		Approved:   false,
+		ReasonCode: ReasonCode,
+		Message:    "output rejected by always-reject validator",
+	}, nil
+}
+
+func Spec() pipeline.ValidatorSpec {
+	return pipeline.ValidatorSpec{
+		Key:            Key,
+		ExecutionClass: contracts.ExecutionClassDeterministic,
+	}
+}
+
+func Register(registry *pipeline.ValidatorRegistry) error {
+	return registry.RegisterWithSpec(Spec(), func() (contracts.Validator, error) {
+		return New(), nil
+	})
+}

internal/validators/generic/always_reject/validator_test.go

@@ -0,0 +1,43 @@
+package alwaysreject
+
+import (
+	"context"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+func TestValidatorRejects(t *testing.T) {
+	result, err := New().Validate(context.Background(), contracts.ValidationRequest{})
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if result.Approved {
+		t.Fatalf("Approved = true, want false")
+	}
+	if result.ReasonCode != ReasonCode {
+		t.Fatalf("ReasonCode = %q, want %q", result.ReasonCode, ReasonCode)
+	}
+	if result.Message == "" {
+		t.Fatal("Message = empty, want rejection message")
+	}
+}
+
+func TestSpecAndRegister(t *testing.T) {
+	if Spec().Key != Key || Spec().ExecutionClass != contracts.ExecutionClassDeterministic {
+		t.Fatalf("Spec() = %#v, want key and deterministic execution", Spec())
+	}
+
+	registry := pipeline.NewValidatorRegistry()
+	if err := Register(registry); err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+	validator, err := registry.Build(Key)
+	if err != nil {
+		t.Fatalf("Build(%q) error = %v, want nil", Key, err)
+	}
+	if validator.Name() != Key {
+		t.Fatalf("Name() = %q, want %q", validator.Name(), Key)
+	}
+}

internal/validators/generic/valid_json/validator.go

@@ -0,0 +1,52 @@
+package validjson
+
+import (
+	"context"
+	"encoding/json"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+const Key = "generic/valid_json"
+const ReasonCodeInvalidJSON = "invalid_json"
+
+var _ contracts.Validator = (*Validator)(nil)
+
+type Validator struct{}
+
+func New() *Validator {
+	return &Validator{}
+}
+
+func (v *Validator) Name() string {
+	return Key
+}
+
+func (v *Validator) ExecutionClass() contracts.ExecutionClass {
+	return contracts.ExecutionClassDeterministic
+}
+
+func (v *Validator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	if !json.Valid(req.Payload.Content) {
+		return contracts.ValidationResult{
+			Approved:   false,
+			ReasonCode: ReasonCodeInvalidJSON,
+			Message:    "payload is not valid JSON",
+		}, nil
+	}
+	return contracts.ValidationResult{Approved: true}, nil
+}
+
+func Spec() pipeline.ValidatorSpec {
+	return pipeline.ValidatorSpec{
+		Key:            Key,
+		ExecutionClass: contracts.ExecutionClassDeterministic,
+	}
+}
+
+func Register(registry *pipeline.ValidatorRegistry) error {
+	return registry.RegisterWithSpec(Spec(), func() (contracts.Validator, error) {
+		return New(), nil
+	})
+}

