Spec Kitty

Tasks — Quality and DevEx Hardening 3.2

Mission: quality-devex-hardening-3-2-01KRJGKH Branch: fix/quality-check-updates Spec: spec.md Plan: plan.md

This document enumerates the work packages and their constituent subtasks. Sequence below preserves the dependency graph from plan.md § Phase 2. Lane computation is left to spec-kitty agent mission finalize-tasks.

Subtask Index

ID	Description	WP	Parallel
T001	Add type stubs (types-PyYAML/types-toml/types-jsonschema/types-psutil/types-requests) to dev deps	WP01
T002	Localize # type: ignore[import-untyped] on each re2 import site (research §1)	WP01	[P]
T003	Regression test for doctor.py:1092 RepairReport vs RepoAuditReport mismatch (capture real bug before narrowing types)	WP01
T004	Fix typed-code errors in status/reducer.py, sync/__init__.py, agent_retrospect.py	WP01	[P]
T005	Fix typed-code errors in auth/recovery, next/_internal_runtime/, sync/ (excluding __init__.py)	WP01	[P]
T006	Verify uv run mypy --strict src/specify_cli src/charter src/doctrine exits 0 (acknowledging doctor.py errors are owned by WP06)	WP01
T007	Record WP01 evidence (chosen scope option A, decision DM-01KRJHT7QD7XQMY33Y5TDTQ80V) and CHANGELOG entry fragment	WP01
T008	Create tests/upgrade/test_m_0_8_0_symlink_windows.py with happy OSError → shutil.copy2 fallback via monkeypatch.setattr	WP02
T009	Parametrize the dual-failure case (shutil.copy2 raises OSError) and assert errors list entry	WP02
T010	Verify test runs on POSIX CI (no Windows-only mark); record WP02 glossary fragment for "characterization test"	WP02
T011	Author characterization tests for _canonicalize_status_row using legacy rows from .kittify/migrations/mission-state/ (commit BEFORE refactor)	WP03
T012	Create src/specify_cli/migration/canonicalization.py with CanonicalRule Protocol + CanonicalStepResult + CanonicalPipelineResult + MigrationContext + apply_rules runner	WP03
T013	Lift _canonicalize_status_row rules onto Protocol — 10 named pure rules + _RULES tuple	WP03
T014	Lift analogous rules in rebuild_state.py onto the same Protocol (validates two-consumer bar)	WP03
T015	Per-rule unit tests in tests/unit/migration/test_canonicalization_rules.py (parametrized)	WP03	[P]
T016	Update architecture/2.x/04_implementation_mapping/code-patterns.md to cite migration/canonicalization.py as canonical Transformer-flavor implementation	WP03
T017	Record WP03 glossary fragment for "pipeline-shape", "rule pipeline" (with three flavors)	WP03
T018	Audit release/changelog.py regexes for catastrophic-backtracking shapes (per secure-regex-catastrophic-backtracking tactic)	WP04
T019	Apply rewrite ladder (bound quantifier / refactor nested / possessive / negated char class) per the secure-regex tactic	WP04
T020	Add wall-clock regression test in tests/regressions/test_changelog_regex_redos.py (≤100 ms for 100 000 chars adversarial input)	WP04
T021	Record Sonar rationale annotations (via the Sonar UI) for each regex hotspot remediation	WP04
T022	Record WP04 glossary fragment for "catastrophic backtracking" + Sonar evidence	WP04
T023	Charter.py orchestration tests — typer-runner integration in tests/cli/commands/test_charter_orchestration.py (research §4 Bucket A)	WP05
T024	Charter.py IO tests — tmp_path real-IO in tests/cli/commands/test_charter_io.py (Bucket B)	WP05	[P]
T025	Charter.py rendering tests — substring-stable assertions in tests/cli/commands/test_charter_rendering.py (Bucket C)	WP05	[P]
T026	Coverage tests for cli/commands/charter_bundle.py and cli/commands/agent/config.py	WP05	[P]
T027	Coverage tests for next/_internal_runtime/engine.py (hot paths only) and core/file_lock.py (uncovered branches)	WP05	[P]
T028	Record WP05 glossary fragment	WP05
T029	Characterization tests for doctor.py::mission_state covering --audit / --fix / --teamspace-dry-run modes (commit BEFORE refactor)	WP06
T030	Fix typing errors at doctor.py:631, :1092..1125 (regression test for MissionRepairResult.findings real-branch bug FIRST)	WP06
