Spec Kitty

---

description: "Work packages for 054-constitution-interview-compiler-and-bootstrap"

---

Work Packages: Constitution Interview Compiler and Context Bootstrap

Inputs: Design documents from kitty-specs/054-constitution-interview-compiler-and-bootstrap/ Prerequisites: plan.md ✅ · spec.md ✅ · research.md ✅ · data-model.md ✅ · quickstart.md ✅ · contracts/constitution-cli-contract.md ✅

Tests: This feature explicitly requires CLI, compiler, context, migration, and integration regression coverage.

Organization: Fine-grained subtasks (Txxx) roll up into focused work packages (WPxx). Each work package is scoped so an implementer can complete it end-to-end in one focused session.

Prompt Files: Each work package references a matching prompt file in kitty-specs/054-constitution-interview-compiler-and-bootstrap/tasks/.

Subtask Format: [Txxx] [P?] Description

• [P] indicates the subtask can proceed in parallel once package prerequisites are met.
• Every subtask points to concrete modules, assets, or tests.

Path Conventions

• Runtime code: src/specify_cli/, src/doctrine/
• Planning artifacts: kitty-specs/054-constitution-interview-compiler-and-bootstrap/
• Tests: tests/specify_cli/, tests/integration/, tests/e2e/

---

Work Package WP01: Harden Interview and Generate CLI Contracts (Priority: P1) 🎯 MVP

Goal: Make the interview/generate workflow strict and scriptable: interview writes the canonical answers file, generate hard-fails when interview answers are missing, and overwrite/error behavior is deterministic. Independent Test: spec-kitty constitution interview --defaults --json returns the canonical answers payload; spec-kitty constitution generate fails without .kittify/constitution/interview/answers.yaml; when generated outputs already exist and --force is absent, the command warns listing conflicting files and prompts interactively — proceeding only on confirmation, aborting on rejection; --force bypasses the prompt entirely. Prompt: tasks/WP01-cli-contract-hardening.md Requirement Refs: FR-001, FR-002, FR-010, FR-011, SC-001, SC-002

Included Subtasks

• ✅ T001 Audit the current interview, generate, and generate-for-agent command flows in src/specify_cli/cli/commands/constitution.py
• ✅ T002 Normalize interview --json and failure payloads to the feature contract
• ✅ T003 Remove silent default fallback for generate / generate-for-agent when --from-interview is active and answers.yaml is missing
• ✅ T004 Enforce overwrite checks across all generated files, not just constitution.md
• ✅ T005 Add CLI regression tests for interview/generate hard-fail, JSON errors, and --force

Implementation Notes

• Use _interview_path(repo_root) as the single source for the path shown in human and JSON errors.
• Keep --no-from-interview behavior intact for explicit default-based generation paths.
• Align JSON failures on {"error": "<message>"} and success responses on {"result": "success", ...}.
• Ensure overwrite guards account for constitution.md, references.yaml, governance.yaml, directives.yaml, and metadata.yaml. Without --force, list all conflicting files in the warning before prompting; do not write anything until the user confirms.

Parallel Opportunities

• None inside this WP; all subtasks touch the same CLI contract surface.

Dependencies

• None.

Risks & Mitigations

• generate-for-agent shares most of the same control flow as generate; keep both commands aligned or reviewers will find inconsistent behavior.

Estimated Prompt Size

~250 lines

---

Work Package WP02: Model Local Support Declarations and Generated References (Priority: P1)

Goal: Add explicit local doctrine-support file declarations to the interview/compiler pipeline, emit them in references.yaml, report them via library_files, and stop materializing generated library/ content. Independent Test: A declared local markdown file appears in references.yaml and generate --json library_files; globs/directories are rejected; shipped-only runs emit an empty library_files; no generated library/ directory remains. Prompt: tasks/WP02-local-support-and-references.md Requirement Refs: FR-004, FR-017, FR-018, NFR-001, SC-001

