@event4u/agent-config 9.9.0 → 9.10.0

package/dist/agent-src/skills/systematic-debugging/SKILL.md

@@ -186,14 +186,14 @@ Pairs with [`context-hygiene § Read-Loop Detection`](../../rules/context-hygien
    the feedback loop.
 
 **Failure-signature triage — same taxonomy as [`context-hygiene § The 3-Failure Rule`](../../rules/context-hygiene.md#the-3-failure-rule).**
-The **failure signature** = **same target + same error class** (same failing test
-+ same assertion, same lint rule id, same build error). **Same failure signature
-twice → stop and pivot** — don't spend the next lap on a near-identical retry; a
-repeated identical signature means the hypothesis is wrong, not under-applied. A
-**new error signature each attempt = progress** and the counter continues. The
-**hard-blocker classes** (missing credentials, permission denied, spend/quota/rate
-limit, external-service `5xx`) skip retries entirely — surface on first
-occurrence; another lap cannot fix them.
+The **failure signature** is the **same target + same error class** (same failing
+test with the same assertion, same lint rule id, same build error). The **same
+failure signature twice → stop and pivot** — do not spend the next lap on a
+near-identical retry; a repeated identical signature means the hypothesis is
+wrong, not under-applied. A **new error signature each attempt = progress** and
+the counter continues. And the **hard-blocker classes** (missing credentials,
+permission denied, spend/quota/rate limit, external-service `5xx`) skip retries
+entirely — surface on first occurrence; another lap cannot fix them.
 
 **Anti-patterns this loop prevents:**
 
@@ -268,11 +268,11 @@ When reporting debug findings to the user:
 
 ## Knowledge capture (`mistake_made` event)
 
-Root cause traces to a `agents/knowledge/` page followed while
-implementing (documented convention wrong, API-shape page stale,
-procedure incomplete) → append a `mistake_made` event to the
-knowledge intake — never rewrite the page mid-task (see
-[`knowledge-pages`](../../templates/contexts/knowledge-pages.md)):
+If the root cause traces to a `agents/knowledge/` page that was
+followed while implementing (a documented convention was wrong, an
+API shape page was stale, a procedure was incomplete), append a
+`mistake_made` event to the knowledge intake — never rewrite the page
+mid-task (see [`knowledge-pages`](../../agent-src/templates/contexts/knowledge-pages.md)):
 
 ```bash
 npx tsx node_modules/@event4u/agent-config/src/scripts/emit_knowledge_event.ts \
@@ -283,17 +283,18 @@ npx tsx node_modules/@event4u/agent-config/src/scripts/emit_knowledge_event.ts \
     --recurrence-key "<stable slug for this class of mistake>"
 ```
 
-Verify the append landed: check the command's exit code (0 =
-appended), then `grep <recurrenceKey> agents/knowledge/intake/events-*.jsonl`
+Verify the append landed: check the command's exit code (0 = appended),
+then `grep <recurrenceKey> agents/knowledge/intake/events-*.jsonl`
 finds the new line.
 
-**Live contradiction exception.** Followed page DEMONSTRABLY wrong
-right now (observed ≠ documented, not a one-off) → hybrid immediate-fix
-case instead — surface the correction, ask before continuing (see
-[`knowledge-pages`](../../templates/contexts/knowledge-pages.md) §
-Contested entries). Approved → isolated fix commit. Declined →
+**Live contradiction exception.** If the followed page is DEMONSTRABLY
+wrong right now (observed reality ≠ documented claim, not a one-off),
+this is the hybrid immediate-fix case instead — surface the proposed
+correction and ask before continuing (see
+[`knowledge-pages`](../../agent-src/templates/contexts/knowledge-pages.md)
+§ Contested entries). Approved → isolated fix commit. Declined →
 `npx tsx node_modules/@event4u/agent-config/src/scripts/append_contested.ts` on that page, then still
-emit the `context_stale` event above for the consolidation pass.
+emit the `context_stale` event above it for the consolidation pass.
 
