AI Dev OS Plugin — Claude Code

A Claude Code plugin that brings the AI Dev OS 4-layer model to life as executable Skills, Hooks, and Sub-agents.

Part of AI Dev OS — a framework for turning tacit knowledge into enforceable AI coding rules (Lifespan Layers). Requires AI Dev OS Rules set up in your project.

Why This Plugin?

Turn AI Dev OS guidelines into automated workflows for Claude Code:

• 11 Skills — /ai-dev-os-check, /ai-dev-os-scan, /ai-dev-os-extract and more
• 3 Sub-agents — Philosophy advisor, principle checker, guideline auditor
• Lifecycle Hooks — Auto-check on code changes, pre-commit reminders
• One command setup — npx ai-dev-os init --rules typescript --plugin claude-code

Quick Start

npx ai-dev-os init --rules typescript --plugin claude-code

Adds submodules, copies CLAUDE.md template, and merges hooks automatically. See AI Dev OS CLI for details.

Requires Claude Code >= 1.0.0 and AI Dev OS layer files (L1-L3) in your project (TypeScript / Python).

Manual Setup

# 1. Add AI Dev OS rules as submodule
git submodule add https://github.com/yunbow/ai-dev-os-rules-typescript.git docs/ai-dev-os
# For Python projects:
# git submodule add https://github.com/yunbow/ai-dev-os-rules-python.git docs/ai-dev-os

# 2. Add this plugin as submodule
git submodule add https://github.com/yunbow/ai-dev-os-plugin-claude-code.git .claude/plugins/ai-dev-os

1. Copy the hooks from hooks/hooks.json into your project's .claude/settings.json
2. Run /ai-dev-os-init in Claude Code chat to set up the 4-layer structure
3. Start coding — hooks will guide you automatically

See Operation Guide for detailed instructions.

Skills

Skill	Command	Description
Init	/ai-dev-os-init	Setup wizard — introduces AI Dev OS to a project in 30 minutes
Check	/ai-dev-os-check	Checks code changes against guidelines (supports git diff, --staged, branch comparison)
Scan	/ai-dev-os-scan	Full project-wide compliance scan of ALL source files
Review	/ai-dev-os-review	Pre-PR self-review combining L3 compliance + L2 design review + L1 alignment
Extract	/ai-dev-os-extract	Extracts new rules from code review diffs (Rule Harvesting)
Why	/ai-dev-os-why	Traces a rule back through L3→L2→L1 to explain its rationale
Plan	/ai-dev-os-plan	Creates an implementation plan with guideline checklists before coding
Ticket	/ai-dev-os-ticket	Generates a ticket (local file or GitHub Issue) with implementation summary and checklist candidates
Audit	/ai-dev-os-audit	Audits 4-layer health: dependency rule, freshness, coverage, consistency
Evolve	/ai-dev-os-evolve	Analyzes recent commits to propose L1-L2 updates (SECI spiral)
Report	/ai-dev-os-report	Generates compliance summary for teams and stakeholders

Sub-agents

Agent	Model	Role
philosophy-advisor	Opus	Architecture decisions based on L1 philosophy
principle-checker	Sonnet	Verifies code changes align with L2 principles
guideline-auditor	Sonnet	Audits L3 guideline coverage, consistency, and freshness

Hooks

Event	Trigger	Action
PreToolUse (Write/Edit)	Code changes	Lightweight guideline compliance check
PreToolUse (Bash: git commit)	Before commit	Reminder to run /ai-dev-os-check
PostToolUse (Write/Edit)	After L1-L2 file edits	Dependency rule violation warning

Package Structure

ai-dev-os-plugin-claude-code/
├── .claude-plugin/
│   └── plugin.json              # Plugin manifest
├── skills/
│   ├── ai-dev-os-init/          # Setup wizard
│   ├── ai-dev-os-check/         # Guideline compliance check (git diff)
│   ├── ai-dev-os-scan/          # Full project-wide compliance scan
│   ├── ai-dev-os-review/        # Pre-PR self-review (L1-L3)
│   ├── ai-dev-os-extract/       # Rule Harvesting from code
│   ├── ai-dev-os-why/           # Explain rule rationale (L3→L2→L1)
│   ├── ai-dev-os-audit/         # 4-layer health audit
│   ├── ai-dev-os-plan/           # Guideline-aware implementation planning
│   ├── ai-dev-os-ticket/        # Ticket generation with checklist candidates
│   ├── ai-dev-os-evolve/        # SECI spiral feedback (L4→L1)
│   └── ai-dev-os-report/        # Compliance report generation
├── agents/
│   ├── philosophy-advisor.md     # L1-based architecture judgment
│   ├── principle-checker.md      # L2 alignment verification
│   └── guideline-auditor.md      # L3 coverage & consistency audit
├── hooks/
│   └── hooks.json                # Lifecycle event hooks
├── templates/
│   ├── CLAUDE.md.template        # CLAUDE.md template with cascade
│   ├── ai-dev-os-starter/        # Minimal setup template
│   └── ai-dev-os-full/           # Full 4-layer template
└── settings.json                  # Default permission settings

Specificity Cascade

When rules conflict: framework-specific > common > project-specific > decision criteria > philosophy. → Details

Related

Repository	Description
ai-dev-os	Framework specification and theory
rules-typescript	TypeScript / Next.js / Node.js guidelines
rules-python	Python / FastAPI guidelines
plugin-kiro	Steering Rules and Hooks for Kiro
plugin-cursor	Cursor Rules (.mdc)
cli	npx ai-dev-os init
benchmark	Quantitative benchmark — guideline impact data

License

MIT

---

Languages: English | 日本語 | 简体中文 | 한국어 | Español