Included Subtasks

• ✅ T006 Extend the interview data model in src/specify_cli/constitution/interview.py for local_supporting_files
• ✅ T007 Validate and normalize local support declarations: explicit file paths only, optional action scoping, optional target metadata
• ✅ T008 Build additive local support references and conflict warnings in src/specify_cli/constitution/compiler.py
• ✅ T009 Rewrite generated reference/output bookkeeping for references.yaml, generated_files, and library_files
• ✅ T010 Remove generated library/ materialization and stale library cleanup logic from bundle writing/status assumptions
• ✅ T011 Add compiler and CLI tests for local supporting files, overlap warnings, and empty shipped-only library_files

Implementation Notes

• Local support files are discoverable only from answers.yaml and mirrored references.yaml; there is no implicit filesystem scan.
• Path declarations must reject directories and glob-like values (, ?, [], recursive * patterns).
• If target_kind and target_id overlap a shipped artifact, shipped doctrine stays primary and the compiler surfaces a warning instead of replacing shipped content.
• Free-form markdown is acceptable for local files; do not force them into shipped YAML schemas.

Parallel Opportunities

• T006 and T007 can be split once the target field shape is agreed from data-model.md.

Dependencies

• Depends on WP01.

Risks & Mitigations

• The current compiler writes physical reference content into .kittify/constitution/library/; remove that behavior carefully and update any downstream callers/tests that still count those files.

Estimated Prompt Size

~320 lines

---

Work Package WP03: Enforce Shipped-Only Validation Authority (Priority: P1)

Goal: Make the shipped doctrine catalog the authoritative validation source by default, keep _proposed/ behind explicit opt-in, close template-set validation gaps, and confirm sync output excludes agents.yaml. Independent Test: Invalid shipped selections fail with named-ID errors; invalid template_set values fail even when template-set metadata is empty; _proposed/ content is excluded unless explicitly requested; sync output remains governance.yaml, directives.yaml, and metadata.yaml only. Prompt: tasks/WP03-shipped-validation-authority.md Requirement Refs: FR-003, FR-005, NFR-003, SC-003, SC-004

Included Subtasks

• ✅ T012 Audit src/specify_cli/constitution/catalog.py, resolver.py, and sync output expectations
• ✅ T013 Keep shipped-only catalog loading as the default contract, with explicit _proposed/ opt-in for curation flows
• ✅ T014 Add template-set filesystem fallback against packaged mission directories when catalog metadata is empty
• ✅ T015 Ensure shipped selection validation remains strict while local support files bypass catalog-ID enforcement
• ✅ T016 Update sync/status assumptions so agents.yaml is absent and any stale assertions are removed
• ✅ T017 Add catalog/resolver/sync regression tests for named-ID failures and shipped-only behavior

Implementation Notes

• Validation errors must name the offending value directly; vague “invalid selection” failures are insufficient.
• If mission directories are unavailable in a packaged install, preserve the current offline-friendly behavior instead of inventing a new hard failure mode.
• Do not reintroduce _proposed/ visibility through secondary validation paths.

Parallel Opportunities

• T013 and T014 can be developed in parallel after T012.

Dependencies

• None.

Risks & Mitigations

• load_doctrine_catalog() is already used in multiple places; keep behavior changes explicit and covered by tests so curation-only callers can still opt in intentionally.

Estimated Prompt Size

~280 lines

---

Work Package WP04: Extract Software-Dev Action Doctrine Assets (Priority: P1)

Goal: Remove embedded governance prose from the software-dev specify, plan, implement, and review command templates, and relocate that content into per-action doctrine assets with action indexes. Independent Test: The four source templates keep workflow/bootstrap instructions only; src/doctrine/missions/software-dev/actions/<action>/guidelines.md and index.yaml exist for each action; extracted guidance is retrievable at runtime instead of embedded inline. Prompt: tasks/WP04-action-doctrine-assets.md Requirement Refs: FR-012, FR-015, SC-008, SC-009

Included Subtasks

• ✅ T018 Audit the four software-dev command templates and mark governance prose to extract vs workflow content to preserve
• ✅ T019 Create actions/specify/ and actions/plan/ guidelines.md + index.yaml
• ✅ T020 Create actions/implement/ and actions/review/ guidelines.md + index.yaml
• ✅ T021 Strip extracted prose from src/doctrine/missions/software-dev/command-templates/*.md without removing bootstrap, $ARGUMENTS, or workflow steps
• ✅ T022 Populate action indexes with shipped doctrine IDs relevant to each action and verify the IDs are valid
• ✅ T023 Add source-asset tests or golden assertions for the extracted templates and action files

