@kensaurus/skills 0.0.0-stage → 2.1.0

package/commands/native-rn-monorepo/rn-ship-ios.md

@@ -0,0 +1,67 @@
+---
+description: "Verify locally, push, trigger iOS CI, and watch the run through the TestFlight upload step"
+effort: low
+---
+
+# /rn-ship-ios
+
+End-to-end: verify locally what can be verified, push, trigger iOS
+CI, and watch.
+
+Use this when you've made changes that need to reach TestFlight
+(bridge changes, RN version bumps, `Podfile` edits, or any
+cross-platform feature where iOS-specific verification is needed).
+
+## Steps
+
+1. **Pre-flight verification** (delegates to `/rn-verify` checks):
+   - `npx tsc --noEmit`
+   - `npm run lint` if available
+   - `cd android && ./gradlew :app:compileDebugKotlin && cd ..` —
+     Kotlin failure means a likely RN-version or Gradle issue that
+     probably hits iOS too.
+   - Stop here if anything fails.
+
+2. **Confirm git state**:
+   - `git status` — must be clean OR all relevant changes
+     committed.
+   - `git log --oneline -5` — confirm the last commit is what you
+     expect to ship.
+
+3. **Push to the working branch**:
+   ```
+   git push origin <branch>
+   ```
+
+4. **Trigger iOS CI** (delegates to `/ios-ci-trigger` logic):
+   ```
+   gh workflow run ios.yml --ref <branch>
+   sleep 10
+   gh run list --workflow=ios.yml --limit=1
+   ```
+
+5. **Watch the run**:
+   ```
+   gh run watch <run-id>
+   ```
+   Or, for a non-blocking flow, just print the run URL and let the
+   user check later.
+
+6. **On success**:
+   - Confirm the TestFlight upload step ran (check the job list).
+   - Remind the user that Apple processing takes 5–30 min before
+     the build is installable.
+   - Print bundle version + build number from the workflow output
+     if the workflow surfaces them.
+
+7. **On failure**:
+   - Delegate to `/ios-ci-logs` for diagnosis.
+   - Do not auto-rerun without user confirmation.
+
+## Don't
+
+- Don't ship a dirty working tree.
+- Don't ship without TS + Android compile passing — the iOS macOS
+  runner is expensive (10x Linux), don't burn it on shipping
+  known-broken code.
+- Don't skip step 1 even if "it's a small change."

package/commands/native-rn-monorepo/rn-verify.md

@@ -0,0 +1,53 @@
+---
+description: "Smoke-compile TypeScript, lint, Android Kotlin/Java, and web typecheck; iOS is verified via CI"
+effort: low
+---
+
+# /rn-verify
+
+Quick compile check across all locally-buildable targets. iOS is
+verified via CI (use `/ios-ci-trigger`).
+
+## Steps
+
+Run in sequence. Stop and report at the first failure.
+
+1. **TypeScript**:
+   ```
+   npx tsc --noEmit
+   ```
+2. **ESLint** (skip if no `lint` script):
+   ```
+   npm run lint
+   ```
+3. **Android compile only** (faster than full assembleDebug):
+   ```
+   cd android && ./gradlew :app:compileDebugKotlin :app:compileDebugJavaWithJavac --console=plain && cd ..
+   ```
+4. **Web** (skip if no `web/` directory):
+   ```
+   cd web && npm run typecheck && cd ..
+   ```
+
+Note: iOS is NOT compiled locally. The shared JS layer running
+through `tsc` covers most iOS-relevant errors; native iOS bridge
+code can only be verified by triggering CI.
+
+## Output format
+
+```
+✓ TypeScript
+✓ ESLint
+✗ Android — compileDebugKotlin failed: <first error>
+✓ Web
+- iOS (skipped — verify via CI: `/ios-ci-trigger`)
+```
+
+## Don't
+
+- Don't run full `assembleDebug` — that's `/android-build`. This is
+  a smoke test.
+- Don't keep going past the first failure unless asked for a full
+  report.
+- Don't pretend iOS is verified just because TS passed — bridge
+  changes need CI.

package/commands/perf-plan.md

@@ -0,0 +1,18 @@
+---
+description: "Measure-don't-guess performance audit + optimization plan — no fixes until approved"
+argument-hint: "[url or scope]"
+disable-model-invocation: true
+effort: high
+---
+
+# Performance Audit Plan
+
+Run the **`plan-perf-audit`** skill: measure performance (Core Web Vitals,
+bundle, queries) and produce an optimization plan grounded in measurements, not
+guesses. **Audit and plan only — optimize nothing until the plan is approved.**
+
+Emit findings ranked by measured impact with locations and a phased checklist.
+After approval, execute with `audit-performance`, `backend-db-performance`, and
+`workflow-refactor`.
+
+The full playbook lives in the **`plan-perf-audit`** skill.

package/commands/plan-mode.md

@@ -0,0 +1,74 @@
+---
+description: "Research, clarify requirements, and produce an approved implementation plan before writing code"
+argument-hint: "[task or feature]"
+disable-model-invocation: true
+effort: high
+---
+
+# /plan-mode
+
+> Research the codebase, clarify requirements, and produce an approved plan before writing code.
+
+> **Named `/plan-mode` (not `/plan`).** Claude Code, the Cursor CLI, and Gemini CLI all ship a built-in `/plan` that enters Plan mode. This file is the procedure; the built-in is the mode switch. The portable copy is named `plan-mode` for the same reason.
+
+## Purpose
+
+Use Plan mode to research, clarify requirements, and produce an approved implementation plan before writing code.
+
+> **Runtime note:** In Cursor, `Shift+Tab` from the chat input rotates to Plan mode, and the mode picker does the same. Plans save to the home directory by default; "Save to workspace" moves one into the repo (`.cursor/plans/`) for sharing and later sessions. Claude Code also toggles plan mode with `Shift+Tab` and approves through its plan-mode prompt. There, a workspace plan file is not guaranteed.
+
+## Process
+
+### 1. Enter Plan mode
+
+Press `Shift+Tab` in the agent input, use the mode picker, or run the host's built-in `/plan`. Do not treat this `/plan-mode` file as the mode switch.
+
+The agent will:
+
+- Search the codebase for relevant files and context
+- Ask clarifying questions about requirements
+- Produce a detailed plan with file paths and code references
+- Wait for approval before building
+
+### 2. Review and refine the plan
+
+Plans open as Markdown. Edit directly to:
+
+- Remove unnecessary steps
+- Adjust the approach or architecture
+- Add context the agent missed
+- Specify patterns, libraries, or constraints
+
+**Save to workspace** when the plan should outlive the session. Cursor stores that copy under `.cursor/plans/`.
+
+### 3. Approve and execute
+
+Once the plan looks right, approve it. The agent switches to implementation mode.
+
+If the result doesn't match expectations:
+
+1. Revert the changes
+2. Refine the plan with more specifics
+3. Re-run — faster than fixing a wrong implementation
+
+### 4. Checklist
+
+- [ ] Requirements clearly stated
+- [ ] Existing patterns identified (check codebase first)
+- [ ] Affected files listed
+- [ ] Edge cases noted
+- [ ] Plan saved to the workspace if it should be shared or resumed
+- [ ] Plan approved before implementation starts
+
+## When to use
+
+- Complex features with more than one valid approach
+- Tasks that touch many files or systems
+- Unclear requirements
+- Architectural decisions you want to review first
+
+## When to skip
+
+- Small, isolated changes you've done many times
+- Bug fixes with an obvious root cause
+- Single-file edits with clear requirements

package/commands/pr.md

@@ -0,0 +1,16 @@
+---
+description: "Validate and open or manage a PR from an already-committed branch"
+argument-hint: "[title]"
+disable-model-invocation: true
+---
+
+# /pr
+
+> Validate and open or manage a PR from an already-committed branch.
+
+This command is a thin entry point. The full PR lifecycle playbook lives in the **`workflow-pr`** skill, which covers validation, monitoring, bot-feedback handling, and merge criteria.
+
+Use `/pr` when you want to create or manage a PR from an already-committed
+branch. Otherwise, say "open a PR" or "manage this PR".
+
+Related: `workflow-git-commit` for commit hygiene, `audit-code-review` for pre-merge review, `babysit` (Cursor skill) for keeping the PR merge-ready after open. Dirty working tree that still needs review + commit + a merge-ready PR is `/release-prep` (`workflow-release-prep`).

package/commands/pricing-plan.md

@@ -0,0 +1,20 @@
+---
+description: "Pricing and packaging audit — value metric, tiers, price corridor, research plan — plan only, no price edits"
+argument-hint: "[product or pricing route]"
+disable-model-invocation: true
+effort: high
+---
+
+# Pricing Plan
+
+Run the **`plan-pricing`** skill: inventory the plans, gates, metering, and
+price points in the repo, score candidate value metrics, design ≤3 tiers with
+an anchor, propose a price corridor with the research that confirms it
+(Van Westendorp → Gabor-Granger → MaxDiff → conjoint), and emit
+`plan-pricing.md`. **Plan only — no prices, plan enums, Stripe objects, or
+copy change until approved.**
+
+After approval: `enhance-web-conversion` (page), `audit-payment-system`
+(billing), `audit-analytics` (pricing events).
+
+The full playbook lives in the **`plan-pricing`** skill.

package/commands/privacy-plan.md

@@ -0,0 +1,18 @@
+---
+description: "Privacy / GDPR / APPI / store-label audit — plan only, no policy or SDK edits until approved"
+argument-hint: "[path or scope]"
+disable-model-invocation: true
+effort: high
+---
+
+# Privacy Compliance Plan
+
+Run the **`plan-privacy-compliance`** skill: map personal-data flows from the
+code, diff them against the privacy policy and store labels, then produce a
+phased remediation plan. **Audit and plan only — change nothing until the
+plan is approved.**
+
+Consent-gated analytics firing is shared with `audit-analytics` — run both
+when tracking SDKs are in play.
+
+The full playbook lives in the **`plan-privacy-compliance`** skill.

package/commands/readability.md

@@ -0,0 +1,12 @@
+---
+description: "Audit and fix how easily content is understood — CPL, Gestalt grouping, deadspace, visuals that cut verbosity"
+argument-hint: "[route, surface, or dense copy]"
+disable-model-invocation: true
+---
+
+# /readability
+
+Follow the **`enhance-readability`** skill. Comprehension first: measure text,
+then grouping / deadspace / icons, then generate a table or diagram only if it
+lowers cognitive load. Looks-good → `enhance-web-ui`. Breakpoints →
+`audit-responsive`. WCAG floor → `audit-accessibility`.

package/commands/readme.md

@@ -0,0 +1,15 @@
+---
+description: "End-of-session README sync — detect convention, smart change detection, stale-reference cleanup, visual showcase"
+disable-model-invocation: true
+---
+
+# /readme
+
+> End-of-session doc sync — detect convention, smart change detection, stale-reference cleanup.
+
+This command is a thin entry point. Two skills cover the README workflow:
+
+- **`enhance-readme`** — visual showcase: hero image, feature tour grid, optional GIF, theme-aware screenshots via playwright-cli.
+- **`docs-writer`** — text content, structure, accuracy.
+
+Use `/readme` to explicitly trigger doc sync. Otherwise, say "update the README", "enhance README", or "make README prettier" and the appropriate skill auto-fires.

package/commands/refactor.md

@@ -0,0 +1,15 @@
+---
+description: "Systematic refactoring: analyze → split → extract → verify behavior, with no truncation or omitted lines"
+argument-hint: "[file or module]"
+disable-model-invocation: true
+---
+
+# /refactor
+
+> Systematic refactoring: analyze → split → extract → verify behavior. Move code, do not retype it: edit surgically, keep every public API and behavior identical, and leave code the refactor does not touch untouched. A bug you find on the way is a follow-up, not part of the refactor.
+
+This command is a thin entry point. The full playbook lives in the **`workflow-refactor`** skill, which covers safe incremental transformations, splitting large files, extracting components/hooks, and dependency-aware verification.
+
+Use `/refactor` when you want to explicitly trigger refactoring. Otherwise, say "refactor this", "split this file", "extract a hook", or "clean up this module" and the skill auto-fires.
+
+Related: `audit-code-quality` for spotting smells before refactoring, `audit-code-review` for verifying behavior preservation after.

package/commands/release-prep.md

@@ -0,0 +1,17 @@
+---
+description: "Review the local working tree, commit, push, open a PR, drive it merge-ready — do not merge"
+argument-hint: "[optional title]"
+disable-model-invocation: true
+---
+
+# /release-prep
+
+> Take the local working tree to a merge-ready PR against main. Do not merge.
+
+This command is a thin entry point. The full playbook lives in the
+**`workflow-release-prep`** skill.
+
+Use `/release-prep` when the dirty tree (uncommitted + staged + untracked
++ unpushed) should become one reviewed, merge-ready PR. Existing-PR
+lifecycle stays on `/pr` (`workflow-pr`). Product launch sweep stays on
+`workflow-launch-ready`. Commit-message-only stays on `/commit`.

package/commands/research.md

@@ -0,0 +1,25 @@
+---
+description: "Research current best practices with Context7, Firecrawl, and built-in search before implementing anything non-trivial"
+argument-hint: "[topic or question]"
+disable-model-invocation: true
+---
+
+# research
+
+Follow the **`research`** skill end to end. This command is a thin
+entry point; do not replace the skill with a shorter improvised workflow.
+
+In brief:
+
+1. Read the current repo first — manifests, the files under change, and
+   explicit research questions.
+2. Fetch official, version-matched docs (Context7 when available).
+3. Run Firecrawl broad search → deep scrape → discovery (or WebSearch /
+   WebFetch if Firecrawl is unavailable).
+4. Produce a gap analysis and a file-mapped implementation plan.
+5. Do not implement until the user asks or approves.
+
+The full playbook lives in the **`research`** skill.
+
+Related: `workflow-onboard` for first-contact orientation, `plan-*` for
+audit-only burndowns, `docs-adr` once a decision is made.

package/commands/responsive-audit.md

@@ -0,0 +1,21 @@
+---
+description: "Responsive layout & IA audit — desktop is not a wide phone"
+argument-hint: "[path or route scope]"
+disable-model-invocation: true
+effort: high
+---
+
+# /responsive-audit
+
+Thin entry for the **`audit-responsive`** skill. Optional scope:
+`/responsive-audit src/pages/dashboard`.
+
+Finds and fixes linearized mobile layouts at every breakpoint — stacked,
+left-aligned, full-width regardless of screen size, with no hierarchy or
+grouping. **Desktop is not a wide phone.**
+
+Do **not** route this to `design-mobile-first` (touch / mobile-up) or
+`audit-ux-journeys` (cross-page IA). Those are neighbors, not this job.
+
+If the user said `/uiux` and the symptom is "1440 looks like a stretched
+phone", use this skill.

package/commands/review-code.md

@@ -0,0 +1,18 @@
+---
+description: "Agent review pass plus manual checklist for correctness, security, performance, and accessibility"
+argument-hint: "[path or PR]"
+disable-model-invocation: true
+effort: high
+---
+
+# /review-code
+
+> Agent review pass plus manual checklist for correctness, security, performance, and accessibility.
+
+This command is a thin entry point. The full playbook lives in the **`audit-code-review`** skill, which integrates Firecrawl for current best-practice verification and Sentry MCP to check whether the change relates to production errors.
+
+> **Named `/review-code` (not `/review`)** to avoid colliding with Claude Code's built-in `/review` command.
+
+Use `/review-code` when you want to explicitly trigger review. Otherwise, say "review this code", "check my PR", or "review this diff" and the skill auto-fires.
+
+Related: `audit-security` for OWASP-focused review, `audit-performance` for Core Web Vitals impact, `workflow-coding-discipline` for behavioral guardrails.

package/commands/rls-plan.md

@@ -0,0 +1,18 @@
+---
+description: "Supabase RLS + access-control audit + remediation plan — no SQL until approved"
+argument-hint: "[project or scope]"
+disable-model-invocation: true
+effort: high
+---
+
+# RLS Audit Plan
+
+Run the **`plan-rls-audit`** skill: audit Supabase/Postgres row-level security
+and access control, then produce a remediation plan. **Audit and plan only —
+write no SQL and change no policies until the plan is approved.**
+
+Emit findings ranked by severity (tables without RLS, over-permissive policies,
+`service_role` exposure) with a phased fix checklist. After approval, execute
+with `db-migrator` and `backend-patterns`.
+
+The full playbook lives in the **`plan-rls-audit`** skill.

package/commands/secrets-plan.md

@@ -0,0 +1,18 @@
+---
+description: "Secrets & key-scope audit (rotate vs relocate) — plan only, no rotation until approved"
+argument-hint: "[path or scope]"
+disable-model-invocation: true
+effort: high
+---
+
+# Secrets Audit Plan
+
+Run the **`plan-secrets-audit`** skill: scan the codebase and git history for
+exposed secrets and over-scoped keys, then produce a rotate-vs-relocate plan.
+**Audit and plan only — rotate or move nothing until the plan is approved.**
+
+Emit findings ranked by severity with locations (never print secret values) and
+a phased remediation checklist. After approval, execute rotation/relocation
+with provider consoles and `backend-patterns`.
+
+The full playbook lives in the **`plan-secrets-audit`** skill.