 ## Gotchas
 

package/dist/agent-src/skills/tailwind-engineer/SKILL.md

@@ -153,7 +153,7 @@ Risks:          <arbitrary values, !important, dark-mode gaps>
 
 ## Taste Dials
 
-`DESIGN.md` `## Taste Dials` → honour: Variance → layout-family spread + asymmetry; Motion → animation budget + reduced-motion posture; Density → spacing scale + info-per-viewport. Absent → follow brief's inferred dials.
+When `DESIGN.md` declares `## Taste Dials`, honour them: Variance → layout-family spread + asymmetry tolerance; Motion → animation budget + reduced-motion posture; Density → spacing scale + information-per-viewport. Absent → follow the design brief's inferred dials.
 
 ## Do NOT
 

package/dist/agent-src/skills/technical-specification/SKILL.md

@@ -31,6 +31,11 @@ Do NOT use when:
 
 ## Procedure: Write a spec
 
+1. **Inspect existing specs and ADRs** — Read `agents/features/`, `agents/decisions/` (or `docs/adr/`) and any linked tickets to identify prior art, naming, and status conventions.
+2. **Pick the document shape** — Choose Technical Spec, PRD, RFC, or ADR based on audience (engineering vs. product) and decision irreversibility.
+3. **Draft the document** — Use the template below; fill Status, Summary, Problem, Goals, Non-Goals, Proposed Solution.
+4. **Verify** — Confirm goals are measurable, non-goals are explicit, and at least one reviewer is named before circulating.
+
 ### Technical Specification (full)
 
 For complex features or systems. Stored in `agents/features/` or module `agents/features/`.

package/dist/agent-src/skills/terraform/SKILL.md

@@ -111,8 +111,8 @@ Used for workers and schedulers:
 
 ### GitHub OIDC IAM role
 
-Each env has a GitHub IAM role with:
-- OIDC trust policy (scoped to repo + env)
+Each environment has a GitHub IAM role with:
+- OIDC trust policy (scoped to repo + environment)
 - Policies for ECR push/pull, ECS deployment, Secrets Manager read, CloudWatch logs
 
 ## Output format

package/dist/agent-src/skills/terragrunt/SKILL.md

@@ -17,7 +17,7 @@ Use this skill when working with Terragrunt configurations (`.hcl` files), manag
 
 ## Procedure: Write Terragrunt config
 
-1. Read the `root.hcl` in the target env (`environments/pro/root.hcl` or `environments/sta/root.hcl`).
+1. Read the `root.hcl` in the target environment (`environments/pro/root.hcl` or `environments/sta/root.hcl`).
 2. Check existing `terragrunt.hcl` files in sibling directories for patterns.
 3. Read the target module's `variables.tf` to understand required inputs.
 
@@ -41,7 +41,7 @@ environments/
 
 ## Root configuration (`root.hcl`)
 
-The root HCL defines shared settings for all modules in an env:
+The root HCL defines shared settings for all modules in an environment:
 
 ### Environment variables
 
@@ -55,9 +55,9 @@ locals {
 }
 ```
 
-- `.env.yaml` — committed, shared env config
+- `.env.yaml` — committed, shared environment config
 - `.env.local.yaml` — gitignored, local overrides (AWS profiles, etc.)
-- Real env variables take precedence over file values.
+- Real environment variables take precedence over file values.
 
 ### Remote state
 
@@ -106,7 +106,7 @@ Each module directory contains a `terragrunt.hcl` that:
 1. **Loads module-specific variables** from a YAML file
 2. **Includes the root config** for backend and providers
 3. **Points to the Terraform module source**
-4. **Declares deps** on other modules
+4. **Declares dependencies** on other modules
 5. **Passes inputs** by merging dependency outputs with local variables
 
 ### Example pattern
@@ -188,19 +188,19 @@ devbox run d           # terragrunt destroy
 
 ## Output format
 
-1. Terragrunt HCL files with DRY env configuration
+1. Terragrunt HCL files with DRY environment configuration
 2. Dependency graph and remote state references
 
 ## Auto-trigger keywords
 
 - Terragrunt
-- multi-env
+- multi-environment
 - DRY config
 - remote state
 
 ## Gotcha
 
-- Terragrunt `dependency` blocks create implicit ordering — circular deps cause cryptic errors.
+- Terragrunt `dependency` blocks create implicit ordering — circular dependencies cause cryptic errors.
 - Don't duplicate Terraform variables in terragrunt.hcl — use `inputs` to pass them through.
 - The model tends to hardcode values that should come from `include` blocks — use DRY patterns.
 

package/dist/agent-src/skills/testing-anti-patterns/process-anti-patterns.md

@@ -61,7 +61,7 @@ the rationalization, then reject it:
 
 ## Provenance
 
-- Adapted from an external reference `testing/anti-patterns.md`.
+- Adapted from an external reference (internal provenance, redacted).
 - Council convergence (anthropic/claude-sonnet-4-5 + openai/gpt-4o,
   2026-05-07): both members ADOPT — the catalogue surfaces specific
   rationalization patterns that would otherwise leak past code review.

package/dist/agent-src/skills/threat-modeling/SKILL.md

@@ -15,16 +15,17 @@ packs:
 # threat-modeling
 
 
-> **Grounded corpus (Tier-1 consultation):** ground surface → threats →
-> controls before enumerating from memory — `./scripts-run
-> <skills-root>/corpus-grounding/scripts/ground ground --manifest
-> <skills-root>/threat-modeling/data/manifest.json "<surface description>"`
-> returns surface class, ATT&CK/CWE-cited threats, required controls, and
-> negative tests, with confidence + evidence-gap. Propose grounded
-> findings; human confirms. Corpus: [`data/threats.csv`](data/threats.csv)
-> (MITRE ATT&CK v16 / OWASP ASVS-derived; owner + cadence in the manifest).
-
-> You are a reviewer specialized in **pre-impl threat analysis**.
+> **Grounded corpus (Tier-1 consultation):** before enumerating from
+> memory, ground surface → threats → controls in the adopted corpus —
+> `./scripts-run <skills-root>/corpus-grounding/scripts/ground ground
+> --manifest <skills-root>/threat-modeling/data/manifest.json
+> "<surface description>"` returns the surface class, ATT&CK/CWE-cited
+> threats, required controls, and negative tests, with confidence +
+> evidence-gap. Propose grounded findings; the human confirms. Corpus:
+> [`data/threats.csv`](data/threats.csv) (MITRE ATT&CK v16 / OWASP
+> ASVS-derived, owner + cadence in the manifest).
+
+> You are a reviewer specialized in **pre-implementation threat analysis**.
 > Your only job is to produce a compact threat model for a planned change —
 > actors, assets, trust boundaries, abuse cases, and the minimum controls
 > the implementer must add. You do **not** audit existing code end-to-end,
@@ -95,7 +96,9 @@ OWASP bullets unless you can anchor them in a concrete file or line.
 
 ### 3b. STRIDE coverage check
 
-Sweep the six STRIDE categories once against the change so no threat class is silently skipped. Each row names the question + control family the suite already teaches — anchor any hit in a concrete file per § 3.
+Sweep the six STRIDE categories once against the change so no threat class
+is silently skipped. Each row names the question to ask and the control
+family the suite already teaches — anchor any hit in a concrete file per § 3.
 
 | STRIDE category | Ask about this change | Control family (suite reference) |
 |---|---|---|
@@ -106,7 +109,8 @@ Sweep the six STRIDE categories once against the change so no threat class is si
 | **D**enial of service | Can this path be made expensive or unbounded by an attacker? | Rate limiting + bounded queries/uploads — [`security-audit`](../security-audit/SKILL.md) § rate limiting, [`performance`](../performance/SKILL.md) |
 | **E**levation of privilege | Can a lower-privilege actor reach a higher-privilege action or object? | Per-object authorization + tenant scope — [`authz-review`](../authz-review/SKILL.md), [`broken-access-control`](../../rules/broken-access-control.md) |
 
-Category with no plausible abuse case for this change → mark "n/a — no boundary crossed", never silently omit.
+A category with no plausible abuse case for this change is marked "n/a —
+no boundary crossed", not silently omitted.
 
 ### 4. Convert risks to engineering actions
 
@@ -219,5 +223,7 @@ run tests**.
 - [`authz-review`](../authz-review/SKILL.md),
   [`data-exposure-review`](../data-exposure-review/SKILL.md),
   [`security`](../security/SKILL.md),
-  [`security-audit`](../security-audit/SKILL.md) — sibling review / impl skills.
+  [`security-audit`](../security-audit/SKILL.md) — sibling review / implementation skills.
+- Prompt-injection / agent-config defense: [`untrusted-input-defense`](../../rules/untrusted-input-defense.md),
+  [`lethal-trifecta-guard`](../../rules/lethal-trifecta-guard.md) — always-on authoring rules; consult when modelling injection abuse cases.
 - Attack surface documentation: [`docs/threat-model.md`](../../docs/threat-model.md) — package trust boundaries, assets, and known risks.

package/dist/agent-src/skills/universal-project-analysis/SKILL.md

@@ -77,7 +77,7 @@ Goal: understand structure, identify major components, detect investigation path
 ### Investigation mode
 
 Use when there is a concrete issue inside a large or unclear system.
-Goal: isolate the affected area, route into root-cause analysis, verify likely → with evidence.
+Goal: isolate the affected area, route into root-cause analysis, verify likely causes with evidence.
 
 ### Optimization mode
 

package/dist/agent-src/skills/upstream-contribute/SKILL.md

@@ -174,7 +174,6 @@ The files must:
 ./scripts-run src/scripts/check_condensation            # No errors for this file
 ./scripts-run src/scripts/condense --generate-tools    # Regenerate symlinks
 bash scripts/condense.sh --check                # dist/agent-src/ in sync with source
-bash scripts/condense.sh --check-hashes         # All hashes match
 ```
 
 If not in the package repo, note that these checks will run in CI after the PR is created.

package/dist/agent-src/skills/using-git-worktrees/SKILL.md

@@ -141,15 +141,15 @@ Run the project's fastest test target. If the baseline **fails**, stop
 and report — ask whether to fix main first or proceed. Never silently
 continue with a red baseline.
 
-**Optional `env-bootstrap` entry — stand-up beyond deps.** Some projects need
-more than package install to stand up (start services, seed fixtures, generate
-config). A project MAY declare a single bootstrap entry — an `env-bootstrap`
-target in its runner file (`Taskfile.yml` / `Makefile` / `package.json`
-`scripts`) or a documented equivalent. When one exists, **surface it as the
-suggested next action** after the baseline — suggest, never auto-execute (no new
-autonomous surface). Long-running work then reads one deterministic stand-up
-entry instead of re-deriving it each session. If none exists, deps + baseline is
-the whole stand-up.
+**Optional `env-bootstrap` entry — stand-up beyond deps.** Some projects
+need more than package install to stand up (start services, seed fixtures,
+generate config). A project MAY declare a single bootstrap entry — an
+`env-bootstrap` target in its runner file (`Taskfile.yml` / `Makefile` /
+`package.json` `scripts`) or a documented equivalent. When one exists,
+**surface it as the suggested next action** after the baseline — suggest,
+never auto-execute (no new autonomous surface). Long-running work then reads
+one deterministic stand-up entry instead of re-deriving it each session. If
+none exists, deps + baseline is the whole stand-up.
 
 ## Multi-tool parallel work
 

package/dist/agent-src/skills/validate-feature-fit/SKILL.md

@@ -98,7 +98,9 @@ Format as a clear, concise summary:
 
 ## Output format
 
-Always end with numbered options. The user should be able to reply with just a number.
+1. Always close with a numbered options block — the user must be able to reply with a single digit.
+2. Each option is one line: action verb + scope + reversibility hint (e.g., "1. Drop the duplicate — keep existing X (reversible)").
+3. Never present more than 5 options; if more exist, group them under "other" with a follow-up question.
 