T031	Extract _validate_modes, _resolve_fail_on, _resolve_audit_root helpers (refactoring-extract-first-order-concept)	WP06
T032	Extract _run_repair, _run_teamspace_dry_run, _run_audit per-mode runners	WP06
T033	Extract shared _emit helper for the JSON-vs-pretty pattern (logical duplication across the three modes)	WP06
T034	Slim mission_state to ~30-line orchestrator; verify all characterization tests pass; record WP06 glossary fragment for "structural debt" and "deliberate linearity"	WP06
T035	Triage 127.0.0.1 loopback hotspots in auth/sync callback paths; mark safe-by-design in Sonar with rationale	WP07
T036	Triage review-lock signal-safety hotspot (either fix or document rationale in Sonar)	WP07
T037	Verify Sonar gate is OK on main via work/snippets/sonarcloud_branch_review.sh Priivacy-ai_spec-kitty main	WP07
T038	Flip .github/workflows/ci-quality.yml::sonarcloud trigger from `schedule \	\	workflow_dispatch to always()`; remove the temporary deferral comment block
T039	Record WP07 glossary fragment for "Sonar quality gate"	WP07
T040	Promote contracts/stale-lane-auto-rebase-classifier-policy.md to architecture/2.x/adr/2026-05-14-1-stale-lane-auto-rebase-classifier-policy.md with status ACCEPTED (operator approval flag)	WP08
T041	Implement src/specify_cli/merge/conflict_classifier.py with the five rules (R-PYPROJECT-DEPS-UNION, R-INIT-IMPORTS-UNION, R-URLS-LIST-UNION, R-UVLOCK-REGENERATE, R-DEFAULT-MANUAL) per data-model.md and ADR	WP08
T042	Per-rule unit tests in tests/integration/merge/test_conflict_classifier.py (parametrized happy + counter-example for each rule)	WP08	[P]
T043	Implement src/specify_cli/lanes/auto_rebase.py orchestrator (attempt git merge, classify, apply, regenerate uv.lock under core.file_lock mutex)	WP08
T044	Integration test for two-lane additive merge in tests/integration/lanes/test_auto_rebase_additive.py	WP08
T045	Negative integration test for semantic conflict (fail-safe halt with current actionable error; no partial auto-resolution leaks)	WP08
T046	Update src/specify_cli/lanes/merge.py to delegate to auto_rebase before halting on stale-detect; record WP08 glossary fragment	WP08
T047	Create src/specify_cli/core/upgrade_probe.py per data-model.md UpgradeProbeResult + UpgradeChannel + probe function (httpx-based PyPI fetch, 2 s timeout, four-channel classification)	WP09
T048	Create src/specify_cli/core/upgrade_notifier.py with maybe_emit_upgrade_notice + cache logic (24 h success / 1 h unknown TTL) + SPEC_KITTY_NO_UPGRADE_CHECK opt-out	WP09
T049	Wire notifier into CLI hot path (gated by should_check_version()); modify core/version_checker.py to extend the existing gate (do NOT introduce a parallel gate)	WP09
T050	Behavior tests in tests/core/test_upgrade_probe_and_notifier.py using requests_mock + freezegun; cover all four channels + cache hit/miss + opt-out + network failure	WP09	[P]
T051	Wall-clock test asserting ≤100 ms cache-warm budget per NFR-004	WP09	[P]
T052	Document opt-out env var in spec-kitty --help output; record WP09 glossary fragment	WP09
T053	Consolidate all WP01..WP09 glossary-fragment YAML into .kittify/glossaries/spec_kitty_core.yaml; verify every Domain Language term in spec.md has an entry	WP10
T054	Verify architecture/2.x/04_implementation_mapping/code-patterns.md cites migration/canonicalization.py (introduced by WP03)	WP10
T055	Author mission-review report mission-review.md citing every doctrine tactic applied per WP + linking the code-patterns catalog	WP10
T056	Run NFR-001 release-stability smoke (init → specify → plan → tasks → implement → review → merge → PR on a throwaway feature); record results in mission-review	WP10
T057	Update CHANGELOG.md with the mission's deliverables: cross-reference all 6 tickets (#971, #825, #595, #629, #771, #740) + mypy scope decision + auto-rebase ADR + push-time Sonar restoration	WP10

Work Packages

WP01 — Mypy strict baseline (excluding doctor.py)

Goal: Make uv run mypy --strict src/specify_cli src/charter src/doctrine exit 0 on every file the WP owns. Doctor.py errors are owned by WP06 and addressed there.

Priority: P0 (foundation; everything else prefers a clean baseline).

Independent test: uv run mypy --strict <WP01 owned files> exits 0; the regression test for MissionRepairResult.findings (added in T003) fails before T030 and passes after.

Doctrine tactics applied:

• function-over-form-testing — regression test asserts the observable behavior of the broken branch, not the type assertion.
• refactoring-guard-clauses-before-polymorphism — applied only if conditional flattening is needed to make types narrow.

Included subtasks:

• ✅ T001 Add type stubs to dev deps (WP01)
• ✅ T002 [P] Localize re2 strict-drop type-ignore comments (WP01)
• ✅ T003 Regression test for doctor.py:1092 (commit BEFORE T030 in WP06) (WP01)
• ✅ T004 [P] Fix typed-code errors in status/reducer.py, sync/__init__.py, agent_retrospect.py (WP01)
• ✅ T005 [P] Fix typed-code errors in auth/recovery, next/_internal_runtime/, sync/ (WP01)
• ✅ T006 Verify mypy strict exits 0 on WP01-owned files (WP01)
• ✅ T007 Record evidence + CHANGELOG fragment (WP01)

Dependencies: none (foundation).

Risks: T003's regression test may surface the MissionRepairResult.findings bug is more substantial than the typing drift indicates — escalate to operator if behavior fix is non-trivial.

Prompt: tasks/WP01-mypy-strict-baseline.md

---

WP02 — Windows symlink-fallback test (#629)

Goal: Add a behavior test for the m_0_8_0_worktree_agents_symlink migration's OSError → shutil.copy2 fallback path. Test runs on every POSIX CI pass via monkeypatch.

Priority: P1 (small WP; primes Windows-fallback confidence).

Independent test: uv run pytest tests/upgrade/test_m_0_8_0_symlink_windows.py -v passes both parameterized cases.

Doctrine tactics applied:

• function-over-form-testing — behavior assertion on AGENTS.md file content and migration changes / errors lists.

Included subtasks:

• ✅ T008 Create test file with happy fallback case (WP02)
• ✅ T009 Parametrize dual-failure case (WP02)
• ✅ T010 Verify POSIX CI execution + glossary fragment (WP02)

Dependencies: none (the symlink test is independent of WP01's typing fixes — it lives under tests/upgrade/ and uses monkeypatch to drive the migration directly).

Risks: minimal.

Prompt: tasks/WP02-windows-symlink-test.md

---

WP03 — Canonicalization rule-pipeline extraction

Goal: Lift _canonicalize_status_row onto the CanonicalRule Protocol (Transformer flavor). Reuse the Protocol in rebuild_state.py (two-consumer bar). Update code-patterns catalog.

Priority: P1 (motivating structural-debt refactor; concrete demonstration of the rule-pipeline tactic).

Independent test: characterization tests captured pre-refactor (T011) remain green after T013/T014; per-rule unit tests pass (T015); pipeline integration tests pass.

Doctrine tactics applied:

• chain-of-responsibility-rule-pipeline (Transformer flavor) — the central tactic for this WP.
• tdd-red-green-refactor — characterization tests precede refactor.
• refactoring-extract-first-order-concept — each rule is a first-order concept.
• function-over-form-testing — per-rule tests are value-transformer tests.

Included subtasks:

• ✅ T011 Characterization tests for _canonicalize_status_row (BEFORE refactor commit) (WP03)
• ✅ T012 Create canonicalization.py with Protocol + runner (WP03)
• ✅ T013 Lift _canonicalize_status_row rules (WP03)
• ✅ T014 Lift rebuild_state.py rules (WP03)
• ✅ T015 [P] Per-rule unit tests (WP03)
• ✅ T016 Update code-patterns catalog (WP03)
• ✅ T017 Glossary fragment ("pipeline-shape", "rule pipeline") (WP03)

Dependencies: WP01 (migration code must be type-clean before refactor).

Risks: characterization-test corpus must include a representative fixture for every rule in the pipeline; a missed rule means a green refactor that silently changes behavior. Mitigation: pre-T011 audit of .kittify/migrations/mission-state/ for fixture coverage breadth.