Implementation Notes

• This feature is software-dev mission only; do not touch other mission templates.
• Keep the exact ## Constitution Context Bootstrap (required) blocks in source templates.
• Preserve command-template structure so existing agent-generation machinery still copies valid prompts.
• Extraction uses a hybrid approach: a script (or migration step) detects governance prose sections and stubs them into guidelines.md files; a contributor then reviews and refines the stubbed content before committing. Do not hand-author guidelines.md from scratch without the stub-generation step.

Parallel Opportunities

• T019 and T020 are parallel once T018 has identified the extractable content.

Dependencies

• None.

Risks & Mitigations

• Over-extracting workflow instructions will break downstream command prompts; reviewers should diff the before/after template structure, not just search for removed headings.

Estimated Prompt Size

~300 lines

---

Work Package WP05: Implement Action-Scoped Iterative Context Retrieval (Priority: P1)

Goal: Rebuild constitution context around action indexes, per-type repository retrieval, explicit depth control, bootstrap state, and action-scoped local support files. Independent Test: First context --action plan call returns depth-2 bootstrap output; subsequent calls default to compact depth 1; --depth 3 adds styleguide/toolguide detail; plan does not leak implement-only doctrine; plan-scoped local support files appear only in plan. Prompt: tasks/WP05-action-scoped-context.md Requirement Refs: FR-006, FR-007, FR-013, FR-014, FR-015, FR-016, NFR-002, SC-005, SC-009, SC-010, SC-011

Included Subtasks

• ✅ T024 Add ActionIndex loading/export support under src/doctrine/missions/
• ✅ T025 Rework build_constitution_context() for bootstrap state, explicit --depth, and graceful degradation on missing/corrupt constitution artifacts
• ✅ T026 Fetch directives, tactics, styleguides, toolguides, and action guidelines through the correct repositories using action-index intersections
• ✅ T027 Include action-scoped local support references from references.yaml without letting them override shipped doctrine
• ✅ T028 Update constitution context CLI and downstream consumers for the new JSON shape (context, text, mode, depth)
• ✅ T029 Add context and consumer regression tests for depth semantics, action isolation, missing files, and local-support scoping

Implementation Notes

• Default depth is derived from bootstrap state only when --depth is omitted: first call -> depth 2, later calls -> depth 1.
• context-state.json is keyed by action and should remain deterministic JSON.
• context and text must contain the same rendered payload for compatibility.
• Each artifact type must be loaded from its own repository service; do not shortcut through generic file scans.

Parallel Opportunities

• T024 can proceed in parallel with design work for T025/T026.

Dependencies

• Depends on WP02.
• Depends on WP04.

Risks & Mitigations

• build_constitution_context() is used by both direct CLI commands and workflow/prompt-builder consumers; keep signatures and compatibility updates synchronized.

Estimated Prompt Size

~360 lines

---

Work Package WP06: Migrate Generated Agent Templates to Bootstrap Context (Priority: P2)

Goal: Ship an idempotent migration that updates generated agent prompt files across all configured agents to use bootstrap context, removes stale inline governance prose, and cleans up obsolete constitution library output. Independent Test: The migration updates all four command prompts per configured agent, inserts bootstrap calls once, strips targeted prose blocks, removes .kittify/constitution/library/ if present, and produces no duplicate changes on a second run. Prompt: tasks/WP06-template-bootstrap-migration.md Requirement Refs: FR-008, FR-009, SC-006, SC-007, SC-008

Included Subtasks

• ✅ T030 Audit existing migration patterns and the generated prompt inventory for configured agents
• ✅ T031 Implement migration detect/can-apply logic for stale inline-prose or missing bootstrap-call states
• ✅ T032 Implement idempotent prompt rewriting for the four command prompts across markdown and TOML agent formats
• ✅ T033 Remove obsolete .kittify/constitution/library/ artifacts during migration apply when present
• ✅ T034 Register the migration and add parametrized tests across all 12 supported agents
• ✅ T035 Add second-run idempotency and configured-agent filtering coverage

Implementation Notes

• Follow the migration style already used in src/specify_cli/upgrade/migrations/; do not invent a new migration harness.
• Action mapping must derive the correct --action from each prompt filename.
• A configured-agent project should not rewrite orphaned prompt directories for agents that are not active in .kittify/config.yaml.

Parallel Opportunities

• T034 and T035 can be split once T031-T033 are stable.

Dependencies

• Depends on WP04.
• Depends on WP05.

Risks & Mitigations

• Generated prompts live in multiple directory conventions and formats; keep the migration logic centralized and heavily fixture-tested rather than scattering format-specific ad hoc replacements.

Estimated Prompt Size

~300 lines

---

Work Package WP07: Close the Integration and Compatibility Gaps (Priority: P2)

Goal: Refresh cross-cutting tests and downstream compatibility points so the end-to-end interview -> generate -> context workflow, next-prompt consumers, and reconstruction tests all match the new contract. Independent Test: End-to-end tests cover explicit local support files, bootstrap-state transitions, absence of generated library/ and agents.yaml, and compatibility with prompt-building/workflow consumers. Prompt: tasks/WP07-integration-compatibility-sweep.md Requirement Refs: FR-010, FR-017, FR-018, NFR-001, NFR-002, SC-001, SC-010

Included Subtasks

• ✅ T036 Update existing compiler, integration, and e2e tests that still assume materialized library docs or legacy JSON fields
• ✅ T037 Add an end-to-end scenario for explicit local support declarations, additive conflict warnings, and first-vs-subsequent context behavior
• ✅ T038 Verify src/specify_cli/next/prompt_builder.py and src/specify_cli/cli/commands/agent/workflow.py still consume bootstrap context correctly, adding targeted regressions where needed

Implementation Notes

• This is the convergence package after the functional WPs land; avoid duplicating unit coverage already owned by earlier work packages.
• Use quickstart.md as the scenario source of truth for the end-to-end checks.

Parallel Opportunities

• T036 and T038 can proceed in parallel after WP05 stabilizes.

Dependencies

• Depends on WP01.
• Depends on WP02.
• Depends on WP03.
• Depends on WP05.
• Depends on WP06.

Risks & Mitigations

• Cross-cutting test updates are easy to under-scope. Reviewers should confirm that legacy references to success, constitution_path, files_written, and library/ have either been updated intentionally or are explicitly preserved for compatibility.

Estimated Prompt Size

~210 lines

---

Work Package WP08: Architectural Review, Glossary, and User Docs (Priority: P2)

Goal: Close the feature with a reviewer-oriented architectural pass, align glossary terminology, and update user-facing documentation for the new constitution workflow. Independent Test: The implementation is cross-checked against the spec/plan contract; glossary terms are consistent across feature artifacts and user docs; user documentation reflects the shipped-only catalog rule, explicit local support declarations, no generated library/, and the interview -> generate -> context flow. Prompt: tasks/WP08-architecture-glossary-and-user-docs.md Requirement Refs: FR-013, FR-016, FR-017, FR-018, NFR-001

Included Subtasks

• ✅ T039 Perform an architectural review against spec.md, plan.md, research.md, and the implemented runtime surfaces, using a reviewer-role mindset
• ✅ T040 Update glossary terms and canonical wording across feature artifacts and touched doctrine/CLI docs
• ✅ T041 Update user-facing documentation for the constitution workflow, including local support declarations and bootstrap depth behavior
• ✅ T042 Add or refresh documentation/regression checks so stale examples do not keep referring to library/, agents.yaml, or legacy JSON payloads

Implementation Notes

• Treat this as a reviewer-oriented polish package, not another feature-design package.
• Canonical terms to keep consistent include: local support file, action index, bootstrap context, compact context, shipped doctrine, and library_files.
• User documentation should explain the additive conflict model for local files without implying they override shipped doctrine.

Parallel Opportunities

• T040 and T041 can proceed in parallel after the architectural review notes from T039 are clear.

Dependencies

• Depends on WP05.
• Depends on WP06.
• Depends on WP07.

Risks & Mitigations

• Documentation polish often drifts from implementation reality. Ground every change in the actual CLI output and feature artifacts rather than rephrasing the spec abstractly.

Estimated Prompt Size

~180 lines

---

Work Package WP09: Extract Canonical ArtifactKind Enum and Consolidate Repetition (Priority: P2)

Goal: Create a single ArtifactKind StrEnum in src/doctrine/artifact_kinds.py that defines all doctrine artifact types once, with derived properties (plural, glob_pattern, singular). Replace the three duplicate reference-type enums, ARTIFACT_TYPES, _GLOB_PATTERNS, and _REF_TYPE_MAP with derivations from the enum. Independent Test: ArtifactKind enum has 8 members with correct plural and glob_pattern properties; old DirectiveReferenceType, ReferenceType (tactics), and ProcedureReferenceType are fully removed; ARTIFACT_TYPES now includes agent_profiles; _REF_TYPE_MAP now includes paradigm; all existing tests pass. Prompt: tasks/WP09-doctrine-artifact-kind-enum.md

Included Subtasks

• ✅ T043 Create src/doctrine/artifact_kinds.py with ArtifactKind enum
• ✅ T044 Replace the three duplicate ReferenceType enums
• ✅ T045 Replace ARTIFACT_TYPES and _GLOB_PATTERNS constants
• ✅ T046 Replace _REF_TYPE_MAP in reference_resolver
• ✅ T047 Add tests and verify no regressions

Implementation Notes

• Keep artifact_kinds.py zero-dependency within doctrine (no imports from specify_cli).
• TEMPLATE kind exists in reference enums but has no repository or shipped directory; keep it in the enum but exclude from iteration helpers like ARTIFACT_TYPES.
• Verify Pydantic model deserialization still works with ArtifactKind as a StrEnum.

Parallel Opportunities

• T043 can proceed independently; T044/T045/T046 can start once T043 is complete but are parallelizable with each other.

Dependencies

• Depends on WP05.

Risks & Mitigations

• Pydantic models use StrEnum for validation; verify that ArtifactKind serializes/deserializes identically to the old per-model enums.

Estimated Prompt Size

~140 lines

---

Work Package WP10: Migrate Remaining Consumers to ArtifactKind Enum (Priority: P2)

Goal: Align all remaining consumer-side naming with ArtifactKind.plural — rename DoctrineCatalog.profiles to agent_profiles, ensure ActionIndex, GovernanceResolution, and ResolvedReferenceGraph field names are consistent, and derive CLI help text from the enum. Independent Test: DoctrineCatalog.profiles no longer exists (fully renamed to agent_profiles); domains_present uses "agent_profiles" not "profiles"; CLI help text no longer hardcodes artifact type lists; all tests pass. Prompt: tasks/WP10-artifact-kind-consumers.md

Included Subtasks

• ✅ T048 Align DoctrineCatalog and catalog loading
• ✅ T049 Align ActionIndex, GovernanceResolution, and ResolvedReferenceGraph
• ✅ T050 Update doctrine CLI help text
• ✅ T051 Run full test suite and fix regressions

Implementation Notes

• The profiles → agent_profiles rename in DoctrineCatalog is a breaking change within this codebase — update all callers.
• Use thorough grep to catch attribute access patterns that don't show up in simple string searches.

Parallel Opportunities

• T048, T049, and T050 can proceed in parallel; T051 runs last as a validation pass.

Dependencies

• Depends on WP09.

Risks & Mitigations

• The profiles → agent_profiles rename touches multiple callers; downstream code may use attribute access patterns that don't show up in simple string searches.

Estimated Prompt Size

~100 lines

---

Work Package WP11: Create MissionRepository and Redirect Package Asset Resolution (Priority: P2)

Goal: Create a MissionRepository service in src/doctrine/missions/repository.py following the established doctrine repository pattern, redirect get_package_asset_root() to resolve from doctrine.missions instead of the stale specify_cli.missions, and update pyproject.toml packaging. Independent Test: MissionRepository lists missions and resolves command-templates; tier-5 resolution serves content from doctrine/missions/; installed distributions include doctrine/missions/ data files; all existing tests pass. Prompt: tasks/WP11-doctrine-mission-repository.md

Included Subtasks

• ✅ T052 Create MissionRepository in src/doctrine/missions/repository.py
• ✅ T053 Redirect get_package_asset_root() to doctrine.missions
• ✅ T054 Update pyproject.toml packaging
• ✅ T055 Update copy_specify_base_from_package() to use doctrine source
• ✅ T056 Add tests and verify no regressions

Implementation Notes

• Keep the repository read-only (no save/delete) — missions are shipped assets.
• The 5-tier resolver in specify_cli handles overrides; the repository just provides the shipped defaults.
• Keep specify_cli/missions as a fallback during the transition (WP12 removes it).

Parallel Opportunities

• T052 can proceed independently; T053/T054/T055 can start once T052 is complete.

Dependencies

• Depends on WP05 and WP09 (both touch doctrine/missions/).

Risks & Mitigations

• Changing get_package_asset_root() affects all tier-5 resolution. Keep a backward-compatible fallback until WP12.

Estimated Prompt Size

~160 lines

---

Work Package WP12: Remove Stale specify_cli/missions Content and Clean Up References (Priority: P2)

Goal: Remove the stale command-templates and templates from src/specify_cli/missions/ (58 stale template copies, all divergent from doctrine source), update migration code to source from doctrine/missions/, remove the fallback in get_package_asset_root(), and fix all stale path references. Independent Test: src/specify_cli/missions/ contains only Python modules (no .md/.yaml template content); migrations source from doctrine/missions/; no stale specify_cli/missions/*/command-templates/ references remain; all tests pass. Prompt: tasks/WP12-remove-stale-cli-missions.md

Included Subtasks

• □ T057 Remove command-templates and templates from specify_cli/missions/
• □ T058 Update migration code references
• □ T059 Remove specify_cli/missions fallback and update packaging
• □ T060 Fix stale path references and run full test suite

Implementation Notes

• Use git rm -r for tracked content removal.
• Keep __init__.py, glossary_hook.py, primitives.py, and mission.yaml files.
• Migration code may need dual-path fallback logic for older projects.

Parallel Opportunities

• T057 and T058 can proceed in parallel; T059 and T060 run after.

Dependencies

• Depends on WP11.

Risks & Mitigations

• Removing template content from specify_cli/missions/ could break installed distributions if WP11's packaging update is incomplete. Verify with a test build.

Estimated Prompt Size

~120 lines

---

Dependency & Execution Summary

• Recommended sequence: WP01 + WP03 + WP04 can start immediately. WP02 follows WP01. WP05 follows WP02 and WP04. WP06 follows WP04 and WP05. WP07 closes the integration sweep after the main runtime and migration work lands. WP08 is the final reviewer/documentation closeout. WP09 follows WP05 (doctrine consolidation). WP10 follows WP09 (consumer alignment). WP11 follows WP05 and WP09 (mission repository). WP12 follows WP11 (stale content removal).
• Best parallel lanes: WP01 / WP03 / WP04 in parallel first; then WP02; then WP05; then WP06 / WP09 in parallel; then WP07 / WP10 / WP11; then WP12; finish with WP08.
• MVP scope: WP01 through WP05. WP06 through WP08 are release-hardening. WP09-WP10 are doctrine consolidation (Phase 3). WP11-WP12 are mission consolidation (Phase 4).

