bigpowers 2.90.2 → 2.91.1

package/.agents/skills/build-epic/SKILL.md

@@ -66,7 +66,7 @@ Orchestrates the **build** flow for a single epic: survey → plan tasks → kic
 After step 5 (verify-work) completes successfully, step 6 runs `audit-code` automatically in `--gate` mode:
 
 1. **Run audit:** Invoke `audit-code --gate` on the complete diff for this story.
-2. **Pass (exit 0):** All checklist sections pass → advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
+2. **Pass (exit 0):** All checklist sections pass → run `bash scripts/lib/assert-epic-gate.sh advance` to mechanically confirm the audit pass, then advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
 3. **Fail (exit 1):** One or more checklist sections fail → **reset `epic_cycle.current_step` to `4`** (develop-tdd) and add the failing section IDs to `completed_steps` as `"1,2,3,4,5,6(fail: ...)"`. Record `epic_cycle.audit_result: fail` in `state.yaml`. Do NOT advance past step 6 until audit passes.
 4. **Audit artifact:** Full audit report saved to `specs/verifications/AUDIT-<epic>-<story>.md` regardless of pass/fail, for reviewer traceability.
 5. **Enforce F.I.R.S.T:** After audit-code passes, run `enforce-first --quick` on new/modified tests. Append F.I.R.S.T violations (if any) to the audit report. Failing F.I.R.S.T criteria trigger the same loop-back to step 4.

package/.agents/skills/commit-message/SKILL.md

@@ -36,6 +36,7 @@ description: "Reviews working-tree changes, then drafts a Conventional Commits t
 
 ## Checklist before finalizing
 
+- [ ] **Audit pass confirmed (HARD GATE — Issue #182):** In active epic flow, `bash scripts/lib/assert-epic-gate.sh commit` passes.
 - [ ] Type matches the **dominant** user-visible outcome (`feat` vs `fix` vs `perf`, etc.).
 - [ ] **Scope** is a short noun in parentheses if it helps (e.g. `fix(api): …`).
 - [ ] Breaking changes are explicit (`!` and/or `BREAKING CHANGE:` in the body/footer).

package/.agents/skills/guard-git/SKILL.md

@@ -12,7 +12,8 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## What gets blocked/enforced
 
-- **Safety**: `git push --force`, `git reset --hard`, `git clean -f`, `git branch -D`, `git checkout .`, `git restore .`.
+- **Safety**: force push (`-f`, `--force`, `+refspec`), `reset` with `--hard`, `clean` with any `-f*`, `branch -D` / `--delete --force`, `checkout`/`restore` that discard work (`-f`, `.` pathspec).
+- **Argv, not substrings**: the hook tokenizes each command, strips git's global options (`-C <path>`, `-c k=v`, `--no-pager`, `--git-dir=…`) and judges the `(subcommand, arguments)` pair — so `git -C repo reset -q --hard` is blocked, while `git push --force-with-lease` is allowed. Rules live in `scripts/lib/git-argv.py`, mirrored in `extensions/omp-hooks.ts`; `scripts/test-guard-git-normalize.sh` fails if the two disagree.
 - **Discipline**: Blocks direct commits or pushes to protected branches (`main`, `master`) unless `GIT_BIGPOWERS_LAND=1` (set only by `scripts/land-branch.sh`).
 - **Allows**: `git push origin <feature-branch>` for backup/CI; solo land push to `main` only inside `land-branch.sh`.
 - **Standardization**: Enforces [Conventional Commits](https://www.conventionalcommits.org/) for all `git commit` commands.
@@ -37,7 +38,11 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## Customization
 
-To add or remove patterns or protected branches, edit `pre-tool-use.sh`.
+The rule table lives in `scripts/lib/git-argv.py` (shell hooks) and is mirrored in `extensions/omp-hooks.ts`. Add a rule to both and a row to `specs/verifications/fixtures/guard-git-normalize.yaml`.
+
+The TS side is judged through `evaluateBashCall()` — the single pure function the `tool_call` handler calls — so the fixture covers **every** rule (destructive argv, protected-branch push, Conventional Commits), not just the argv table. The gate (`bash scripts/test-guard-git-normalize.sh`) runs both implementations with `GIT_BIGPOWERS_LAND` unset and fails if they disagree, so a verdict can never depend on the developer's shell.
+
+Protected-branch rules answer differently on `main` than on a feature branch, so a row that touches commit/push semantics states its `branch:` and both sides are run in it (the shell hook in a throwaway repo checked out on that branch). A row without `branch:` is judged on a feature branch. Do not pin the branch inside either implementation — that is what made the gate pass on PRs and fail on `main`.
 
 ## Advanced
 

package/.agents/skills/release-branch/SKILL.md

@@ -56,13 +56,12 @@ git log main...HEAD --format="%B" | grep -qiE 'co[- ]authored[- ]by' && echo "
 ### 2a. Security gate
 
 - [ ] `specs/security/REVIEW.md` exists and is fresh (matches current branch diff)
-- [ ] No unresolved HIGH findings with confidence ≥ 8 (or all documented in `specs/security/EXCEPTIONS.md` with sign-off rationale)
 
-If REVIEW.md is missing or stale → run `security-review` inline. Findings block the merge unless documented in EXCEPTIONS.md.
+### 2b. Audit-code & Traceability gates (HARD GATES — Issue #182)
 
-### 2b. Traceability gate
-
-Run `gate-trace` before merge. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml` (`traceability_override: CONCERNS accepted, reason: <explanation>`). WAIVED if no matrix available.
+- [ ] Run `bash scripts/lib/assert-epic-gate.sh release` (asserts `specs/AUDIT_LATEST.md` passed; blocks release on failure).
+- [ ] No unresolved HIGH security findings with confidence ≥ 8 (or documented in `specs/security/EXCEPTIONS.md`).
+- [ ] Run `gate-trace`. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml`. WAIVED if no matrix available.
 
 > **Adversarial refute framing (e45s32):** The final pre-merge check is **refute, not rubber-stamp**. Before declaring ready, actively try to disprove traceability completeness — missing story tags, absent verify evidence, stale security review. Only proceed when refutation fails.
 

package/.agents/skills/run-planning/SKILL.md

@@ -62,7 +62,12 @@ if written:
    - `plan-release` — sequence epics by WSJF
    - `slice-tasks` — cut vertical slices
 
-4. **Update status** — On successful completion, set `status: done` for that workflow key in `planning-status.yaml`.
+4. **Mechanical artifact assertion before status update (HARD GATE — Issue #183):**
+   Before marking any workflow key `status: done`, run:
+   ```bash
+   bash scripts/lib/assert-planning-artifacts.sh <workflow-key>
+   ```
+   If the artifact assertion fails, do NOT advance status. Remediate missing output files before continuing. On success, set `status: done` for that workflow key in `planning-status.yaml`.
 
 5. **Advance** — Set `state.yaml` `active_flow: planning` while in this chain. When all required keys are done, set `handoff.next_skill` to `plan-work`.
 

package/.agents/skills/verify-work/SKILL.md

@@ -62,7 +62,8 @@ Review answers "is the code good?"; Verify answers "does the built thing do what
 5a. **Blind-spot check** — run `bash scripts/check-blind-spots.sh`. This detects structural quality gaps (verify-gap, test-gap, stale-tag, etc.) beyond percentage coverage. If any HIGH-severity findings exist → **block the verify-work PASS gate**. Findings are written to `specs/blind-spots.json`. MEDIUM/LOW findings warn but don't block.
 5a2. **Completeness critic (e45s05)** — `bash scripts/lib/completeness-critic.sh`. **BLOCKER** aborts merge gate; WARNING → gaps loop; FILLED = evidence only.
 
-5b. **NFR Evidence Gate** (P0 only) — Produces go/no-go output on three dimensions: Performance (response time, throughput), Reliability (error rate, recovery), and Operability (logging, health checks). Reads thresholds from `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md` and writes evidence as OKF verification-report bundles to `specs/verifications/NFR-eNNsYY.json`. FAIL on any dimension blocks the gate.
+5b. **NFR Evidence Gate** (P0 only) — Go/no-go on Performance, Reliability, Operability against `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md`. Evidence written to `specs/verifications/NFR-eNNsYY.json`. FAIL blocks gate.
+5c. **Adversarial Defect Mutation Gate (HARD GATE — e86 / ab-20)** — `bash scripts/verify-adversarial.sh`. Boundary fuzzing & parameter injection against modified interfaces. Crashes or failures block PASS gate; evidence in `specs/verifications/ADVERSARIAL_LATEST.yaml`.
 6. **Step-by-step UAT** (skip if P2/P3) — one user-observable action at a time.
 7. **Gaps loop** — failures → log → `plan-work` → re-verify. Unaddressed HIGH findings from step 5 feed into this loop alongside other quality gaps.
 

package/.cline/skills/build-epic/SKILL.md

@@ -66,7 +66,7 @@ Orchestrates the **build** flow for a single epic: survey → plan tasks → kic
 After step 5 (verify-work) completes successfully, step 6 runs `audit-code` automatically in `--gate` mode:
 
 1. **Run audit:** Invoke `audit-code --gate` on the complete diff for this story.
-2. **Pass (exit 0):** All checklist sections pass → advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
+2. **Pass (exit 0):** All checklist sections pass → run `bash scripts/lib/assert-epic-gate.sh advance` to mechanically confirm the audit pass, then advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
 3. **Fail (exit 1):** One or more checklist sections fail → **reset `epic_cycle.current_step` to `4`** (develop-tdd) and add the failing section IDs to `completed_steps` as `"1,2,3,4,5,6(fail: ...)"`. Record `epic_cycle.audit_result: fail` in `state.yaml`. Do NOT advance past step 6 until audit passes.
 4. **Audit artifact:** Full audit report saved to `specs/verifications/AUDIT-<epic>-<story>.md` regardless of pass/fail, for reviewer traceability.
 5. **Enforce F.I.R.S.T:** After audit-code passes, run `enforce-first --quick` on new/modified tests. Append F.I.R.S.T violations (if any) to the audit report. Failing F.I.R.S.T criteria trigger the same loop-back to step 4.

package/.cline/skills/commit-message/SKILL.md

@@ -36,6 +36,7 @@ description: "Reviews working-tree changes, then drafts a Conventional Commits t
 
 ## Checklist before finalizing
 
+- [ ] **Audit pass confirmed (HARD GATE — Issue #182):** In active epic flow, `bash scripts/lib/assert-epic-gate.sh commit` passes.
 - [ ] Type matches the **dominant** user-visible outcome (`feat` vs `fix` vs `perf`, etc.).
 - [ ] **Scope** is a short noun in parentheses if it helps (e.g. `fix(api): …`).
 - [ ] Breaking changes are explicit (`!` and/or `BREAKING CHANGE:` in the body/footer).

package/.cline/skills/guard-git/SKILL.md

@@ -12,7 +12,8 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## What gets blocked/enforced
 
-- **Safety**: `git push --force`, `git reset --hard`, `git clean -f`, `git branch -D`, `git checkout .`, `git restore .`.
+- **Safety**: force push (`-f`, `--force`, `+refspec`), `reset` with `--hard`, `clean` with any `-f*`, `branch -D` / `--delete --force`, `checkout`/`restore` that discard work (`-f`, `.` pathspec).
+- **Argv, not substrings**: the hook tokenizes each command, strips git's global options (`-C <path>`, `-c k=v`, `--no-pager`, `--git-dir=…`) and judges the `(subcommand, arguments)` pair — so `git -C repo reset -q --hard` is blocked, while `git push --force-with-lease` is allowed. Rules live in `scripts/lib/git-argv.py`, mirrored in `extensions/omp-hooks.ts`; `scripts/test-guard-git-normalize.sh` fails if the two disagree.
 - **Discipline**: Blocks direct commits or pushes to protected branches (`main`, `master`) unless `GIT_BIGPOWERS_LAND=1` (set only by `scripts/land-branch.sh`).
 - **Allows**: `git push origin <feature-branch>` for backup/CI; solo land push to `main` only inside `land-branch.sh`.
 - **Standardization**: Enforces [Conventional Commits](https://www.conventionalcommits.org/) for all `git commit` commands.
@@ -37,7 +38,11 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## Customization
 
-To add or remove patterns or protected branches, edit `pre-tool-use.sh`.
+The rule table lives in `scripts/lib/git-argv.py` (shell hooks) and is mirrored in `extensions/omp-hooks.ts`. Add a rule to both and a row to `specs/verifications/fixtures/guard-git-normalize.yaml`.
+
+The TS side is judged through `evaluateBashCall()` — the single pure function the `tool_call` handler calls — so the fixture covers **every** rule (destructive argv, protected-branch push, Conventional Commits), not just the argv table. The gate (`bash scripts/test-guard-git-normalize.sh`) runs both implementations with `GIT_BIGPOWERS_LAND` unset and fails if they disagree, so a verdict can never depend on the developer's shell.
+
+Protected-branch rules answer differently on `main` than on a feature branch, so a row that touches commit/push semantics states its `branch:` and both sides are run in it (the shell hook in a throwaway repo checked out on that branch). A row without `branch:` is judged on a feature branch. Do not pin the branch inside either implementation — that is what made the gate pass on PRs and fail on `main`.
 
