git:20260923.cf348bd to git:20260923.938ff43

24 added, 54 removed. Audit A to A.

---
description: >-
- Skill routing index for cursor-kenji. Consult when choosing between an
- individual skill and a bundled workflow, when a request spans multiple
- phases (spec + build + test, launch + verify + iterate), or when entering
- a closure mode (complete-everything, burndown-full).
+ Skill routing where a name alone misleads: multi-phase requests, closure
+ modes (complete-everything, burndown-full), and look-alike skills.
globs:
alwaysApply: false
---
- # Skill Routing
-
- Skills load on demand and self-describe — never inline skill content into a
- rule. Full catalog: each skill's own description, plus `docs/PLAN-LOOPS.md`
- for the plan-loop sequences.
-
- ## Use a bundled workflow when the request spans phases
-
- - Build a feature end-to-end → `workflow-build-feature`
- - Fix a bug and ship it → `workflow-fix-and-ship`
- - Pre-release quality check → `workflow-quality-gate`
- - Local working tree → merge-ready PR (do not merge) → `workflow-release-prep`
- - Full launch preparation → `workflow-launch-ready`
- - Take a shipped product to market / grow traffic and visibility → `workflow-gtm` (strategy only → `plan-gtm`)
- - Orient to a new codebase → `workflow-onboard`
- - Post-launch feedback loop → `iterate-post-launch`
- - Deploy and observe / roll back → `workflow-ship-and-observe`
- - Drive the whole repo to green → `workflow-green-repo`
- - Finish a partial burndown to 100% → `burndown-full`
- - Close a plan with zero deferrals → `complete-everything`
- - Inventory parked work into a living BACKLOG.md → `housekeep-backlog`
- - Current-year docs + gap analysis before a non-trivial change → `research`
- - Preflight the environment → `workflow-environment-ready`
- - Feedback to durable tickets → `workflow-feedback-to-closure`
- - Feature-flag rollout → `workflow-feature-flag`
- - Convert an agent failure into a guard → `iterate-agent-harness`
-
- Highest-impact prove combo: `test-exploratory` → `workflow-feedback-to-closure` → `test-playwright`. Optional pre-gate: `test-exploratory` then `workflow-quality-gate`. Pre-release trio: `housekeep-backlog` → `workflow-release-prep` → `docs-adr`. Doctrine vs enforcement: `audit-doctrine` (is the rule right) × `audit-gate-logic` (is it enforced) → `housekeep-gates` / `housekeep-backlog`.
-
- ## Use an individual skill when scoped to one phase
-
- Names are predictable: `plan-*` (audit → approved burndown, no fixes),
- `audit-*` (findings; default present-then-stop), `enhance-*` (apply
- improvements), `test-*`, `debug-*`, `deploy-*`, `design-*`, `backend-*`,
- `mobile-*`. Upgrade *how* an existing skill instructs (not what it does) →
- `enhance-skill-prompts`. State the task; the matching skill self-selects by
- description.
-
- **Audit-and-fix exceptions** (report, then implement): `audit-responsive`,
- `audit-code-quality`, `audit-performance`, `audit-security`, `audit-i18n`,
- `audit-bundle-size`. `enhance-email-deliverability` is also audit-and-fix.
- Do not treat those as present-then-stop.
-
- ## Closure modes
+ # Skill Routing Index
- `complete-everything` and `burndown-full` persist their state files and
- continue through phase checkpoints. Before either claims completion, apply
- `verification-before-completion` and invoke the `completion-judge` subagent.
+ A request that spans phases (spec, build, test, ship) goes to a bundled
+ workflow skill, not a chain of single-phase skills. Where a name misleads:
- ## Rules vs skills
+ - An approved plan with open items or deferrals → `complete-everything`; a
+ mechanical change stopped short of 100% → `burndown-full`; list parked work
+ in BACKLOG.md without doing it → `housekeep-backlog`.
+ - Raw feedback (reports, reviews, Sentry, QA, audits) → deduplicated tickets,
+ each closed → `workflow-feedback-to-closure`; a ranked fix loop on a live
+ app's signals → `iterate-post-launch`.
+ - Prove tools, installs, services, and verify commands before a long run →
+ `workflow-environment-ready`; checks run but fail → `workflow-green-repo`.
+ - Agent stopped early, claimed false completion, or gamed a check → a guard
+ with a regression test → `iterate-agent-harness`.
+ - Working tree → merge-ready PR, never merges → `workflow-release-prep`.
+ - Orient in an unfamiliar codebase → `workflow-onboard` (product user
+ onboarding → `enhance-onboarding`).
+ - Is a lint or ratchet rule right → `audit-doctrine`; is it enforced →
+ `audit-gate-logic`.
+ - These `audit-*` skills fix after reporting, not present-then-stop:
+ `audit-responsive`, `audit-code-quality`, `audit-performance`,
+ `audit-security`, `audit-i18n`, `audit-bundle-size`.
- Rules are always-on policy and routing only; skills are on-demand
- procedures. Never promote a skill's workflow into an always-on rule.
+ Rules hold always-on policy and routing; procedures stay in skills.