internal/validators/generic/valid_json/validator_test.go

@@ -0,0 +1,66 @@
+package validjson
+
+import (
+	"context"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+func TestValidatorAcceptsValidJSON(t *testing.T) {
+	tests := []string{
+		`{"value":true}`,
+		`[1,2,3]`,
+		`"value"`,
+	}
+	for _, payload := range tests {
+		result, err := New().Validate(context.Background(), requestWithPayload(payload))
+		if err != nil {
+			t.Fatalf("Validate(%s) error = %v, want nil", payload, err)
+		}
+		if !result.Approved {
+			t.Fatalf("Validate(%s) = %#v, want approved", payload, result)
+		}
+	}
+}
+
+func TestValidatorRejectsInvalidJSON(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithPayload(`{"value":`))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if result.Approved {
+		t.Fatalf("Approved = true, want false")
+	}
+	if result.ReasonCode != ReasonCodeInvalidJSON {
+		t.Fatalf("ReasonCode = %q, want %q", result.ReasonCode, ReasonCodeInvalidJSON)
+	}
+}
+
+func TestSpecAndRegister(t *testing.T) {
+	if Spec().Key != Key || Spec().ExecutionClass != contracts.ExecutionClassDeterministic {
+		t.Fatalf("Spec() = %#v, want key and deterministic execution", Spec())
+	}
+
+	registry := pipeline.NewValidatorRegistry()
+	if err := Register(registry); err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+	validator, err := registry.Build(Key)
+	if err != nil {
+		t.Fatalf("Build(%q) error = %v, want nil", Key, err)
+	}
+	if validator.Name() != Key {
+		t.Fatalf("Name() = %q, want %q", validator.Name(), Key)
+	}
+}
+
+func requestWithPayload(payload string) contracts.ValidationRequest {
+	return contracts.ValidationRequest{
+		Payload: contracts.RawPayload{
+			Content:   []byte(payload),
+			MediaType: "application/json",
+		},
+	}
+}

internal/validators/generic/valid_json_schema/validator.go

@@ -0,0 +1,81 @@
+package validjsonschema
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+
+	"github.com/santhosh-tekuri/jsonschema/v6"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+const Key = "generic/valid_json_schema"
+const ReasonCodeInvalidJSON = "invalid_json"
+const ReasonCodeSchemaInvalid = "json_schema_invalid"
+
+var _ contracts.Validator = (*Validator)(nil)
+
+type Validator struct{}
+
+func New() *Validator {
+	return &Validator{}
+}
+
+func (v *Validator) Name() string {
+	return Key
+}
+
+func (v *Validator) ExecutionClass() contracts.ExecutionClass {
+	return contracts.ExecutionClassDeterministic
+}
+
+func (v *Validator) Validate(ctx context.Context, req contracts.ValidationRequest) (contracts.ValidationResult, error) {
+	if len(req.Schema.JSONSchema) == 0 {
+		return contracts.ValidationResult{}, fmt.Errorf("response schema content is not available")
+	}
+
+	instance, err := jsonschema.UnmarshalJSON(bytes.NewReader(req.Payload.Content))
+	if err != nil {
+		return contracts.ValidationResult{
+			Approved:   false,
+			ReasonCode: ReasonCodeInvalidJSON,
+			Message:    "payload is not valid JSON",
+		}, nil
+	}
+
+	schemaDocument, err := jsonschema.UnmarshalJSON(bytes.NewReader(req.Schema.JSONSchema))
+	if err != nil {
+		return contracts.ValidationResult{}, fmt.Errorf("parse response schema: %w", err)
+	}
+	compiler := jsonschema.NewCompiler()
+	if err := compiler.AddResource("schema.json", schemaDocument); err != nil {
+		return contracts.ValidationResult{}, fmt.Errorf("load response schema: %w", err)
+	}
+	schema, err := compiler.Compile("schema.json")
+	if err != nil {
+		return contracts.ValidationResult{}, fmt.Errorf("compile response schema: %w", err)
+	}
+	if err := schema.Validate(instance); err != nil {
+		return contracts.ValidationResult{
+			Approved:   false,
+			ReasonCode: ReasonCodeSchemaInvalid,
+			Message:    "payload does not conform to response schema",
+		}, nil
+	}
+	return contracts.ValidationResult{Approved: true}, nil
+}
+
+func Spec() pipeline.ValidatorSpec {
+	return pipeline.ValidatorSpec{
+		Key:            Key,
+		ExecutionClass: contracts.ExecutionClassDeterministic,
+	}
+}
+
+func Register(registry *pipeline.ValidatorRegistry) error {
+	return registry.RegisterWithSpec(Spec(), func() (contracts.Validator, error) {
+		return New(), nil
+	})
+}

internal/validators/generic/valid_json_schema/validator_test.go

@@ -0,0 +1,111 @@
+package validjsonschema
+
+import (
+	"context"
+	"strings"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/contracts"
+	"gitea.maximumdirect.net/eric/notarius/internal/framework/pipeline"
+)
+
+func TestValidatorAcceptsSchemaConformantJSON(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithSchema(`{"name":"Aria"}`, objectSchema()))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if !result.Approved {
+		t.Fatalf("Validate() = %#v, want approved", result)
+	}
+}
+
+func TestValidatorRejectsInvalidPayloadJSON(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithSchema(`{"name":`, objectSchema()))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if result.Approved {
+		t.Fatalf("Approved = true, want false")
+	}
+	if result.ReasonCode != ReasonCodeInvalidJSON {
+		t.Fatalf("ReasonCode = %q, want %q", result.ReasonCode, ReasonCodeInvalidJSON)
+	}
+}
+
+func TestValidatorRejectsSchemaNonConformance(t *testing.T) {
+	result, err := New().Validate(context.Background(), requestWithSchema(`{"name":3}`, objectSchema()))
+	if err != nil {
+		t.Fatalf("Validate() error = %v, want nil", err)
+	}
+	if result.Approved {
+		t.Fatalf("Approved = true, want false")
+	}
+	if result.ReasonCode != ReasonCodeSchemaInvalid {
+		t.Fatalf("ReasonCode = %q, want %q", result.ReasonCode, ReasonCodeSchemaInvalid)
+	}
+}
+
+func TestValidatorErrorsWhenSchemaContentMissing(t *testing.T) {
+	_, err := New().Validate(context.Background(), requestWithSchema(`{"name":"Aria"}`, nil))
+	if err == nil {
+		t.Fatal("Validate() error = nil, want missing schema content error")
+	}
+	if !strings.Contains(err.Error(), "schema content") {
+		t.Fatalf("Validate() error = %q, want schema content context", err.Error())
+	}
+}
+
+func TestValidatorErrorsWhenSchemaContentIsMalformed(t *testing.T) {
+	_, err := New().Validate(context.Background(), requestWithSchema(`{"name":"Aria"}`, []byte(`{"type":`)))
+	if err == nil {
+		t.Fatal("Validate() error = nil, want malformed schema error")
+	}
+	if !strings.Contains(err.Error(), "parse response schema") {
+		t.Fatalf("Validate() error = %q, want parse schema context", err.Error())
+	}
+}
+
+func TestSpecAndRegister(t *testing.T) {
+	if Spec().Key != Key || Spec().ExecutionClass != contracts.ExecutionClassDeterministic {
+		t.Fatalf("Spec() = %#v, want key and deterministic execution", Spec())
+	}
+
+	registry := pipeline.NewValidatorRegistry()
+	if err := Register(registry); err != nil {
+		t.Fatalf("Register() error = %v, want nil", err)
+	}
+	validator, err := registry.Build(Key)
+	if err != nil {
+		t.Fatalf("Build(%q) error = %v, want nil", Key, err)
+	}
+	if validator.Name() != Key {
+		t.Fatalf("Name() = %q, want %q", validator.Name(), Key)
+	}
+}
+
+func requestWithSchema(payload string, schema []byte) contracts.ValidationRequest {
+	return contracts.ValidationRequest{
+		Schema: contracts.ResponseSchema{
+			ID:         "test.schema",
+			Name:       "test_schema",
+			Version:    "v1",
+			JSONSchema: append([]byte(nil), schema...),
+		},
+		Payload: contracts.RawPayload{
+			Content:   []byte(payload),
+			MediaType: "application/json",
+		},
+	}
+}
+
+func objectSchema() []byte {
+	return []byte(`{
+		"$schema": "https://json-schema.org/draft/2020-12/schema",
+		"type": "object",
+		"required": ["name"],
+		"properties": {
+			"name": {"type": "string"}
+		},
+		"additionalProperties": false
+	}`)
+}