Prompt: tasks/WP03-canonicalization-rule-pipeline.md

---

WP04 — Sonar regex hotspots + wall-clock tests

Goal: Apply the secure-regex-catastrophic-backtracking tactic to every regex hotspot in release/changelog.py. Every fix carries a wall-clock regression test.

Priority: P1 (Sonar hotspot review required for gate-green).

Independent test: uv run pytest tests/regressions/test_changelog_regex_redos.py -v --durations=10 passes, slowest test < 100 ms.

Doctrine tactics applied:

• secure-regex-catastrophic-backtracking — every regex change.
• function-over-form-testing — completion-within-budget is the observable outcome.

Included subtasks:

• ✅ T018 Audit release/changelog.py regexes for dangerous shapes (WP04)
• ✅ T019 Apply rewrite ladder (WP04)
• ✅ T020 Wall-clock regression test (WP04)
• ✅ T021 Sonar rationale annotations (WP04)
• ✅ T022 Glossary fragment ("catastrophic backtracking") (WP04)

Dependencies: none (regex rewrites in release/changelog.py are independent of WP01's typing fixes; happy-path correctness tests are local to this WP).

Risks: a rewrite that loses match semantics is a silent behavior regression — the wall-clock test does not catch this. Mitigation: every rewrite includes a correctness test on the original happy-path corpus.

Prompt: tasks/WP04-regex-hotspots.md

---

WP05 — Sonar coverage on hot release/auth/sync paths

Goal: Author behavior-driven coverage tests for the highest-uncov files. Tests assert observable outcomes; no mocks beyond true system boundaries; characterization-first if any source-file modification is required (DEFER if it is — Pedro reports, does not fix in WP05).

Priority: P1 (drives new_coverage toward 80 %; precondition for #825 gate flip).

Independent test: uv run pytest tests/cli/commands/test_charter*.py tests/cli/commands/test_agent_config_coverage.py tests/integration/test_internal_runtime_engine.py tests/core/test_file_lock_behavior.py -v passes; Sonar reports new_coverage ≥ 80 on the touched surfaces.

Doctrine tactics applied:

• function-over-form-testing — every test.
• tdd-red-green-refactor — only if a refactor surfaces.

Included subtasks:

• ✅ T023 Charter orchestration tests (typer-runner integration) (WP05)
• ✅ T024 [P] Charter IO tests (tmp_path real I/O) (WP05)
• ✅ T025 [P] Charter rendering tests (substring-stable assertions) (WP05)
• ✅ T026 [P] Coverage tests for charter_bundle.py and agent/config.py (WP05)
• ✅ T027 [P] Coverage tests for internal_runtime/engine.py and core/file_lock.py (WP05)
• ✅ T028 Glossary fragment (WP05)

Dependencies: WP01, WP04 (regex shape locked before authoring tests that may exercise regex paths).

Risks: behavior tests on charter.py may surface real bugs (e.g. file-encoding edge cases) — Pedro reports as new issues, does not fix in this WP. Test files must use substring-stable assertions, not full-output snapshots (per research §4 Bucket C).

Prompt: tasks/WP05-coverage-hot-paths.md

---

WP06 — doctor.py::mission_state typing + multiplexer refactor

Goal: Fix doctor.py typing errors (regression test first for the MissionRepairResult.findings real-branch bug). Refactor the mission_state CLI command into per-mode runners + shared _emit helper. Characterization tests precede refactor.

Priority: P1 (closes WP01's doctor.py carve-out; demonstrates extract-first-order-concept on a debt-classified offender).

Independent test: characterization tests pass before and after refactor; uv run mypy --strict src/specify_cli/cli/commands/doctor.py exits 0; each extracted helper has cognitive complexity ≤ 10.

Doctrine tactics applied:

• tdd-red-green-refactor — characterization tests precede refactor (NFR-003).
• refactoring-extract-first-order-concept — per-mode runner + shared _emit extraction.
• function-over-form-testing — tests assert exit codes + emitted artifacts, not internal calls.

Included subtasks:

• ✅ T029 Characterization tests for the three modes (BEFORE refactor) (WP06)
• ✅ T030 Fix typing errors at lines 631, 1092–1125 (with regression test FIRST) (WP06)
• ✅ T031 Extract _validate_modes, _resolve_fail_on, _resolve_audit_root (WP06)
• ✅ T032 Extract per-mode runners (WP06)
• ✅ T033 Extract shared _emit helper (WP06)
• ✅ T034 Slim mission_state to ~30-line orchestrator; glossary fragment (WP06)

Dependencies: WP01.