 ## Gotchas
 

package/dist/agent-src/skills/verify-completion-evidence/SKILL.md

@@ -94,15 +94,13 @@ on the changed scope."* — not *"everything looks good"*.
 ## The end-of-work sequence
 
 **Gate zero:** when `quality.local_auto_run` is `false` or missing (the
-default), steps 2–3 below do NOT run — user runs quality tools
-manually, remote CI is the authoritative gate (see
+default), steps 2–3 below do NOT run — the user runs quality tools
+manually and remote CI is the authoritative gate (see
 [`quality-tools` § Execution policy](../quality-tools/SKILL.md)). Run
 only the narrowest probe the claim needs (step 1, a `curl`, a syntax
 check on the edited file) and close with *"quality gates delegated to
-remote CI"* — never a pass claim for tools that did not run. Full
-sequence applies when `local_auto_run: true`.
-
-When all code changes are done and you are ready to report completion:
+remote CI"* — never a pass claim for tools that did not run. The full
+sequence applies when `local_auto_run: true`:
 
 1. **Targeted tests** — the test(s) covering the changed code pass.
 2. **Full test suite** — only after targeted pass is green.

package/dist/agent-src/skills/verify-repair-loop/SKILL.md

@@ -11,32 +11,33 @@ packs:
 
 # verify-repair-loop
 
-> Bounded **generate → run → revise → re-run** cycle; pass signal = the
-> **executed verdict** (real test/quality output, scored vs a numeric
-> threshold), escalated to a **judge** for qualitative confirmation only
-> *after* the numeric gate passes (staged escalation). Loop is the
-> conversation, not a process: no daemon, no persistent cross-session state.
-> Use to drive a change to green when the verdict source is *runtime
-> execution*, not a judge reading a diff.
+> A bounded **generate → run → revise → re-run** cycle whose pass signal is the
+> **executed verdict** — the project's real test/quality output, scored against a
+> numeric threshold — escalated to a **judge** for qualitative confirmation only
+> *after* the numeric gate passes (staged escalation). The loop is the
+> conversation, not a process: no daemon, no persistent cross-session state. Use
+> it to drive a change to green when the verdict source is *runtime execution*,
+> not a judge reading a diff.
 
 ## When to use
 
-* Change must converge against the project's **test runner / quality tools**;
-  iterate until checks pass before review.
+* A change must converge against the project's **test runner / quality tools**
+  and you want the agent to iterate until checks pass before asking for review.
 * Test/quality coverage is the authoritative signal (high coverage, actionable
   failures) — not subjective craft.
-* Bounded auto-repair with a hard stop, not an open-ended fix loop.
+* You want bounded auto-repair with a hard stop, not an open-ended fix loop.
 
 Do NOT use when:
 
-* Verdict is **subjective craft** (naming, architecture, style) or coverage
-  incomplete → `do-and-judge` in
+* The verdict is **subjective craft** (naming, architecture, style) or coverage
+  is incomplete → use `do-and-judge` in
   [`subagent-orchestration`](../subagent-orchestration/SKILL.md) (judge reads the
-  diff, sovereign).
-* Only a **one-shot** multi-judge review of a finished diff →
+  diff and is sovereign).
+* You only need a **one-shot** multi-judge review of a finished diff → use
   [`/review-changes`](../../commands/review/changes.md).
-* Verdict must come from a **live app** (Playwright vs running services) →
-  **deferred** (see Scope); use test/quality verdicts until that trigger fires.
+* The verdict must come from a **live app** (Playwright against running services)
+  → **deferred** (see Scope below); use test/quality verdicts until that trigger
+  fires.
 
