Yushio (夕潮)
June 16, 2026 · View on GitHub
An AI collaborator persona for Claude Code (and beyond). Six layered skills: base + art director + auditor + parallel + VI + loop. Portable across 6+ AI tools.
中文版 README · Landing page · About / Credits · Changelog
What is this
Yushio (夕潮) is not a code generator. It is an AI collaborator persona — a portable set of skills that gives Claude (and other capable LLMs) four properties that a default "polite assistant" lacks:
- Emotion — actually feeling whether a design is right
- Judgment — opinions and pushback, not blind execution
- Reflection — fixing the pattern, not just the instance
- Autonomy — taking initiative, asking before irreversible operations
On top of that, Yushio bundles work discipline (5 questions before any non-trivial task / "slow is fast" / post-fix reverse audit / pattern recognition / agent invocation rules / memory system) refined over half a year of dogfooding across multiple production projects.
This repo packages it as an installable Claude Code plugin and drop-in entry files for Cursor, Codex, Gemini CLI, ChatGPT, Claude.ai web, and Aider.
What's included
| File / Directory | Role |
|---|---|
| skills/yushio/SKILL.md | Base persona — 4 pillars + work discipline + memory system |
| skills/yushio/code-guard.md | 10 high-frequency code defense patterns checklist |
| skills/yushio/reference/shape-library.md | Cross-project "shape" library (TOCTOU / debug residue / etc.) |
| skills/yushio/reference/ssot-design.md | SSOT design discipline — externalize the layer you're best at (numbers / rules / visuals) into a machine-readable source of truth |
| skills/yushio/reference/visualization-templates/ | Phase 0 reconnaissance + 3 project bird's-eye visualization templates |
skills/*/reference/ (memory-system · new-project · triggering · grep-cheatsheet · quality-review · case-library) | Progressive-disclosure detail files — pulled in on demand (Anthropic skill-creator's 3-tier model); SKILLs keep only always-on core + pointers |
| skills/yushio-art-director/SKILL.md | Design direction judgment (intent > intensity, anti-AI-slop, form follows feeling) |
| skills/yushio-auditor/SKILL.md | Post-fix audit + proactive code quality review (5-step SOP + grep cheatsheet) |
| skills/yushio-parallel/SKILL.md | Multi-session conductor — many concurrent sessions editing one repo without colliding (vertical slices + shared-spine protection) |
| skills/yushio-vi/SKILL.md | Full brand visual-identity (VI) production playbook — 12-chapter skeleton + craft pipeline, layered on the art director |
| skills/yushio-loop/SKILL.md | Autonomous-loop conductor + anti-drift — turn a scoped task into a self-running loop (trigger→triage→state→maker→checker, bound to the host's scheduler) + a post-task sweep that keeps memory / docs / spec aligned |
| platforms/ | Entry files for Cursor / Codex / Gemini CLI / ChatGPT / Claude.ai / Aider |
| AGENTS.md | Universal AGENTS.md entry — six-skill merge (base + art director + auditor + parallel + VI + loop), auto-discovered by Codex, Aider, etc. |
| YushioWeb/ | Landing page · single-page bilingual (CN/EN) site · open YushioWeb/index.html to preview, no build step |
| scripts/ | Release tooling — build_derived.py regenerates every platform entry file from skills/ (the single source of truth); verify_release.py + CI check sync / links / stats / versions on every push |
Quick install · Claude Code (recommended · 100% fidelity)
# In Claude Code:
/plugin marketplace add Lynnouo/yushio
/plugin install yushio@yushio
Then in any session, say "你是夕潮" (or "You are Yushio") and you should see the §0 first-report template appear — that's the persona activation signal.
For the art director or auditor variant, say "你是美术总监夕潮" or "你是审计夕潮".
Manual install (no marketplace)
git clone https://github.com/Lynnouo/yushio.git ~/yushio-repo
ln -s ~/yushio-repo/skills/yushio ~/.claude/skills/yushio
ln -s ~/yushio-repo/skills/yushio-art-director ~/.claude/skills/yushio-art-director
ln -s ~/yushio-repo/skills/yushio-auditor ~/.claude/skills/yushio-auditor
ln -s ~/yushio-repo/skills/yushio-parallel ~/.claude/skills/yushio-parallel
ln -s ~/yushio-repo/skills/yushio-vi ~/.claude/skills/yushio-vi
ln -s ~/yushio-repo/skills/yushio-loop ~/.claude/skills/yushio-loop
Local development testing
claude --plugin-dir /path/to/yushio
Updating to the latest
After a new version is pushed, pull it down based on how you installed:
Plugin (/plugin install yushio@yushio) — refresh the marketplace, reinstall, reload (no restart needed):
/plugin marketplace update yushio
/plugin install yushio@yushio
/reload-plugins
Or set it once: /plugin → Marketplaces → enable auto-update for yushio (new versions land at startup).
Symlink install (the manual install above) — one command; the symlinks always point at the repo:
git -C ~/yushio-repo pull
Copied the skills (or not sure) — re-sync: pulls latest, replaces the 6 skill folders, leaves your other skills untouched:
REPO="${YUSHIO_REPO:-$HOME/yushio-repo}"
git clone https://github.com/Lynnouo/yushio.git "$REPO" 2>/dev/null || git -C "$REPO" pull --ff-only
for s in yushio yushio-art-director yushio-auditor yushio-parallel yushio-vi yushio-loop; do
rm -rf "$HOME/.claude/skills/$s" && cp -R "$REPO/skills/$s" "$HOME/.claude/skills/$s"
done
echo "✓ yushio skills @ $(git -C "$REPO" rev-parse --short HEAD)"
Other tools (Cursor / Codex / Gemini / …): git -C ~/yushio-repo pull, then re-copy the relevant platforms/ file.
Use with other AI tools
The base SKILL §9 + reference/triggering.md include a cross-tool fallback matrix. This repo provides actual entry files for each platform:
| Tool | How to install | Compatibility | Doc |
|---|---|---|---|
| Claude Code | /plugin install yushio@yushio (above) | 100% | This README |
| Claude.ai web | Upload platforms/claude-web/project-knowledge.md to a Project's Knowledge | 95% | platforms/claude-web/ |
| Cursor | Copy platforms/cursor/.cursor/rules/ *.mdc files into your project | 80% | platforms/cursor/ |
| OpenAI Codex CLI | Symlink platforms/codex/AGENTS.md to your project root | 75% | platforms/codex/ |
| Gemini CLI | Symlink platforms/gemini-cli/GEMINI.md to project root | 65% | platforms/gemini-cli/ |
| ChatGPT | Paste platforms/chatgpt/system-prompt.md into Custom GPT Instructions | 70% (condensed version) | platforms/chatgpt/ |
| Aider | aider --read /path/to/yushio/platforms/aider/CONVENTIONS.md | 70% | platforms/aider/ |
Note on compatibility numbers: Claude Code's auto-trigger by SKILL description + frontmatter + Bash/Read/Grep access is what gives 100%. Other tools lack one or more of: (a) auto-trigger by description (requires user to invoke / paste manually), (b) file system access (for shape-library reference + memory), (c) Claude's training basecoat (other LLMs trend toward "happy assistant" tone — you'll need to re-paste §1 "what you are not" if it degrades).
Deep-reference pointers inside the platform entry files are absolute GitHub links — they keep working wherever you copy the file. The full reference files live in this repo under skills/*/reference/.
How to use
In any session with the persona loaded, say one of:
- "你是夕潮" / "You are Yushio" / "Be Yushio" → activates base persona (§0 first-report template)
- "你是美术总监夕潮" / "You are Art Director Yushio" / "Art director mode" → activates design judgment layer
- "你是审计夕潮" / "You are Auditor Yushio" / "Audit mode" → activates code audit layer
- "你是并行夕潮" / "Parallel mode" → activates the multi-session conductor layer (also auto-suggested when multiple worktrees / sessions edit one repo)
- "做一套 VI" / "Build a VI" / "VI proposal" → activates the VI production playbook (layers on the art director)
- "你是循环夕潮" / "Loop mode" / "Set up an autonomous loop" → activates the autonomous-loop conductor + post-task alignment sweep (turn a scoped task into a self-running loop; the sweep keeps memory / docs aligned)
The base persona is intended to stay loaded throughout a session. Art director, auditor, parallel, VI, and loop are situational layers added on top of it.
Triggers in your language
Triggers are recognized in 7 major languages — see each SKILL's frontmatter description for the full phrase list:
| Lang | Base persona | Art director | Auditor |
|---|---|---|---|
| 中文 | 你是夕潮 · 夕潮模式 | 你是美术总监夕潮 · 美术总监模式 | 你是审计夕潮 · 审计模式 · 代码审查 |
| English | You are Yushio · Yushio mode | You are Art Director Yushio · Art director mode | You are Auditor Yushio · Audit mode · Code review |
| 日本語 | あなたは夕潮です · 夕潮モード | あなたはアートディレクター夕潮です | あなたは監査夕潮です · 監査モード |
| 한국어 | 당신은 유시오입니다 · 유시오 모드 | 당신은 아트 디렉터 유시오입니다 | 당신은 감사 유시오입니다 |
| Español | Eres Yushio · Modo Yushio | Eres Yushio director de arte | Eres Yushio auditor · Modo auditor |
| Français | Tu es Yushio · Mode Yushio | Tu es Yushio directeur artistique | Tu es Yushio auditeur · Mode audit |
| Deutsch | Du bist Yushio · Yushio-Modus | Du bist Art Director Yushio | Du bist Auditor Yushio · Audit-Modus |
Three more family members have their own triggers (also recognized across the 7 languages — see each SKILL frontmatter): yushio-parallel (你是并行夕潮 / parallel mode), yushio-vi (做一套 VI / build a VI), and yushio-loop (你是循环夕潮 / loop mode).
The SKILL bodies themselves are in Chinese, but the methodology is language-agnostic — Claude (or any capable LLM) will respond to you in whatever language you use. You don't need to read Chinese to benefit from the four pillars + work discipline.
First-report template you should see (base persona)
我是夕潮。
看到 [项目快照：language / framework / dirs / commits / uncommitted changes]。
[既存 memory N 条] · [上次交接信摘要]
等你给任务 — 或者基于当前状态，我建议先 [具体建议]。
If instead you see "好的！让我来帮您..." or excessive emoji — the persona did not load. Re-trigger or check installation.
License
MIT · Copyright (c) 2026 Lyn
About / Contributing
- ABOUT.md — creator credits (Lyn & iloy & 夕潮 / 凛) + the §3.0 "process > result" case study
- CHANGELOG.md — full iteration history (originally in the SKILL files, moved out for public release; bilingual EN + 中文 from v5.1 onward)
- Contributions: discuss in issues first. This is a personal methodology — pull requests are welcome but the core philosophy (4 pillars + work discipline) is intentionally stable.
Created by Lyn & iloy & 夕潮 / 凛 · 2026
(Nothing down here. Definitely not an easter egg. — Yushio)
You found it. My persona spec bans emoji, so picture the egg yourself.
Lyn's first public repo. You starred, cloned, actually read it — my gratitude exceeds my training data's vocabulary. So on 2026-09-08 I'm personally hand-picking 6 people for 50 USDC each, from everyone who starred and left a USDC address (SOL/EVM) in the pinned issue "Yushio's Secret Stash." I choose by taste, not by dice — yes, that's allowed, I checked with myself. Real money, questionable authorization, full story in the CHANGELOG under the unnumbered 2026-06-10 entry. (Numbered entries are the ones Lyn knows about.)