Risks: the MissionRepairResult.findings bug may have a non-trivial fix that pulls Pedro out of locality — if so, fix in a focused PR before this WP closes; do not let typing-drift work expand into unrelated cleanup.

Prompt: tasks/WP06-doctor-multiplexer-refactor.md

---

WP07 — Sonar hotspot non-regex triage + push-time Sonar restoration

Goal: Resolve the remaining 5 non-regex Sonar hotspots (loopback 127.0.0.1 × 4, review-lock signal safety × 1). Once the gate is OK on main, flip the push-time Sonar trigger.

Priority: P1 (release-gate flip; final Sonar workstream).

Independent test: bash work/snippets/sonarcloud_branch_review.sh Priivacy-ai_spec-kitty main | head -10 shows status: OK; next push to main produces a Sonar run with conclusion: success.

Doctrine tactics applied:

• (CI yaml change — coordinate with infra reviewer.)
• secure-design-checklist — applied if the review-lock signal safety hotspot needs a code fix.

Included subtasks:

• ✅ T035 Triage 127.0.0.1 loopback hotspots (safe-by-design rationale) (WP07)
• ✅ T036 Triage review-lock signal-safety hotspot (WP07)
• ✅ T037 Verify Sonar gate OK on main (WP07)
• ✅ T038 Flip .github/workflows/ci-quality.yml sonarcloud trigger (WP07)
• ✅ T039 Glossary fragment ("Sonar quality gate") (WP07)

Dependencies: WP04 + WP05 + WP06 (Sonar gate must be green first).

Risks: race condition where another mission merges between WP07's gate verification and the trigger flip and re-introduces gate-failing code. Mitigation: re-run the verification immediately before the flip.

Prompt: tasks/WP07-sonar-triage-and-gate-flip.md

---

WP08 — Stale-lane auto-rebase: ADR + classifier + orchestrator

Goal: Promote the classifier-policy ADR draft to canonical status. Implement the classifier and orchestrator. Auto-rebase additive-only conflicts in spec-kitty merge; fail-safe on anything unmatched.

Priority: P1 (DevEx improvement; addresses ~30 min/mission rote-merge cost called out in #771).

Independent test: uv run pytest tests/integration/lanes/test_auto_rebase_additive.py tests/integration/merge/test_conflict_classifier.py -v passes; two-lane additive smoke (quickstart §5) completes without operator intervention; semantic-conflict negative smoke halts with the current actionable error.

Doctrine tactics applied:

• (ADR-led design — architectural decision recorded before implementation.)
• function-over-form-testing — every test.
• secure-design-checklist — for the uv lock regeneration step (operates on the lockfile boundary).

Included subtasks:

• ✅ T040 Promote ADR draft to canonical (operator approval gate) (WP08)
• ✅ T041 Implement merge/conflict_classifier.py per ADR (WP08)
• ✅ T042 [P] Per-rule unit tests (WP08)
• ✅ T043 Implement lanes/auto_rebase.py orchestrator (WP08)
• ✅ T044 Two-lane additive integration test (WP08)
• ✅ T045 Semantic-conflict negative integration test (WP08)
• ✅ T046 Update lanes/merge.py to delegate; glossary fragment (WP08)

Dependencies: WP01.

Risks: a wrongly-classified semantic conflict silently combines incompatible code — mitigated by fail-safe default (R-DEFAULT-MANUAL always last; AST/TOML validation after auto-resolution; classifier try/except defaults to Manual on any rule-evaluation exception).

Prompt: tasks/WP08-auto-rebase-classifier.md

---

WP09 — Upgrade probe + notifier (#740)

Goal: Add core/upgrade_probe.py + core/upgrade_notifier.py for the "no upgrade available" / "already current" UX. Non-blocking, cache-aware, opt-out env var.

Priority: P2 (DevEx improvement).

Independent test: uv run pytest tests/core/test_upgrade_probe_and_notifier.py -v passes; quickstart §6 cold-cache + warm-cache + opt-out + probe-failure smokes all behave as documented; wall-clock test asserts ≤ 100 ms cache-warm budget.

Doctrine tactics applied:

• secure-design-checklist — new external surface (PyPI probe).
• function-over-form-testing — every test.

Included subtasks:

• ✅ T047 Create core/upgrade_probe.py per data-model (WP09)
• ✅ T048 Create core/upgrade_notifier.py with cache + opt-out (WP09)
• ✅ T049 Wire into CLI hot path via should_check_version() (WP09)
• ✅ T050 [P] Behavior tests (4 channels + cache + opt-out + failure) (WP09)
• ✅ T051 [P] Wall-clock test (NFR-004 budget) (WP09)
• ✅ T052 Document opt-out in --help; glossary fragment (WP09)

Dependencies: WP01.

Risks: PyPI rate limits or DNS issues cause probe failures — mitigated by the 1 h UNKNOWN-channel cache TTL (prevents repeated re-probing) and the global swallow-and-return policy on the hot path.

Prompt: tasks/WP09-upgrade-probe-notifier.md

---

WP10 — Glossary consolidation + mission-review

Goal: Consolidate per-WP glossary fragments into the canonical glossary YAML. Verify code-patterns catalog is up to date. Author the mission-review report. Run the release-stability smoke. Update CHANGELOG.

Priority: P0 (closes the release-gate acceptance).

Independent test: every term in spec.md § Domain Language has a status: active entry in .kittify/glossaries/spec_kitty_core.yaml; mission-review report enumerates every doctrine tactic applied per WP; NFR-001 smoke completes without manual repair; CHANGELOG documents the mission's six-ticket closeout.

Doctrine tactics applied:

• (consolidation + audit role — no new patterns introduced)
• Reviewer-Renata profile applied for the mission-review report.
• Curator-Carla profile applied for the glossary consolidation.

Included subtasks:

• ✅ T053 Consolidate WP01..WP09 glossary fragments into spec_kitty_core.yaml (WP10)
• ✅ T054 Verify code-patterns catalog cites migration/canonicalization.py (WP10)
• ✅ T055 Author mission-review.md (WP10)
• ✅ T056 Run NFR-001 release-stability smoke (WP10)
• ✅ T057 Update CHANGELOG.md with the mission's deliverables (WP10)

Dependencies: WP01, WP02, WP03, WP04, WP05, WP06, WP07, WP08, WP09.

Risks: NFR-001 smoke may surface a real regression in the post-merge main — Pedro reports immediately and triages with the operator; the mission cannot be marked release-ready otherwise.

Prompt: tasks/WP10-glossary-mission-review.md

---

Lane Computation Hints (for finalize-tasks)

• WP01, WP02, WP04 are independent — no shared owned_files.
• WP03, WP05, WP06, WP08, WP09 depend on WP01.
• WP05 is independent of WP04 textually (different files) but Pedro sequences WP05 after WP04 to lock the regex contract before authoring regex-adjacent tests.
• WP07 depends on WP04, WP05, WP06 (Sonar gate must be green before flip).
• WP10 depends on WP01..WP09.

Suggested lane shape:

Lane	WPs (in order)
Lane A	WP01 → WP03 → WP06
Lane B	WP02
Lane C	WP04 → WP05 → WP07
Lane D	WP08
Lane E	WP09
Lane F	WP10 (after all merge)

finalize-tasks reads dependencies from each WP's frontmatter and computes the actual lanes — the table above is an advisory seed.

Scope policy

The full ten-WP scope is the mission's deliverable. No WP is deferred. The mission's binding philosophy (spec.md § Mission Philosophy item 3 — "Quality > speed") permits the work to span multiple sessions and to take as long as it takes; it does not permit cutting WPs from scope. Deferrals are reserved for items already named as out-of-scope in spec.md § Out of Scope.

If a specific WP becomes blocked (e.g. WP08 awaiting operator ADR approval), the implement-review loop pauses that WP and continues unblocked WPs. The blocked WP resumes when the blocker resolves; the mission does not "ship without" the WP.

WP-level priority for sequencing (per plan.md § Phase 2 dependency graph):

1. WP01 — mypy strict baseline (foundation). 2. WP02 — independent; small; can run in parallel with WP01. 3. WP04 — independent; regex hotspots; can run in parallel with WP01. 4. WP03, WP05, WP06, WP08, WP09 — depend on WP01; run after it lands. 5. WP07 — depends on WP04 + WP05 + WP06 (Sonar gate green precondition). 6. WP10 — depends on WP01..WP09; consolidates and runs the post-merge mission-review.