agent-folder-init（Agent 文件夹初始化）

Agent Folder Init

Create a comprehensive .agents/ folder structure for AI-first development workflows.

Purpose

This skill scaffolds a complete AI agent documentation system including:

• Session tracking (daily files)
• Task management
• Coding standards and rules
• Architecture decision records
• Security checklists
• SOPs for common workflows
• Agent config folders (.claude,.codex,.cursor) with commands, rules, and agents

When to Use

Use this skill when:

• Starting a new project that will use AI coding assistants
• Setting up AI-first development workflows
• Migrating an existing project to use structured AI documentation

Usage

Run the scaffold script:

python3 scripts/scaffold.py --help

# Basic usage
python3 scripts/scaffold.py \
  --root /path/to/project \
  --name "My Project"

# With custom options
python3 scripts/scaffold.py \
  --root /path/to/project \
  --name "My Project" \
  --tech "nextjs,nestjs" \
  --allow-outside

Generated Structure

Documentation (.agents/)

.agents/
├── README.md                    # Navigation hub
├── SYSTEM/
│   ├── README.md
│   ├── RULES.md                 # Coding standards
│   ├── ARCHITECTURE.md          # What's implemented
│   ├── SUMMARY.md               # Current state
│   ├── PRD.md                   # Product Requirements Document
│   ├── ENTITIES.md              # Entity documentation
│   ├── ai/
│   │   ├── SESSION-QUICK-START.md
│   │   ├── SESSION-DOCUMENTATION-PROTOCOL.md
│   │   └── USER-PREFERENCES.md
│   ├── architecture/
│   │   ├── DECISIONS.md         # ADRs
│   │   └── PROJECT-MAP.md
│   ├── critical/
│   │   ├── CRITICAL-NEVER-DO.md
│   │   └── CROSS-PROJECT-RULES.md
│   └── quality/
│       └── SECURITY-CHECKLIST.md
├── TASKS/
│   ├── README.md
│   └── INBOX.md
├── SESSIONS/
│   ├── README.md
│   └── TEMPLATE.md
├── SOP/
│   └── README.md
├── EXAMPLES/
│   └── README.md
└── FEEDBACK/
    └── README.md

Agent Configs

.claude/
├── commands/                    # Slash commands (project-specific)
│   ├── start.md
│   ├── end.md
│   ├── new-session.md
│   ├── commit-summary.md
│   ├── code-review.md
│   ├── bug.md
│   ├── quick-fix.md
│   ├── refactor-code.md
│   ├── inbox.md
│   ├── task.md
│   ├── validate.md
│   └── clean.md
├── agents/                      # Specialized agents (project-specific)
│   ├── senior-backend-engineer.md
│   └── senior-frontend-engineer.md
└── skills/                      # Project-specific skills

.codex/
├── commands/
└── skills/

.cursor/
└── commands/

Note: Agent configs (agents/, commands/) are copied from the installed library bundle so projects get the latest version. Rules are not copied because they are expected to be managed at the user or repo level to avoid duplication and drift.

Root Files

• AGENTS.md - Points to .agents/README.md
• CLAUDE.md - Claude-specific entry point
• CODEX.md - Codex-specific entry point
• .editorconfig - Editor configuration

Key Patterns

Naming Conventions

• Top-level directories: ALL-CAPS (SYSTEM/, TASKS/, SESSIONS/)
• Files: ALL-CAPS for critical files (README.md, RULES.md), kebab-case for others

Session Files

• One file per day: YYYY-MM-DD.md
• Multiple sessions same day use Session 1, Session 2, etc. in the same file

Customization

After scaffolding, customize:

1. SYSTEM/PRD.md - Fill in your product requirements (use with fullstack-workspace-init)
2. SYSTEM/ENTITIES.md - Document your data entities
3. SYSTEM/RULES.md - Add project-specific coding standards
4. SYSTEM/ARCHITECTURE.md - Document your architecture
5. SYSTEM/critical/CRITICAL-NEVER-DO.md - Add project-specific violations
6. SOP/ - Add your standard operating procedures
7. .claude/rules/ - Add project-specific rules
8. .claude/commands/ - Add project-specific commands

Integration with Other Skills

This skill integrates with: