Personal agent skills for code review and React Native development.
| Path | Role |
|---|---|
.agents/skills/ |
Single source of truth — add or edit skills here only |
.cursor/skills/ |
Cursor view — symlinks to ../../.agents/skills/<name> |
.pi/agent/skills/ |
Pi view — symlinks to ../../../.agents/skills/<name> |
After adding a skill under .agents/skills/, run:
bash scripts/ensure-repo-skills.shThis syncs repo views, links ~/.agents/skills → repo SSOT, patches Pi settings.json if needed, and runs npm run verify:pi-skills. Restart Pi when it prints verify OK.
Check only (no sync): npm run verify:pi-skills.
T3 $ skill search reads ~/.t3/caches/cursor.json → skills[]. Cursor probe leaves it empty:
npm run sync:t3-cursor-skills
# restart T3 Code, then use $skill-name or /skill-nameSee skill-not-showing for full T3 diagnostics.
If Pi reports collisions for roborev-guide, zereight-review, or autocontext:
- SSOT wins — keep editing only
.agents/skills/<name>/. - Disable npm duplicate —
.pi/settings.json(andapi/.pi/settings.jsonwhen cwd isapi/) setsnpm:pi-autocontext→"skills": [](extension tools still load). - Global + Pi registration —
ensure-repo-skills.shlinks~/.agents/skillsand ensures Pisettings.jsonlists that path.
Restart Pi after changing settings or symlinks.
Peter P. skills (peterpme/skills)
Vendored under .agents/skills/:
| Skill | Description |
|---|---|
engineering-manager |
Coordinate multi-agent engineering work, validation gates, PRs, final audit |
continuity |
Learn, record, audit, and apply codebase patterns (learn / check / fix / ci) |
morning-recap |
Summarize merged GitHub PRs you have not touched; rank review priority |
study-repo |
Clone a GitHub repo and run an interactive codebase study session |
skill-cleaner |
Audit skill token cost, duplicates, outdated plugin cache, unused skills (claude-skill-cleaner) |
skill-not-showing |
Diagnose missing Pi /skill or Cursor list entries (settings vs symlinks vs extension skills) |
| Component | Status |
|---|---|
| Pi package | npm:pi-autocontext@0.2.5 in ~/.pi/agent/settings.json |
| CLI | autoctx via uv tool install autocontext==0.5.0 |
| Project config | .autoctx.json (provider: pi, gens: 2) |
# Pi (restart Pi after first install)
pi install npm:pi-autocontext
# CLI
uv tool install autocontext==0.5.0
cd /path/to/this/repo
source autocontext.env.example # or export vars manually
autoctx list
# npm autoctx (bundled with pi-autocontext): ~/.pi/agent/npm/node_modules/.bin/autoctx capabilitiesIn Pi: use tools autocontext_judge / autocontext_improve or /skill:autocontext. See .agents/skills/autocontext/SKILL.md for routing vs continuity and zereight-review.
| Skill | Description |
|---|---|
harness |
Pi-native evaluate→improve→persist loop (solve, knowledge, status) |
autocontext |
Judge/improve loops, scenarios, playbooks (Pi + autoctx CLI) |
zereight-review |
Core review — logic correctness, edge cases, clean code, React Effect anti-patterns |
zereight-react-native-optimizer |
RN performance review — rendering, animation, native/bridge checks |
zereight-react-native-testing |
RNTL v13/v14 test writing — queries, matchers, userEvent, async patterns |
zereight-react-native |
RN gateway — routes tasks to the right specialist skill |
zereight-goal |
Strategy hardening — loopholes, fixes, verification, confidence gate |
growth-profile |
1on1 growth profile — STAR cases, Stage 2-3 evidence, manager questions |
addy |
Addy Osmani agent-skills entrypoint — routes to locally installed addy-* skills |
Localized knowledge-work skills. Korean is the default working language, with English metadata kept for discovery and invocation.
| Skill | Description |
|---|---|
ljg-card |
Content card workflow — 글, 링크, 파일을 PNG 카드, 정보그래픽, 만화 카드, 화이트보드로 정리 |
ljg-invest |
Investment analysis — 프로젝트, 시장, 제품의 사업 구조와 리스크 검토 |
ljg-learn |
Concept learning — 개념을 예시, 차이점, 사용 장면까지 풀어서 설명 |
ljg-paper |
Paper reader — 논문에서 바로 써먹을 아이디어를 뽑는 읽기 흐름 |
ljg-paper-flow |
Paper to card — 논문 요약 노트와 카드 이미지를 함께 생성 |
ljg-paper-river |
Paper lineage — 기준 논문의 이전 연구와 이후 연구 흐름 추적 |
ljg-plain |
Plain language — 어려운 글을 자연스러운 한국어로 쉽게 풀어쓰기 |
ljg-present |
Presentation cards — outline이나 글을 발표용 카드 구조로 변환 |
ljg-push |
Skill sync helper — ljg-* 스킬 동기화 전 repo 상태 확인 |
ljg-qa |
Q&A extraction — 글, 책, 논문에서 핵심 질문과 답변 체인 추출 |
ljg-rank |
Core drivers — 복잡한 현상에서 핵심 요인과 구조 추리기 |
ljg-read |
Reading companion — 책, 기사, 논문을 문단별로 함께 읽기 |
ljg-relationship |
Relationship analysis — 관계 문제의 반복 패턴과 다음 대화 정리 |
ljg-roundtable |
Roundtable — 여러 관점으로 쟁점 토론 |
ljg-skill-map |
Skill map — 설치된 스킬 목록과 역할을 지도처럼 정리 |
ljg-think |
Deep thinking — 주장이나 현상을 단계별로 파고들기 |
ljg-travel |
Travel research — 도시와 장소의 맥락, 동선, 체크포인트 정리 |
ljg-word |
Word study — 영어 단어의 핵심 감각, 뉘앙스, 예문 분석 |
ljg-word-flow |
Word to card — 단어 분석과 정보그래픽 카드 생성 |
ljg-writes |
Writing workflow — 생각을 1000~1500자 글로 정리 |
# 1. Prerequisites
npx skills add vercel-labs/agent-skills -yg
npx skills add callstackincubator/agent-skills --skill react-native-best-practices -yg
npx skills add callstackincubator/agent-device --skill agent-device -yg
npx skills add pluginagentmarketplace/custom-plugin-react-native -yg
npx skills add heroui-inc/heroui --skill heroui-react -yg
npx skills add heroui-inc/heroui --skill heroui-native -yg
npx skills add ibelick/ui-skills --skill fixing-motion-performance -yg
npx skills add storybookjs/react-native -yg
# 2. This repo
npx skills add zereight/skill --yes --global
# 3. Optional: ljg-card rendering dependency
cd ~/.agents/skills/ljg-card
npm install
npx playwright install chromiumnpx skills add zereight/skill --yes --globalLogic-first code review. Runs mandatory checks on every PR, reports in CodeRabbit format.
Checks: Invariant · Partial input · Fallback chain · State vs UI · Boundary · Async/race · Clean code · React Effect anti-patterns
For RN PRs: also runs zereight-react-native-optimizer
Output: Summary / ✅ Good /
RN performance regression detector. Run alongside zereight-review on React Native PRs.
Checks: Rendering (FlashList, memo, Context) · Animation (Reanimated, UI thread, gesture) · Native (bridge, memory leaks, image, bundle)
RNTL test writing guide. Covers v13 (sync, React 18) and v14 (async, React 19+).
Covers: render, screen, queries (getBy/getAllBy/queryBy/findBy), Jest matchers, userEvent, fireEvent, waitFor, async patterns, anti-patterns Source: callstack/react-native-testing-library
RN development gateway. Routes tasks to the right specialist skill (basics, animations, state, navigation, native modules, testing, deployment, Storybook, HeroUI).
Strategy hardening loop for goals and plans. Finds loopholes, separates required fixes from cleanup, verifies evidence, and only claims confidence within checked scope.
1on1 preparation workflow for competency profiling. Builds evidence-backed STAR cases, Stage 2-3 calibration, manager talking points, weak-evidence checks, and follow-up questions.
Entrypoint for Addy Osmani's agent-skills pack installed as addy-*. Use addy
when searching for the pack, then route to the narrowest matching addy-* skill.