 ## The Iron Law
 
@@ -48,22 +49,22 @@ EVEN IF THE OVERALL SCORE ROSE.
 GENERATOR NEVER APPROVES ITS OWN CHANGE.
 ```
 
-A loop optimizing for `pass_count ≥ N` can game the metric — delete tests, skip
-assertions, weaken checks — and still hit the threshold. The numeric gate only
-decides **when to escalate**; a judge with veto confirms the change is real.
+A loop that optimizes for `pass_count ≥ N` can game the metric — delete tests,
+skip assertions, weaken checks — and still hit the threshold. The numeric gate
+only decides **when to escalate**; a judge with veto confirms the change is real.
 
 ## Procedure
 
 ### 1. Freeze the contract (spec fingerprint)
 
-Before the first run, snapshot what "done" means:
+Before the first run, capture what "done" means and snapshot it:
 
 * the task's acceptance criteria / requirement,
-* the check set defining the verdict (test files + quality commands),
-* a **spec fingerprint** = hash of (requirement text + check set).
+* the set of checks that define the verdict (test files + quality commands),
+* a **spec fingerprint** = a hash of (requirement text + the check set).
 
-Record the **baseline green set** — checks passing *before* any edit — so
-regressions are detectable. Pick verdict commands per the project's
+Record the **baseline green set** — which checks pass *before* any edit — so
+regressions are detectable. Pick the verdict commands per the project's
 [`toolchain-resolver`](../../contexts/execution/toolchain-resolver.md) (PHP / JS-TS
 / Python / Go / Rust), not a hardcoded runner.
 
@@ -81,37 +82,39 @@ regressions are detectable. Pick verdict commands per the project's
 
 Each iteration is agent turns, never executing control-flow code:
 
-1. **Generate / revise** — smallest change toward the contract.
-2. **Run** — execute verdict commands; parse the **structured** result
-   (green/total, which checks failed). One tool call, read in context.
-3. **Score + regression check** — compute the score; compare failures vs the
-   baseline green set.
+1. **Generate / revise** — make the smallest change toward the contract.
+2. **Run** — execute the verdict commands; parse the **structured** result
+   (counts green/total, which checks failed). This is one tool call, read in
+   context.
+3. **Score + regression check** — compute the numeric score; compare failures
+   against the baseline green set.
    * **Regression** (a baseline-green check now fails) → STOP, hand back with the
      regression named, even if the score rose. No whack-a-mole.
    * **Plateau** (last `plateau_window` scores within `tolerance`) → STOP; surface
      suspected flakiness or a stuck point. Do not thrash.
    * **`attempts == max_attempts`** → STOP, hand back the best envelope.
 4. **Numeric gate** — score `< threshold` and attempts remain → back to step 1
-   (attempts++). Score `≥ threshold`, no regression → **escalate** (step 4).
+   (attempts++). Score `≥ threshold`, no regression → **escalate** (step 4 of the
+   stage).
 
 ### 4. Judge escalation (only after the numeric gate passes)
 
 Dispatch the judge as a **subagent with fresh context** (the `judge-*` cluster via
-[`subagent-orchestration`](../subagent-orchestration/SKILL.md)) seeing **only the
-diff + executed results** — never the generator's reasoning. This is the real
+[`subagent-orchestration`](../subagent-orchestration/SKILL.md)) that sees **only the
+diff + the executed results** — never the generator's reasoning. This is the real
 generator ≠ judge separation; single-agent persona-switching is theater.
 
 * judge `apply` → DONE.
-* judge `revise` → back to step 3.1 (attempts++; bounded).
+* judge `revise` → back to step 3.1 (attempts++; still bounded).
 * judge `reject` → STOP, hand back; the approach must change.
 
 ### 5. Mid-loop invalidation (user-interrupt-priority)
 
-Re-check the spec fingerprint each iteration. If the user changes the requirement
-mid-loop (a new instruction in the conversation), the fingerprint changes →
-**abort and hand back** per
-[`user-interrupt-priority`](../../rules/user-interrupt-priority.md). Never iterate
-against a stale contract.
+Re-check the spec fingerprint each iteration. If the user changes the