 ## Advanced
 

package/.cline/skills/release-branch/SKILL.md

@@ -56,13 +56,12 @@ git log main...HEAD --format="%B" | grep -qiE 'co[- ]authored[- ]by' && echo "
 ### 2a. Security gate
 
 - [ ] `specs/security/REVIEW.md` exists and is fresh (matches current branch diff)
-- [ ] No unresolved HIGH findings with confidence ≥ 8 (or all documented in `specs/security/EXCEPTIONS.md` with sign-off rationale)
 
-If REVIEW.md is missing or stale → run `security-review` inline. Findings block the merge unless documented in EXCEPTIONS.md.
+### 2b. Audit-code & Traceability gates (HARD GATES — Issue #182)
 
-### 2b. Traceability gate
-
-Run `gate-trace` before merge. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml` (`traceability_override: CONCERNS accepted, reason: <explanation>`). WAIVED if no matrix available.
+- [ ] Run `bash scripts/lib/assert-epic-gate.sh release` (asserts `specs/AUDIT_LATEST.md` passed; blocks release on failure).
+- [ ] No unresolved HIGH security findings with confidence ≥ 8 (or documented in `specs/security/EXCEPTIONS.md`).
+- [ ] Run `gate-trace`. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml`. WAIVED if no matrix available.
 
 > **Adversarial refute framing (e45s32):** The final pre-merge check is **refute, not rubber-stamp**. Before declaring ready, actively try to disprove traceability completeness — missing story tags, absent verify evidence, stale security review. Only proceed when refutation fails.
 

package/.cline/skills/run-planning/SKILL.md

@@ -62,7 +62,12 @@ if written:
    - `plan-release` — sequence epics by WSJF
    - `slice-tasks` — cut vertical slices
 
-4. **Update status** — On successful completion, set `status: done` for that workflow key in `planning-status.yaml`.
+4. **Mechanical artifact assertion before status update (HARD GATE — Issue #183):**
+   Before marking any workflow key `status: done`, run:
+   ```bash
+   bash scripts/lib/assert-planning-artifacts.sh <workflow-key>
+   ```
+   If the artifact assertion fails, do NOT advance status. Remediate missing output files before continuing. On success, set `status: done` for that workflow key in `planning-status.yaml`.
 
 5. **Advance** — Set `state.yaml` `active_flow: planning` while in this chain. When all required keys are done, set `handoff.next_skill` to `plan-work`.
 

package/.cline/skills/verify-work/SKILL.md

@@ -62,7 +62,8 @@ Review answers "is the code good?"; Verify answers "does the built thing do what
 5a. **Blind-spot check** — run `bash scripts/check-blind-spots.sh`. This detects structural quality gaps (verify-gap, test-gap, stale-tag, etc.) beyond percentage coverage. If any HIGH-severity findings exist → **block the verify-work PASS gate**. Findings are written to `specs/blind-spots.json`. MEDIUM/LOW findings warn but don't block.
 5a2. **Completeness critic (e45s05)** — `bash scripts/lib/completeness-critic.sh`. **BLOCKER** aborts merge gate; WARNING → gaps loop; FILLED = evidence only.
 
-5b. **NFR Evidence Gate** (P0 only) — Produces go/no-go output on three dimensions: Performance (response time, throughput), Reliability (error rate, recovery), and Operability (logging, health checks). Reads thresholds from `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md` and writes evidence as OKF verification-report bundles to `specs/verifications/NFR-eNNsYY.json`. FAIL on any dimension blocks the gate.
+5b. **NFR Evidence Gate** (P0 only) — Go/no-go on Performance, Reliability, Operability against `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md`. Evidence written to `specs/verifications/NFR-eNNsYY.json`. FAIL blocks gate.
+5c. **Adversarial Defect Mutation Gate (HARD GATE — e86 / ab-20)** — `bash scripts/verify-adversarial.sh`. Boundary fuzzing & parameter injection against modified interfaces. Crashes or failures block PASS gate; evidence in `specs/verifications/ADVERSARIAL_LATEST.yaml`.
 6. **Step-by-step UAT** (skip if P2/P3) — one user-observable action at a time.
 7. **Gaps loop** — failures → log → `plan-work` → re-verify. Unaddressed HIGH findings from step 5 feed into this loop alongside other quality gaps.
 

package/.codebuddy/skills/build-epic/SKILL.md

@@ -66,7 +66,7 @@ Orchestrates the **build** flow for a single epic: survey → plan tasks → kic
 After step 5 (verify-work) completes successfully, step 6 runs `audit-code` automatically in `--gate` mode:
 
 1. **Run audit:** Invoke `audit-code --gate` on the complete diff for this story.
-2. **Pass (exit 0):** All checklist sections pass → advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
+2. **Pass (exit 0):** All checklist sections pass → run `bash scripts/lib/assert-epic-gate.sh advance` to mechanically confirm the audit pass, then advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
 3. **Fail (exit 1):** One or more checklist sections fail → **reset `epic_cycle.current_step` to `4`** (develop-tdd) and add the failing section IDs to `completed_steps` as `"1,2,3,4,5,6(fail: ...)"`. Record `epic_cycle.audit_result: fail` in `state.yaml`. Do NOT advance past step 6 until audit passes.
 4. **Audit artifact:** Full audit report saved to `specs/verifications/AUDIT-<epic>-<story>.md` regardless of pass/fail, for reviewer traceability.
 5. **Enforce F.I.R.S.T:** After audit-code passes, run `enforce-first --quick` on new/modified tests. Append F.I.R.S.T violations (if any) to the audit report. Failing F.I.R.S.T criteria trigger the same loop-back to step 4.

package/.codebuddy/skills/commit-message/SKILL.md

@@ -36,6 +36,7 @@ description: "Reviews working-tree changes, then drafts a Conventional Commits t
 
 ## Checklist before finalizing
 
+- [ ] **Audit pass confirmed (HARD GATE — Issue #182):** In active epic flow, `bash scripts/lib/assert-epic-gate.sh commit` passes.
 - [ ] Type matches the **dominant** user-visible outcome (`feat` vs `fix` vs `perf`, etc.).
 - [ ] **Scope** is a short noun in parentheses if it helps (e.g. `fix(api): …`).
 - [ ] Breaking changes are explicit (`!` and/or `BREAKING CHANGE:` in the body/footer).

package/.codebuddy/skills/guard-git/SKILL.md

@@ -12,7 +12,8 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## What gets blocked/enforced
 