package/commands/security-plan.md

@@ -0,0 +1,19 @@
+---
+description: "OWASP Top 10 + Supabase security audit + hardening plan — no fixes until approved"
+argument-hint: "[path or scope]"
+disable-model-invocation: true
+effort: high
+---
+
+# Security Audit Plan
+
+Run the **`plan-security-audit`** skill: audit against the OWASP Top 10 plus
+Supabase-specific risks (auth, RLS, secrets, injection), then produce a
+hardening plan. **Audit and plan only — change no code until the plan is
+approved.**
+
+Emit findings ranked by severity with concrete `file:line` locations and a
+phased hardening checklist. After approval, execute with `audit-security`,
+`plan-rls-audit`, and `backend-patterns`.
+
+The full playbook lives in the **`plan-security-audit`** skill.

package/commands/ship-and-observe.md

@@ -0,0 +1,36 @@
+---
+description: "Take merged, green code to a verified, monitored production release — deploy, confirm the revision is live, smoke-test, observe, and roll back if needed"
+argument-hint: "[environment]"
+disable-model-invocation: true
+---
+
+# Ship & Observe
+
+Take merged, repository-green code all the way to a **verified, monitored**
+production release for a running app. For npm package releases use
+`/deploy-npm` instead.
+
+Follow the **`workflow-ship-and-observe`** skill end to end. In brief:
+
+1. **Pin the target and revision** — confirm the exact merged SHA and the
+   environment/provider (discover from `vercel.json`, `fly.toml`, `Dockerfile`,
+   CI deploy workflows, EAS, Terraform, etc.). Identify the deploy mechanism,
+   backend dependencies, an executable rollback plan, and the signals +
+   thresholds that mean "roll back".
+2. **Preflight** — repo green, revision merged; deploy and verify backend
+   dependencies first (migrations, functions, RLS, secrets).
+3. **Deploy** via the real pipeline and wait for a terminal state.
+4. **Verify the revision is live** — the deployed build must equal the intended
+   SHA (deployment metadata / version endpoint / asset hash), not just a 200.
+   Smoke-test critical flows against production with evidence.
+5. **Observe** the defined stability window; crossing a threshold triggers
+   rollback or hotfix.
+6. **Report** STABLE / ROLLED BACK / HOTFIXED / BLOCKED with evidence.
+
+**Shipping is not merging.** Deployed-verified needs a proven live revision;
+observed-stable needs the monitoring window to pass.
+
+The full playbook lives in the **`workflow-ship-and-observe`** skill.
+
+Related: `deploy-verify`, `full-stack-ship-discipline`, `iterate-post-launch`,
+`workflow-feedback-to-closure`.

package/commands/skill-conflicts.md

@@ -0,0 +1,19 @@
+---
+description: "Audit the skill pack for contradictions, overlapping triggers, and stale cross-refs — report only, no rewrites until approved"
+argument-hint: "[path or pack root]"
+disable-model-invocation: true
+effort: high
+---
+
+# Skill-Pack Conflict Audit
+
+Run the **`audit-skill-conflicts`** skill: treat the installed skills,
+commands, and always-on rules as one system. Find opposite directives,
+ambiguous trigger descriptions, dangling handoffs, and greedy descriptions
+that waste context.
+
+**Read-only — rewrite no SKILL.md until the user approves the fix plan.**
+Description rewrites first (routing). Per-file spec lint stays on
+`validate:skills`. Authoring how-to stays on `meta-skill-creator`.
+
+The full playbook lives in the **`audit-skill-conflicts`** skill.

package/commands/slop-plan.md

@@ -0,0 +1,18 @@
+---
+description: "AI slop / authenticity / voice audit + de-slop burndown — plan only, no rewrites until approved"
+argument-hint: "[path or scope]"
+disable-model-invocation: true
+effort: high
+---
+
+# Anti-Slop Plan
+
+Run the **`plan-antislop`** skill: audit the codebase, UI, and copy for
+machine-generated slop, authenticity, and voice, then produce a de-slop
+burndown. **Audit and plan only — rewrite nothing until the plan is approved.**
+
+Emit findings ranked by severity with concrete `file:line` locations and a
+phased de-slop checklist. After approval, execute with `docs-writer`,
+`audit-i18n`, `enhance-web-ui`, and `enhance-web-ux`.
+
+The full playbook lives in the **`plan-antislop`** skill.

package/commands/stub-plan.md

@@ -0,0 +1,18 @@
+---
+description: "Stub / dead-link / fake-component audit + wiring plan — no fixes until approved"
+argument-hint: "[path or scope]"
+disable-model-invocation: true
+effort: high
+---
+
+# Stub Checker Plan
+
+Run the **`plan-stub-checker`** skill: exhaustively audit for stubs, dead
+buttons, unwired handlers, fake components, and dead links, then produce a
+wiring plan. **Audit and plan only — wire nothing until the plan is approved.**
+
+Emit findings ranked by severity with concrete `file:line` locations and a
+phased wiring checklist. After approval, execute with `debug-fe-be-integration`
+and `workflow-fix-and-ship`.
+
+The full playbook lives in the **`plan-stub-checker`** skill.

package/commands/test-mutation.md

@@ -0,0 +1,16 @@
+---
+description: "Set up mutation testing to prove tests assert behavior, not just coverage"
+argument-hint: "[paths to mutate]"
+disable-model-invocation: true
+---
+
+# Mutation Testing
+
+Run the **`test-mutation`** skill: install StrykerJS or mutmut, scope to
+critical paths (never whole-repo first), triage survivors, and wire an
+incremental per-PR job plus a score ratchet.
+
+Coverage plan stays on `plan-test-coverage`. Writing tests stays on
+`test-unit`. "Compiles and covers" is not "would notice a bug."
+
+The full playbook lives in the **`test-mutation`** skill.

package/commands/test-plan.md

@@ -0,0 +1,17 @@
+---
+description: "User-story-driven test coverage audit + plan — no tests written until approved"
+argument-hint: "[path or scope]"
+disable-model-invocation: true
+effort: high
+---
+
+# Test Coverage Plan
+
+Run the **`plan-test-coverage`** skill: build a user-story traceability matrix,
+find untested paths and fake-green tests, then produce a coverage plan. **Audit
+and plan only — write no tests until the plan is approved.**
+
+Emit gaps ranked by risk with a phased coverage checklist. After approval,
+execute with `test-unit`, `workflow-spec-tdd`, and `test-playwright`.
+
+The full playbook lives in the **`plan-test-coverage`** skill.

package/commands/test.md

@@ -0,0 +1,29 @@
+---
+description: "Route to the right test skill: unit, headed Playwright, QA, exploratory, visual regression, load, or mobile emulator"
+argument-hint: "[target or flow]"
+disable-model-invocation: true
+---
+
+# /test
+
+> Pick the test skill that matches the target. Do not run one generic suite and call it done.
+
+This command is a thin entry point. Pick the right skill for the target:
+
+- **`test-unit`** — write and improve unit tests (Jest, Vitest, pytest, Go test).
+- **`test-playwright`** — PDCA the changes you just made: drive the live app for the touched flows + blast radius and fix pain points as you go.
+- **`test-qa`** — full-app webapp QA, dynamic user stories, CRUD verification.
+- **`test-exploratory`** — unscripted monkey wander as GUEST then LOGGED-IN, then a guest-vs-authed diff. Discovery only — do not fix in-pass.
+- **`test-visual-regression`** — Playwright screenshot baselines and CI pixel diffs.
+- **`test-load`** — k6/Artillery throughput, latency percentiles, breaking point. Never hit prod unsigned.
+- **`mobile-emulator-test`** — native mobile builds (RN/Expo/Capacitor/Flutter) on Android emulator.
+
+> **All browser QA is manual and headed.** Drive a **visible** browser through playwright-cli
+> (`npx --yes @playwright/cli@latest -s=<session> …`, with `--headed` on `open`) one real user
+> action at a time — never `*.spec.ts`, `npx playwright test`, or driving the
+> UI through `eval`/`run-code`. First read the
+> `protocol-browser-anti-stall` skill (Rule 0).
+
+Use `/test` to explicitly trigger testing. Otherwise, say "test this", "test my changes with playwright", "write tests", "QA the app", "monkey test the app", "guest vs logged in", "add visual regression tests", "load test this", or "smoke-test on emulator" and the appropriate skill auto-fires.
+
+Highest-impact combo: `test-exploratory` → `workflow-feedback-to-closure` → `test-playwright` (wander twice → ticket real bugs → lock the worst ones). Pre-release: wander first, then `workflow-quality-gate`.