Requirements Coverage Summary

Requirement ID	Covered By Work Package(s)
FR-001	WP01
FR-002	WP01
FR-003	WP03
FR-004	WP02
FR-005	WP03
FR-006	WP05
FR-007	WP05
FR-008	WP06
FR-009	WP06
FR-010	WP01, WP05, WP07
FR-011	WP01
FR-012	WP04
FR-013	WP05, WP08
FR-014	WP05
FR-015	WP04, WP05
FR-016	WP05, WP08
FR-017	WP02, WP07, WP08
FR-018	WP02, WP03, WP05, WP07, WP08
NFR-001	WP02, WP07, WP08
NFR-002	WP05, WP07
NFR-003	WP03

Subtask Index (Reference)

Subtask ID	Summary	Work Package	Priority	Parallel?
T001	Audit CLI interview/generate flows	WP01	P1	No
T002	Normalize interview/generate JSON contracts	WP01	P1	No
T003	Hard-fail missing interview answers	WP01	P1	No
T004	Enforce overwrite checks across generated files	WP01	P1	No
T005	Add CLI contract tests	WP01	P1	No
T006	Extend interview model for local supporting files	WP02	P1	No
T007	Validate explicit local declaration shapes	WP02	P1	Yes
T008	Build additive local references and warnings	WP02	P1	No
T009	Emit generated_files and library_files from bundle generation	WP02	P1	No
T010	Remove library materialization	WP02	P1	No
T011	Add local-support compiler/CLI tests	WP02	P1	No
T012	Audit catalog/resolver/sync validation behavior	WP03	P1	No
T013	Keep shipped-only catalog loading authoritative	WP03	P1	Yes
T014	Add template-set fallback validation	WP03	P1	Yes
T015	Preserve strict shipped validation while exempting local support docs	WP03	P1	No
T016	Remove agents.yaml assumptions from sync/status	WP03	P1	No
T017	Add catalog/resolver/sync tests	WP03	P1	No
T018	Audit extractable software-dev template prose	WP04	P1	No
T019	Create specify/plan action assets	WP04	P1	Yes
T020	Create implement/review action assets	WP04	P1	Yes
T021	Strip inline prose from source templates	WP04	P1	No
T022	Validate action-index doctrine IDs	WP04	P1	No
T023	Add source-template asset tests	WP04	P1	No
T024	Add ActionIndex loader/export	WP05	P1	Yes
T025	Rework bootstrap state and depth handling	WP05	P1	No
T026	Implement repository-driven action scoping	WP05	P1	No
T027	Include scoped local support references in context	WP05	P1	No
T028	Update context CLI/consumer JSON contract	WP05	P1	No
T029	Add context and consumer regression tests	WP05	P1	No
T030	Audit migration patterns and target prompt inventory	WP06	P2	No
T031	Implement migration detection	WP06	P2	No
T032	Rewrite generated prompts idempotently	WP06	P2	No
T033	Remove stale library artifacts during migration	WP06	P2	No
T034	Add parametrized 12-agent migration coverage	WP06	P2	Yes
T035	Add idempotency and config-filtering coverage	WP06	P2	Yes
T036	Refresh legacy integration/e2e tests	WP07	P2	Yes
T037	Add end-to-end declared-local-support scenario	WP07	P2	No
T038	Verify workflow/prompt-builder compatibility	WP07	P2	Yes
T039	Perform reviewer-oriented architectural review	WP08	P2	No
T040	Normalize glossary and canonical terms	WP08	P2	Yes
T041	Update user-facing constitution workflow docs	WP08	P2	Yes
T042	Refresh documentation/regression checks for stale examples	WP08	P2	No
T043	Create ArtifactKind enum in src/doctrine/artifact_kinds.py	WP09	P2	No
T044	Replace three duplicate ReferenceType enums	WP09	P2	Yes
T045	Replace ARTIFACT_TYPES and _GLOB_PATTERNS constants	WP09	P2	Yes
T046	Replace _REF_TYPE_MAP in reference_resolver	WP09	P2	Yes
T047	Add ArtifactKind tests and verify no regressions	WP09	P2	No
T048	Align DoctrineCatalog and catalog loading	WP10	P2	Yes
T049	Align ActionIndex, GovernanceResolution, ResolvedReferenceGraph	WP10	P2	Yes
T050	Update doctrine CLI help text	WP10	P2	Yes
T051	Run full test suite and fix regressions	WP10	P2	No

<!-- status-model:start -->

Canonical Status (Generated)

<!-- status-model:end -->

• WP01: done
• WP02: done
• WP03: done
• WP04: done
• WP05: done
• WP06: done
• WP07: done
• WP08: done
• WP09: done
• WP10: done
• WP11: done
• WP12: done