-- **Safety**: `git push --force`, `git reset --hard`, `git clean -f`, `git branch -D`, `git checkout .`, `git restore .`.
+- **Safety**: force push (`-f`, `--force`, `+refspec`), `reset` with `--hard`, `clean` with any `-f*`, `branch -D` / `--delete --force`, `checkout`/`restore` that discard work (`-f`, `.` pathspec).
+- **Argv, not substrings**: the hook tokenizes each command, strips git's global options (`-C <path>`, `-c k=v`, `--no-pager`, `--git-dir=…`) and judges the `(subcommand, arguments)` pair — so `git -C repo reset -q --hard` is blocked, while `git push --force-with-lease` is allowed. Rules live in `scripts/lib/git-argv.py`, mirrored in `extensions/omp-hooks.ts`; `scripts/test-guard-git-normalize.sh` fails if the two disagree.
 - **Discipline**: Blocks direct commits or pushes to protected branches (`main`, `master`) unless `GIT_BIGPOWERS_LAND=1` (set only by `scripts/land-branch.sh`).
 - **Allows**: `git push origin <feature-branch>` for backup/CI; solo land push to `main` only inside `land-branch.sh`.
 - **Standardization**: Enforces [Conventional Commits](https://www.conventionalcommits.org/) for all `git commit` commands.
@@ -37,7 +38,11 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## Customization
 
-To add or remove patterns or protected branches, edit `pre-tool-use.sh`.
+The rule table lives in `scripts/lib/git-argv.py` (shell hooks) and is mirrored in `extensions/omp-hooks.ts`. Add a rule to both and a row to `specs/verifications/fixtures/guard-git-normalize.yaml`.
+
+The TS side is judged through `evaluateBashCall()` — the single pure function the `tool_call` handler calls — so the fixture covers **every** rule (destructive argv, protected-branch push, Conventional Commits), not just the argv table. The gate (`bash scripts/test-guard-git-normalize.sh`) runs both implementations with `GIT_BIGPOWERS_LAND` unset and fails if they disagree, so a verdict can never depend on the developer's shell.
+
+Protected-branch rules answer differently on `main` than on a feature branch, so a row that touches commit/push semantics states its `branch:` and both sides are run in it (the shell hook in a throwaway repo checked out on that branch). A row without `branch:` is judged on a feature branch. Do not pin the branch inside either implementation — that is what made the gate pass on PRs and fail on `main`.
 
 ## Advanced
 

package/.codebuddy/skills/release-branch/SKILL.md

@@ -56,13 +56,12 @@ git log main...HEAD --format="%B" | grep -qiE 'co[- ]authored[- ]by' && echo "
 ### 2a. Security gate
 
 - [ ] `specs/security/REVIEW.md` exists and is fresh (matches current branch diff)
-- [ ] No unresolved HIGH findings with confidence ≥ 8 (or all documented in `specs/security/EXCEPTIONS.md` with sign-off rationale)
 
-If REVIEW.md is missing or stale → run `security-review` inline. Findings block the merge unless documented in EXCEPTIONS.md.
+### 2b. Audit-code & Traceability gates (HARD GATES — Issue #182)
 
-### 2b. Traceability gate
-
-Run `gate-trace` before merge. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml` (`traceability_override: CONCERNS accepted, reason: <explanation>`). WAIVED if no matrix available.
+- [ ] Run `bash scripts/lib/assert-epic-gate.sh release` (asserts `specs/AUDIT_LATEST.md` passed; blocks release on failure).
+- [ ] No unresolved HIGH security findings with confidence ≥ 8 (or documented in `specs/security/EXCEPTIONS.md`).
+- [ ] Run `gate-trace`. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml`. WAIVED if no matrix available.
 
 > **Adversarial refute framing (e45s32):** The final pre-merge check is **refute, not rubber-stamp**. Before declaring ready, actively try to disprove traceability completeness — missing story tags, absent verify evidence, stale security review. Only proceed when refutation fails.
 

package/.codebuddy/skills/run-planning/SKILL.md

@@ -62,7 +62,12 @@ if written:
    - `plan-release` — sequence epics by WSJF
    - `slice-tasks` — cut vertical slices
 
-4. **Update status** — On successful completion, set `status: done` for that workflow key in `planning-status.yaml`.
+4. **Mechanical artifact assertion before status update (HARD GATE — Issue #183):**
+   Before marking any workflow key `status: done`, run:
+   ```bash
+   bash scripts/lib/assert-planning-artifacts.sh <workflow-key>
+   ```
+   If the artifact assertion fails, do NOT advance status. Remediate missing output files before continuing. On success, set `status: done` for that workflow key in `planning-status.yaml`.
 
 5. **Advance** — Set `state.yaml` `active_flow: planning` while in this chain. When all required keys are done, set `handoff.next_skill` to `plan-work`.
 

package/.codebuddy/skills/verify-work/SKILL.md

@@ -62,7 +62,8 @@ Review answers "is the code good?"; Verify answers "does the built thing do what
 5a. **Blind-spot check** — run `bash scripts/check-blind-spots.sh`. This detects structural quality gaps (verify-gap, test-gap, stale-tag, etc.) beyond percentage coverage. If any HIGH-severity findings exist → **block the verify-work PASS gate**. Findings are written to `specs/blind-spots.json`. MEDIUM/LOW findings warn but don't block.
 5a2. **Completeness critic (e45s05)** — `bash scripts/lib/completeness-critic.sh`. **BLOCKER** aborts merge gate; WARNING → gaps loop; FILLED = evidence only.
 
-5b. **NFR Evidence Gate** (P0 only) — Produces go/no-go output on three dimensions: Performance (response time, throughput), Reliability (error rate, recovery), and Operability (logging, health checks). Reads thresholds from `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md` and writes evidence as OKF verification-report bundles to `specs/verifications/NFR-eNNsYY.json`. FAIL on any dimension blocks the gate.
+5b. **NFR Evidence Gate** (P0 only) — Go/no-go on Performance, Reliability, Operability against `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md`. Evidence written to `specs/verifications/NFR-eNNsYY.json`. FAIL blocks gate.
+5c. **Adversarial Defect Mutation Gate (HARD GATE — e86 / ab-20)** — `bash scripts/verify-adversarial.sh`. Boundary fuzzing & parameter injection against modified interfaces. Crashes or failures block PASS gate; evidence in `specs/verifications/ADVERSARIAL_LATEST.yaml`.
 6. **Step-by-step UAT** (skip if P2/P3) — one user-observable action at a time.
 7. **Gaps loop** — failures → log → `plan-work` → re-verify. Unaddressed HIGH findings from step 5 feed into this loop alongside other quality gaps.
 

package/.codex/skills/build-epic/SKILL.md

@@ -66,7 +66,7 @@ Orchestrates the **build** flow for a single epic: survey → plan tasks → kic
 After step 5 (verify-work) completes successfully, step 6 runs `audit-code` automatically in `--gate` mode:
 