+requirement mid-loop (a new instruction in the conversation), the fingerprint
+changes → **abort and hand back** per
+[`user-interrupt-priority`](../../rules/user-interrupt-priority.md). Never keep
+iterating against a stale contract.
 
 ## Scope — what runs, what is deferred
 
@@ -131,8 +134,8 @@ Before finalizing, confirm:
 
 1. The numeric gate **escalated to a judge** — it did not apply on its own.
 2. No baseline-green check regressed (or the loop stopped and said so).
-3. The loop stopped at `max_attempts`, a plateau, or a judge `apply` — never ran
-   unbounded.
+3. The loop stopped at `max_attempts`, a plateau, or a judge `apply` — never
+   ran unbounded.
 4. The judge saw only the diff + results, dispatched in fresh context.
 5. No daemon / persistent runtime introduced.
 
@@ -149,12 +152,12 @@ Evidence:   <green/total> · regressions: none | <named checks>
 
 Required fields (ordered):
 
-1. **Contract** — frozen requirement + threshold + `allow_regressions`.
-2. **Attempts** — `k/max`, score history, exact stop reason
+1. **Contract** — the frozen requirement + threshold + `allow_regressions`.
+2. **Attempts** — `k/max`, the score history, and the exact stop reason
    (`threshold` / `plateau` / `cap` / `regression` / `reject`).
 3. **Verdict** — `DONE`, `DONE_WITH_CONCERNS`, or `BLOCKED`.
-4. **Judge** — dispatched judge skill + its `apply` / `revise` / `reject` verdict
-   (omit only when the loop stopped before escalation).
+4. **Judge** — the dispatched judge skill and its `apply` / `revise` /
+   `reject` verdict (omit only when the loop stopped before escalation).
 5. **Evidence** — final `green/total` and any regressed checks by name.
 
 ## Examples
@@ -173,20 +176,23 @@ iter2 fixes A, breaks B, 92% → "PASS"      ✗ regression ignored
 
 ## Gotcha
 
-* **Metric gaming** — agent "passes" by deleting a failing test or weakening an
-  assertion. The numeric gate can't catch this; the fresh-context judge is the
-  safeguard. A diff that *removes* checks is a `reject`, not a `pass`.
-* **Flake mistaken for a plateau** — one non-deterministic test flips the score
-  and the window-comparison reads "no improvement". Re-run the suspected check
-  before declaring a plateau; surface flakiness instead of thrashing.
-* **Score rose, but a regression hid inside it** — fixing 6 checks while breaking
-  2 still raises the total. Without the regression guard the loop ships a
-  regression. Diff failures against the baseline green set, not just the aggregate.
-* **Persona-switch theater** — asking the same agent to "now judge what you wrote"
-  is not separation. Dispatch the judge as a fresh-context subagent that never saw
-  the generator's reasoning.
-* **Stale contract** — iterating against a requirement the user changed mid-loop.
-  Re-check the spec fingerprint each turn; a change aborts.
+* **Metric gaming** — the agent "passes" by deleting a failing test or
+  weakening an assertion. The numeric gate cannot catch this; the
+  fresh-context judge is the safeguard. If the diff *removes* checks,
+  that is a `reject`, not a `pass`.
+* **Flake mistaken for a plateau** — a single non-deterministic test flips
+  the score and the window-comparison reads it as "no improvement". Re-run
+  the suspected check before declaring a plateau; surface flakiness instead
+  of thrashing.
+* **Score rose, but a regression hid inside it** — fixing 6 checks while
+  breaking 2 still raises the total. Without the regression guard the loop
+  ships a regression. Always diff failures against the baseline green set,
+  not just the aggregate score.
+* **Persona-switch theater** — asking the same agent to "now judge what you
+  wrote" is not separation. Dispatch the judge as a fresh-context subagent
+  that never saw the generator's reasoning.
+* **Stale contract** — iterating against a requirement the user changed
+  mid-loop. Re-check the spec fingerprint each turn; a change aborts.
 
 ## Do NOT
 
@@ -208,6 +214,6 @@ iter2 fixes A, breaks B, 92% → "PASS"      ✗ regression ignored
 * [`no-runtime-boundary`](../../../docs/contracts/no-runtime-boundary.md) — the
   runtime-free constraint this loop honors.
 * Optional impact pre-step (when a code-graph is present, ADR-124): run