 1. **Run audit:** Invoke `audit-code --gate` on the complete diff for this story.
-2. **Pass (exit 0):** All checklist sections pass → advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
+2. **Pass (exit 0):** All checklist sections pass → run `bash scripts/lib/assert-epic-gate.sh advance` to mechanically confirm the audit pass, then advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
 3. **Fail (exit 1):** One or more checklist sections fail → **reset `epic_cycle.current_step` to `4`** (develop-tdd) and add the failing section IDs to `completed_steps` as `"1,2,3,4,5,6(fail: ...)"`. Record `epic_cycle.audit_result: fail` in `state.yaml`. Do NOT advance past step 6 until audit passes.
 4. **Audit artifact:** Full audit report saved to `specs/verifications/AUDIT-<epic>-<story>.md` regardless of pass/fail, for reviewer traceability.
 5. **Enforce F.I.R.S.T:** After audit-code passes, run `enforce-first --quick` on new/modified tests. Append F.I.R.S.T violations (if any) to the audit report. Failing F.I.R.S.T criteria trigger the same loop-back to step 4.

package/.codex/skills/commit-message/SKILL.md

@@ -36,6 +36,7 @@ description: "Reviews working-tree changes, then drafts a Conventional Commits t
 
 ## Checklist before finalizing
 
+- [ ] **Audit pass confirmed (HARD GATE — Issue #182):** In active epic flow, `bash scripts/lib/assert-epic-gate.sh commit` passes.
 - [ ] Type matches the **dominant** user-visible outcome (`feat` vs `fix` vs `perf`, etc.).
 - [ ] **Scope** is a short noun in parentheses if it helps (e.g. `fix(api): …`).
 - [ ] Breaking changes are explicit (`!` and/or `BREAKING CHANGE:` in the body/footer).

package/.codex/skills/guard-git/SKILL.md

@@ -12,7 +12,8 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## What gets blocked/enforced
 
-- **Safety**: `git push --force`, `git reset --hard`, `git clean -f`, `git branch -D`, `git checkout .`, `git restore .`.
+- **Safety**: force push (`-f`, `--force`, `+refspec`), `reset` with `--hard`, `clean` with any `-f*`, `branch -D` / `--delete --force`, `checkout`/`restore` that discard work (`-f`, `.` pathspec).
+- **Argv, not substrings**: the hook tokenizes each command, strips git's global options (`-C <path>`, `-c k=v`, `--no-pager`, `--git-dir=…`) and judges the `(subcommand, arguments)` pair — so `git -C repo reset -q --hard` is blocked, while `git push --force-with-lease` is allowed. Rules live in `scripts/lib/git-argv.py`, mirrored in `extensions/omp-hooks.ts`; `scripts/test-guard-git-normalize.sh` fails if the two disagree.
 - **Discipline**: Blocks direct commits or pushes to protected branches (`main`, `master`) unless `GIT_BIGPOWERS_LAND=1` (set only by `scripts/land-branch.sh`).
 - **Allows**: `git push origin <feature-branch>` for backup/CI; solo land push to `main` only inside `land-branch.sh`.
 - **Standardization**: Enforces [Conventional Commits](https://www.conventionalcommits.org/) for all `git commit` commands.
@@ -37,7 +38,11 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## Customization
 
-To add or remove patterns or protected branches, edit `pre-tool-use.sh`.
+The rule table lives in `scripts/lib/git-argv.py` (shell hooks) and is mirrored in `extensions/omp-hooks.ts`. Add a rule to both and a row to `specs/verifications/fixtures/guard-git-normalize.yaml`.
+
+The TS side is judged through `evaluateBashCall()` — the single pure function the `tool_call` handler calls — so the fixture covers **every** rule (destructive argv, protected-branch push, Conventional Commits), not just the argv table. The gate (`bash scripts/test-guard-git-normalize.sh`) runs both implementations with `GIT_BIGPOWERS_LAND` unset and fails if they disagree, so a verdict can never depend on the developer's shell.
+
+Protected-branch rules answer differently on `main` than on a feature branch, so a row that touches commit/push semantics states its `branch:` and both sides are run in it (the shell hook in a throwaway repo checked out on that branch). A row without `branch:` is judged on a feature branch. Do not pin the branch inside either implementation — that is what made the gate pass on PRs and fail on `main`.
 
 ## Advanced
 

package/.codex/skills/release-branch/SKILL.md

@@ -56,13 +56,12 @@ git log main...HEAD --format="%B" | grep -qiE 'co[- ]authored[- ]by' && echo "
 ### 2a. Security gate
 
 - [ ] `specs/security/REVIEW.md` exists and is fresh (matches current branch diff)
-- [ ] No unresolved HIGH findings with confidence ≥ 8 (or all documented in `specs/security/EXCEPTIONS.md` with sign-off rationale)
 
-If REVIEW.md is missing or stale → run `security-review` inline. Findings block the merge unless documented in EXCEPTIONS.md.
+### 2b. Audit-code & Traceability gates (HARD GATES — Issue #182)
 
-### 2b. Traceability gate
-
-Run `gate-trace` before merge. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml` (`traceability_override: CONCERNS accepted, reason: <explanation>`). WAIVED if no matrix available.
+- [ ] Run `bash scripts/lib/assert-epic-gate.sh release` (asserts `specs/AUDIT_LATEST.md` passed; blocks release on failure).
+- [ ] No unresolved HIGH security findings with confidence ≥ 8 (or documented in `specs/security/EXCEPTIONS.md`).
+- [ ] Run `gate-trace`. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml`. WAIVED if no matrix available.
 
 > **Adversarial refute framing (e45s32):** The final pre-merge check is **refute, not rubber-stamp**. Before declaring ready, actively try to disprove traceability completeness — missing story tags, absent verify evidence, stale security review. Only proceed when refutation fails.
 

package/.codex/skills/run-planning/SKILL.md

@@ -62,7 +62,12 @@ if written:
    - `plan-release` — sequence epics by WSJF
    - `slice-tasks` — cut vertical slices
 
-4. **Update status** — On successful completion, set `status: done` for that workflow key in `planning-status.yaml`.
+4. **Mechanical artifact assertion before status update (HARD GATE — Issue #183):**
+   Before marking any workflow key `status: done`, run:
+   ```bash
+   bash scripts/lib/assert-planning-artifacts.sh <workflow-key>
+   ```
+   If the artifact assertion fails, do NOT advance status. Remediate missing output files before continuing. On success, set `status: done` for that workflow key in `planning-status.yaml`.
 
 5. **Advance** — Set `state.yaml` `active_flow: planning` while in this chain. When all required keys are done, set `handoff.next_skill` to `plan-work`.
 

package/.codex/skills/verify-work/SKILL.md

@@ -62,7 +62,8 @@ Review answers "is the code good?"; Verify answers "does the built thing do what
 5a. **Blind-spot check** — run `bash scripts/check-blind-spots.sh`. This detects structural quality gaps (verify-gap, test-gap, stale-tag, etc.) beyond percentage coverage. If any HIGH-severity findings exist → **block the verify-work PASS gate**. Findings are written to `specs/blind-spots.json`. MEDIUM/LOW findings warn but don't block.
 5a2. **Completeness critic (e45s05)** — `bash scripts/lib/completeness-critic.sh`. **BLOCKER** aborts merge gate; WARNING → gaps loop; FILLED = evidence only.
 
-5b. **NFR Evidence Gate** (P0 only) — Produces go/no-go output on three dimensions: Performance (response time, throughput), Reliability (error rate, recovery), and Operability (logging, health checks). Reads thresholds from `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md` and writes evidence as OKF verification-report bundles to `specs/verifications/NFR-eNNsYY.json`. FAIL on any dimension blocks the gate.
+5b. **NFR Evidence Gate** (P0 only) — Go/no-go on Performance, Reliability, Operability against `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md`. Evidence written to `specs/verifications/NFR-eNNsYY.json`. FAIL blocks gate.
+5c. **Adversarial Defect Mutation Gate (HARD GATE — e86 / ab-20)** — `bash scripts/verify-adversarial.sh`. Boundary fuzzing & parameter injection against modified interfaces. Crashes or failures block PASS gate; evidence in `specs/verifications/ADVERSARIAL_LATEST.yaml`.
 6. **Step-by-step UAT** (skip if P2/P3) — one user-observable action at a time.
 7. **Gaps loop** — failures → log → `plan-work` → re-verify. Unaddressed HIGH findings from step 5 feed into this loop alongside other quality gaps.
 

package/.copilot/skills/build-epic/SKILL.md

@@ -66,7 +66,7 @@ Orchestrates the **build** flow for a single epic: survey → plan tasks → kic
 After step 5 (verify-work) completes successfully, step 6 runs `audit-code` automatically in `--gate` mode:
 
 1. **Run audit:** Invoke `audit-code --gate` on the complete diff for this story.
-2. **Pass (exit 0):** All checklist sections pass → advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
+2. **Pass (exit 0):** All checklist sections pass → run `bash scripts/lib/assert-epic-gate.sh advance` to mechanically confirm the audit pass, then advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
 3. **Fail (exit 1):** One or more checklist sections fail → **reset `epic_cycle.current_step` to `4`** (develop-tdd) and add the failing section IDs to `completed_steps` as `"1,2,3,4,5,6(fail: ...)"`. Record `epic_cycle.audit_result: fail` in `state.yaml`. Do NOT advance past step 6 until audit passes.
 4. **Audit artifact:** Full audit report saved to `specs/verifications/AUDIT-<epic>-<story>.md` regardless of pass/fail, for reviewer traceability.
 5. **Enforce F.I.R.S.T:** After audit-code passes, run `enforce-first --quick` on new/modified tests. Append F.I.R.S.T violations (if any) to the audit report. Failing F.I.R.S.T criteria trigger the same loop-back to step 4.

package/.copilot/skills/commit-message/SKILL.md

@@ -36,6 +36,7 @@ description: "Reviews working-tree changes, then drafts a Conventional Commits t
 
 ## Checklist before finalizing
 
+- [ ] **Audit pass confirmed (HARD GATE — Issue #182):** In active epic flow, `bash scripts/lib/assert-epic-gate.sh commit` passes.
 - [ ] Type matches the **dominant** user-visible outcome (`feat` vs `fix` vs `perf`, etc.).
 - [ ] **Scope** is a short noun in parentheses if it helps (e.g. `fix(api): …`).
 - [ ] Breaking changes are explicit (`!` and/or `BREAKING CHANGE:` in the body/footer).

package/.copilot/skills/guard-git/SKILL.md

@@ -12,7 +12,8 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## What gets blocked/enforced
 
-- **Safety**: `git push --force`, `git reset --hard`, `git clean -f`, `git branch -D`, `git checkout .`, `git restore .`.
+- **Safety**: force push (`-f`, `--force`, `+refspec`), `reset` with `--hard`, `clean` with any `-f*`, `branch -D` / `--delete --force`, `checkout`/`restore` that discard work (`-f`, `.` pathspec).
+- **Argv, not substrings**: the hook tokenizes each command, strips git's global options (`-C <path>`, `-c k=v`, `--no-pager`, `--git-dir=…`) and judges the `(subcommand, arguments)` pair — so `git -C repo reset -q --hard` is blocked, while `git push --force-with-lease` is allowed. Rules live in `scripts/lib/git-argv.py`, mirrored in `extensions/omp-hooks.ts`; `scripts/test-guard-git-normalize.sh` fails if the two disagree.
 - **Discipline**: Blocks direct commits or pushes to protected branches (`main`, `master`) unless `GIT_BIGPOWERS_LAND=1` (set only by `scripts/land-branch.sh`).
 - **Allows**: `git push origin <feature-branch>` for backup/CI; solo land push to `main` only inside `land-branch.sh`.
 - **Standardization**: Enforces [Conventional Commits](https://www.conventionalcommits.org/) for all `git commit` commands.
@@ -37,7 +38,11 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## Customization
 
-To add or remove patterns or protected branches, edit `pre-tool-use.sh`.
+The rule table lives in `scripts/lib/git-argv.py` (shell hooks) and is mirrored in `extensions/omp-hooks.ts`. Add a rule to both and a row to `specs/verifications/fixtures/guard-git-normalize.yaml`.
+
+The TS side is judged through `evaluateBashCall()` — the single pure function the `tool_call` handler calls — so the fixture covers **every** rule (destructive argv, protected-branch push, Conventional Commits), not just the argv table. The gate (`bash scripts/test-guard-git-normalize.sh`) runs both implementations with `GIT_BIGPOWERS_LAND` unset and fails if they disagree, so a verdict can never depend on the developer's shell.
+
+Protected-branch rules answer differently on `main` than on a feature branch, so a row that touches commit/push semantics states its `branch:` and both sides are run in it (the shell hook in a throwaway repo checked out on that branch). A row without `branch:` is judged on a feature branch. Do not pin the branch inside either implementation — that is what made the gate pass on PRs and fail on `main`.
 
 ## Advanced
 

package/.copilot/skills/release-branch/SKILL.md

@@ -56,13 +56,12 @@ git log main...HEAD --format="%B" | grep -qiE 'co[- ]authored[- ]by' && echo "
 ### 2a. Security gate
 
 - [ ] `specs/security/REVIEW.md` exists and is fresh (matches current branch diff)
-- [ ] No unresolved HIGH findings with confidence ≥ 8 (or all documented in `specs/security/EXCEPTIONS.md` with sign-off rationale)
 
-If REVIEW.md is missing or stale → run `security-review` inline. Findings block the merge unless documented in EXCEPTIONS.md.
+### 2b. Audit-code & Traceability gates (HARD GATES — Issue #182)
 
-### 2b. Traceability gate
-
-Run `gate-trace` before merge. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml` (`traceability_override: CONCERNS accepted, reason: <explanation>`). WAIVED if no matrix available.
+- [ ] Run `bash scripts/lib/assert-epic-gate.sh release` (asserts `specs/AUDIT_LATEST.md` passed; blocks release on failure).
+- [ ] No unresolved HIGH security findings with confidence ≥ 8 (or documented in `specs/security/EXCEPTIONS.md`).
+- [ ] Run `gate-trace`. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml`. WAIVED if no matrix available.
 
 > **Adversarial refute framing (e45s32):** The final pre-merge check is **refute, not rubber-stamp**. Before declaring ready, actively try to disprove traceability completeness — missing story tags, absent verify evidence, stale security review. Only proceed when refutation fails.
 

package/.copilot/skills/run-planning/SKILL.md

@@ -62,7 +62,12 @@ if written:
    - `plan-release` — sequence epics by WSJF
    - `slice-tasks` — cut vertical slices
 
-4. **Update status** — On successful completion, set `status: done` for that workflow key in `planning-status.yaml`.
+4. **Mechanical artifact assertion before status update (HARD GATE — Issue #183):**
+   Before marking any workflow key `status: done`, run:
+   ```bash
+   bash scripts/lib/assert-planning-artifacts.sh <workflow-key>
+   ```
+   If the artifact assertion fails, do NOT advance status. Remediate missing output files before continuing. On success, set `status: done` for that workflow key in `planning-status.yaml`.
 
 5. **Advance** — Set `state.yaml` `active_flow: planning` while in this chain. When all required keys are done, set `handoff.next_skill` to `plan-work`.
 

package/.copilot/skills/verify-work/SKILL.md

@@ -62,7 +62,8 @@ Review answers "is the code good?"; Verify answers "does the built thing do what
 5a. **Blind-spot check** — run `bash scripts/check-blind-spots.sh`. This detects structural quality gaps (verify-gap, test-gap, stale-tag, etc.) beyond percentage coverage. If any HIGH-severity findings exist → **block the verify-work PASS gate**. Findings are written to `specs/blind-spots.json`. MEDIUM/LOW findings warn but don't block.
 5a2. **Completeness critic (e45s05)** — `bash scripts/lib/completeness-critic.sh`. **BLOCKER** aborts merge gate; WARNING → gaps loop; FILLED = evidence only.
 
-5b. **NFR Evidence Gate** (P0 only) — Produces go/no-go output on three dimensions: Performance (response time, throughput), Reliability (error rate, recovery), and Operability (logging, health checks). Reads thresholds from `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md` and writes evidence as OKF verification-report bundles to `specs/verifications/NFR-eNNsYY.json`. FAIL on any dimension blocks the gate.
+5b. **NFR Evidence Gate** (P0 only) — Go/no-go on Performance, Reliability, Operability against `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md`. Evidence written to `specs/verifications/NFR-eNNsYY.json`. FAIL blocks gate.
+5c. **Adversarial Defect Mutation Gate (HARD GATE — e86 / ab-20)** — `bash scripts/verify-adversarial.sh`. Boundary fuzzing & parameter injection against modified interfaces. Crashes or failures block PASS gate; evidence in `specs/verifications/ADVERSARIAL_LATEST.yaml`.
 6. **Step-by-step UAT** (skip if P2/P3) — one user-observable action at a time.
 7. **Gaps loop** — failures → log → `plan-work` → re-verify. Unaddressed HIGH findings from step 5 feed into this loop alongside other quality gaps.
 

package/.cursor/rules/build-epic.mdc

@@ -65,7 +65,7 @@ Orchestrates the **build** flow for a single epic: survey → plan tasks → kic
 After step 5 (verify-work) completes successfully, step 6 runs `audit-code` automatically in `--gate` mode:
 
 1. **Run audit:** Invoke `audit-code --gate` on the complete diff for this story.
-2. **Pass (exit 0):** All checklist sections pass → advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
+2. **Pass (exit 0):** All checklist sections pass → run `bash scripts/lib/assert-epic-gate.sh advance` to mechanically confirm the audit pass, then advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
 3. **Fail (exit 1):** One or more checklist sections fail → **reset `epic_cycle.current_step` to `4`** (develop-tdd) and add the failing section IDs to `completed_steps` as `"1,2,3,4,5,6(fail: ...)"`. Record `epic_cycle.audit_result: fail` in `state.yaml`. Do NOT advance past step 6 until audit passes.
 4. **Audit artifact:** Full audit report saved to `specs/verifications/AUDIT-<epic>-<story>.md` regardless of pass/fail, for reviewer traceability.
 5. **Enforce F.I.R.S.T:** After audit-code passes, run `enforce-first --quick` on new/modified tests. Append F.I.R.S.T violations (if any) to the audit report. Failing F.I.R.S.T criteria trigger the same loop-back to step 4.

package/.cursor/rules/commit-message.mdc

@@ -35,6 +35,7 @@ alwaysApply: false
 
 ## Checklist before finalizing
 
+- [ ] **Audit pass confirmed (HARD GATE — Issue #182):** In active epic flow, `bash scripts/lib/assert-epic-gate.sh commit` passes.
 - [ ] Type matches the **dominant** user-visible outcome (`feat` vs `fix` vs `perf`, etc.).
 - [ ] **Scope** is a short noun in parentheses if it helps (e.g. `fix(api): …`).
 - [ ] Breaking changes are explicit (`!` and/or `BREAKING CHANGE:` in the body/footer).

package/.cursor/rules/guard-git.mdc

@@ -11,7 +11,8 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## What gets blocked/enforced
 
-- **Safety**: `git push --force`, `git reset --hard`, `git clean -f`, `git branch -D`, `git checkout .`, `git restore .`.
+- **Safety**: force push (`-f`, `--force`, `+refspec`), `reset` with `--hard`, `clean` with any `-f*`, `branch -D` / `--delete --force`, `checkout`/`restore` that discard work (`-f`, `.` pathspec).
+- **Argv, not substrings**: the hook tokenizes each command, strips git's global options (`-C <path>`, `-c k=v`, `--no-pager`, `--git-dir=…`) and judges the `(subcommand, arguments)` pair — so `git -C repo reset -q --hard` is blocked, while `git push --force-with-lease` is allowed. Rules live in `scripts/lib/git-argv.py`, mirrored in `extensions/omp-hooks.ts`; `scripts/test-guard-git-normalize.sh` fails if the two disagree.
 - **Discipline**: Blocks direct commits or pushes to protected branches (`main`, `master`) unless `GIT_BIGPOWERS_LAND=1` (set only by `scripts/land-branch.sh`).
 - **Allows**: `git push origin <feature-branch>` for backup/CI; solo land push to `main` only inside `land-branch.sh`.
 - **Standardization**: Enforces [Conventional Commits](https://www.conventionalcommits.org/) for all `git commit` commands.
@@ -36,7 +37,11 @@ Installs a shared hook that blocks destructive git operations and enforces workf
 
 ## Customization
 
-To add or remove patterns or protected branches, edit `pre-tool-use.sh`.
+The rule table lives in `scripts/lib/git-argv.py` (shell hooks) and is mirrored in `extensions/omp-hooks.ts`. Add a rule to both and a row to `specs/verifications/fixtures/guard-git-normalize.yaml`.
+
+The TS side is judged through `evaluateBashCall()` — the single pure function the `tool_call` handler calls — so the fixture covers **every** rule (destructive argv, protected-branch push, Conventional Commits), not just the argv table. The gate (`bash scripts/test-guard-git-normalize.sh`) runs both implementations with `GIT_BIGPOWERS_LAND` unset and fails if they disagree, so a verdict can never depend on the developer's shell.
+
+Protected-branch rules answer differently on `main` than on a feature branch, so a row that touches commit/push semantics states its `branch:` and both sides are run in it (the shell hook in a throwaway repo checked out on that branch). A row without `branch:` is judged on a feature branch. Do not pin the branch inside either implementation — that is what made the gate pass on PRs and fail on `main`.
 
 ## Advanced
 

package/.cursor/rules/release-branch.mdc

@@ -55,13 +55,12 @@ git log main...HEAD --format="%B" | grep -qiE 'co[- ]authored[- ]by' && echo "
 ### 2a. Security gate
 
 - [ ] `specs/security/REVIEW.md` exists and is fresh (matches current branch diff)
-- [ ] No unresolved HIGH findings with confidence ≥ 8 (or all documented in `specs/security/EXCEPTIONS.md` with sign-off rationale)
 
-If REVIEW.md is missing or stale → run `security-review` inline. Findings block the merge unless documented in EXCEPTIONS.md.
+### 2b. Audit-code & Traceability gates (HARD GATES — Issue #182)
 
-### 2b. Traceability gate
-
-Run `gate-trace` before merge. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml` (`traceability_override: CONCERNS accepted, reason: <explanation>`). WAIVED if no matrix available.
+- [ ] Run `bash scripts/lib/assert-epic-gate.sh release` (asserts `specs/AUDIT_LATEST.md` passed; blocks release on failure).
+- [ ] No unresolved HIGH security findings with confidence ≥ 8 (or documented in `specs/security/EXCEPTIONS.md`).
+- [ ] Run `gate-trace`. FAIL blocks merge; CONCERNS requires explicit override in `specs/state.yaml`. WAIVED if no matrix available.
 
 > **Adversarial refute framing (e45s32):** The final pre-merge check is **refute, not rubber-stamp**. Before declaring ready, actively try to disprove traceability completeness — missing story tags, absent verify evidence, stale security review. Only proceed when refutation fails.
 

package/.cursor/rules/run-planning.mdc

@@ -61,7 +61,12 @@ if written:
    - `plan-release` — sequence epics by WSJF
    - `slice-tasks` — cut vertical slices
 
-4. **Update status** — On successful completion, set `status: done` for that workflow key in `planning-status.yaml`.
+4. **Mechanical artifact assertion before status update (HARD GATE — Issue #183):**
+   Before marking any workflow key `status: done`, run:
+   ```bash
+   bash scripts/lib/assert-planning-artifacts.sh <workflow-key>
+   ```
+   If the artifact assertion fails, do NOT advance status. Remediate missing output files before continuing. On success, set `status: done` for that workflow key in `planning-status.yaml`.
 
 5. **Advance** — Set `state.yaml` `active_flow: planning` while in this chain. When all required keys are done, set `handoff.next_skill` to `plan-work`.
 

package/.cursor/rules/verify-work.mdc

@@ -61,7 +61,8 @@ Review answers "is the code good?"; Verify answers "does the built thing do what
 5a. **Blind-spot check** — run `bash scripts/check-blind-spots.sh`. This detects structural quality gaps (verify-gap, test-gap, stale-tag, etc.) beyond percentage coverage. If any HIGH-severity findings exist → **block the verify-work PASS gate**. Findings are written to `specs/blind-spots.json`. MEDIUM/LOW findings warn but don't block.
 5a2. **Completeness critic (e45s05)** — `bash scripts/lib/completeness-critic.sh`. **BLOCKER** aborts merge gate; WARNING → gaps loop; FILLED = evidence only.
 
-5b. **NFR Evidence Gate** (P0 only) — Produces go/no-go output on three dimensions: Performance (response time, throughput), Reliability (error rate, recovery), and Operability (logging, health checks). Reads thresholds from `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md` and writes evidence as OKF verification-report bundles to `specs/verifications/NFR-eNNsYY.json`. FAIL on any dimension blocks the gate.
+5b. **NFR Evidence Gate** (P0 only) — Go/no-go on Performance, Reliability, Operability against `specs/tech-architecture/eNN-TEST_PLAN_LATEST.md`. Evidence written to `specs/verifications/NFR-eNNsYY.json`. FAIL blocks gate.
+5c. **Adversarial Defect Mutation Gate (HARD GATE — e86 / ab-20)** — `bash scripts/verify-adversarial.sh`. Boundary fuzzing & parameter injection against modified interfaces. Crashes or failures block PASS gate; evidence in `specs/verifications/ADVERSARIAL_LATEST.yaml`.
 6. **Step-by-step UAT** (skip if P2/P3) — one user-observable action at a time.
 7. **Gaps loop** — failures → log → `plan-work` → re-verify. Unaddressed HIGH findings from step 5 feed into this loop alongside other quality gaps.
 

package/.gemini/extensions/bigpowers/commands/prompts/build-epic.md

@@ -65,7 +65,7 @@ Orchestrates the **build** flow for a single epic: survey → plan tasks → kic
 After step 5 (verify-work) completes successfully, step 6 runs `audit-code` automatically in `--gate` mode:
 
 1. **Run audit:** Invoke `audit-code --gate` on the complete diff for this story.
-2. **Pass (exit 0):** All checklist sections pass → advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
+2. **Pass (exit 0):** All checklist sections pass → run `bash scripts/lib/assert-epic-gate.sh advance` to mechanically confirm the audit pass, then advance to step 7 (commit-message). Record `epic_cycle.audit_result: pass` in `state.yaml`.
 3. **Fail (exit 1):** One or more checklist sections fail → **reset `epic_cycle.current_step` to `4`** (develop-tdd) and add the failing section IDs to `completed_steps` as `"1,2,3,4,5,6(fail: ...)"`. Record `epic_cycle.audit_result: fail` in `state.yaml`. Do NOT advance past step 6 until audit passes.
 4. **Audit artifact:** Full audit report saved to `specs/verifications/AUDIT-<epic>-<story>.md` regardless of pass/fail, for reviewer traceability.
 5. **Enforce F.I.R.S.T:** After audit-code passes, run `enforce-first --quick` on new/modified tests. Append F.I.R.S.T violations (if any) to the audit report. Failing F.I.R.S.T criteria trigger the same loop-back to step 4.