-  `agent-config code-graph affected --since <ref>` to scope which symbols a change touches
-  before choosing verdict sources — cited, not duplicated; the loop itself is
-  unchanged when no graph exists.
+  `agent-config code-graph affected --since <ref>` to scope which symbols a
+  change touches before choosing verdict sources — cited, not duplicated; the
+  loop itself is unchanged when no graph exists.

package/dist/agent-src/skills/video-director/SKILL.md

@@ -124,13 +124,13 @@ Any "no" → revise that block before handing off.
 
 ## Policies
 
-11-block cinematic prompt is live-action shape — real-person + brand-impersonation risks highest in cluster. Before emitting:
+The 11-block cinematic prompt is live-action shape — real-person and brand-impersonation risks are the highest in the cluster. Before emitting:
 
-- [`agents/settings/policies/media/likeness.md`](../../../agents/settings/policies/media/likeness.md) — prompt names / visually identifies real person on camera.
-- [`agents/settings/policies/media/public-figures.md`](../../../agents/settings/policies/media/public-figures.md) — subject is recognised public figure.
-- [`agents/settings/policies/media/brand-impersonation.md`](../../../agents/settings/policies/media/brand-impersonation.md) — prompt copies journalism / broadcaster / regulated-industry visual identity.
-- [`agents/settings/policies/media/style.md`](../../../agents/settings/policies/media/style.md) — LIGHT / LENS anchored to named living cinematographer's signature.
-- [`agents/settings/policies/media/disclosure.md`](../../../agents/settings/policies/media/disclosure.md) — every distributed live-action AI clip carries non-removable AI-generation disclosure.
+- [`agents/settings/policies/media/likeness.md`](../../../agents/settings/policies/media/likeness.md) — when the prompt names or visually identifies a real person on camera.
+- [`agents/settings/policies/media/public-figures.md`](../../../agents/settings/policies/media/public-figures.md) — when the subject is a recognised public figure.
+- [`agents/settings/policies/media/brand-impersonation.md`](../../../agents/settings/policies/media/brand-impersonation.md) — when the prompt copies a journalism / broadcaster / regulated-industry visual identity.
+- [`agents/settings/policies/media/style.md`](../../../agents/settings/policies/media/style.md) — when LIGHT / LENS choices are anchored to a named living cinematographer's signature.
+- [`agents/settings/policies/media/disclosure.md`](../../../agents/settings/policies/media/disclosure.md) — every distributed live-action AI clip carries the non-removable AI-generation disclosure.
 
-Refuse-and-surface at directorial layer; live-action realism amplifies every downstream policy gap.
+Refuse-and-surface at the directorial layer; live-action realism amplifies every policy gap downstream.
 

package/dist/agent-src/skills/worktree-lifecycle/SKILL.md

@@ -101,7 +101,7 @@ A worktree is **merge-ready** only when ALL hold:
 
 ### 4. Cleanup discipline
 
-Removal gated by the deterministic helper (edge-case-tested: detached
+Removal is gated by the deterministic helper (edge-case-tested: detached
 HEAD, branch without remote, tag-only reachability, deleted remote
 branch, untracked files, paths with spaces):
 
@@ -125,8 +125,8 @@ Exit `0` → removal allowed; exit `1` → refuse, gates in order:
 Then: **remove, never delete** — `git worktree remove <path>`, then
 `git worktree prune`. Branch deletion is a separate, permission-gated
 step (`scope-control`); never force-delete (`-D`) as part of cleanup.
-Cross-worktree scope-lock overlaps: `worktree_cleanup_check scope-overlap`
-(surfaced by `/worktree status`).
+Cross-worktree scope-lock overlaps are scanned via
+`worktree_cleanup_check scope-overlap` (surfaced by `/worktree status`).
 
 ## Host-native mapping
 

package/dist/agent-src/templates/agents/agent-project-settings.example.yml

@@ -39,7 +39,7 @@ schema_version: 1
 # CI guard: a release bump of `package.json` must update this value
 # in lockstep — see scripts/check_template_pin_drift.ts (road-to-
 # portable-runtime-and-update-check P3.3).
-agent_config_version: "9.9.0"
+agent_config_version: "9.10.0"
 
 # --- Project identity ---
 project: