From init-workspace
Initializes multi-agent workspace with shared AGENTS.md config, imports for CLAUDE.md/GEMINI.md, and directory structure for Claude Code, Codex CLI, Gemini CLI. Configures gitignore for private/public AI files.
How this skill is triggered — by the user, by Claude, or both
Slash command
/init-workspace:init-workspaceThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Create shared configuration files for Claude Code, Codex CLI, and Gemini CLI, plus the project structure for all three tools.
Create shared configuration files for Claude Code, Codex CLI, and Gemini CLI, plus the project structure for all three tools.
Ask the user two questions:
Create the following directory structure and files:
project/
├── .claude/
│ ├── settings.json # Team-shared settings
│ ├── agents/ # Custom subagents
│ │ └── .gitkeep
│ ├── skills/ # Custom skills
│ │ └── .gitkeep
│ └── rules/ # Modular rules
│ └── .gitkeep
├── .codex/
│ ├── skills/ # Codex CLI skills
│ │ └── .gitkeep
│ └── prompts/ # Codex CLI custom slash commands
│ └── .gitkeep
├── .gemini/
│ ├── skills/ # Gemini CLI skills
│ │ └── .gitkeep
│ └── commands/ # Gemini CLI custom slash commands (TOML)
│ └── .gitkeep
├── .mcp.json # MCP server configuration
├── .gitignore # Comprehensive gitignore
├── AGENTS.md # Shared instructions (Codex CLI native)
├── CLAUDE.md # Claude Code (imports AGENTS.md)
└── GEMINI.md # Gemini CLI (imports AGENTS.md)
# Project Instructions
> $USER_RESPONSE
## Guidelines
<!-- Add your project-specific instructions here -->
## Shared Memory
**Always write new instructions, rules, and memory to `AGENTS.md` only.**
Never modify `CLAUDE.md` or `GEMINI.md` directly - they only import `AGENTS.md`.
This ensures Claude Code, Codex CLI, and Gemini CLI share the same context consistently.
## Project Structure
- `.claude/agents/` - Custom subagents for specialized tasks
- `.claude/skills/` - Claude Code skills (slash commands)
- `.claude/rules/` - Modular rules auto-loaded into context
- `.codex/skills/` - Codex CLI skills
- `.codex/prompts/` - Codex CLI custom slash commands
- `.gemini/skills/` - Gemini CLI skills
- `.gemini/commands/` - Gemini CLI custom slash commands (TOML)
- `.mcp.json` - MCP server configuration
@AGENTS.md
@AGENTS.md
{
"mcpServers": {}
}
{
"permissions": {
"allow": [],
"deny": []
}
}
If user chose PUBLIC (recommended):
# =============================================================================
# AI CODING ASSISTANTS
# =============================================================================
# Claude Code - local settings contain personal preferences and may include tokens
.claude/settings.local.json
.claude/CLAUDE.local.md
.claude/*.local.*
CLAUDE.local.md
# Codex CLI - local files (sessions, logs, config)
.codex/*
!.codex/skills/
!.codex/prompts/
# Gemini CLI - local files (settings, tmp)
.gemini/*
!.gemini/skills/
!.gemini/commands/
# =============================================================================
# SECRETS & ENVIRONMENT
# =============================================================================
# Environment files often contain API keys, database credentials, etc.
.env
.env.*
!.env.example # Keep example files for documentation
# Private keys and certificates
*.pem
*.key
# Credential files
credentials.json
secrets.json
# =============================================================================
# OPERATING SYSTEM
# =============================================================================
# macOS - Finder metadata
.DS_Store
# Windows - thumbnail cache
Thumbs.db
# =============================================================================
# EDITORS & IDEs
# =============================================================================
# JetBrains (IntelliJ, WebStorm, PyCharm, etc.)
.idea/
# Visual Studio Code
.vscode/
# Vim/Neovim swap and backup files
*.swp
*.swo
*~
# =============================================================================
# DEPENDENCIES
# =============================================================================
# Node.js
node_modules/
# PHP (Composer)
vendor/
# Python
__pycache__/
*.pyc
.venv/
venv/
# =============================================================================
# BUILD OUTPUTS
# =============================================================================
# Common build directories
dist/
build/
# Python package build
*.egg-info/
# Rust/Cargo
target/
# =============================================================================
# LOGS & TEMPORARY FILES
# =============================================================================
*.log
logs/
# =============================================================================
# TEST COVERAGE
# =============================================================================
# JavaScript/TypeScript
coverage/
.nyc_output/
# Python
htmlcov/
If user chose PRIVATE:
Use the same gitignore as above, but ADD these lines to the AI CODING ASSISTANTS section:
# AI config files (private mode - not shared with team)
AGENTS.md
CLAUDE.md
GEMINI.md
.claude/
.codex/skills/
.codex/prompts/
.gemini/skills/
.gemini/commands/
.mcp.json
Create .gitkeep files in empty directories to ensure they're tracked by git.
Tell the user:
npx claudepluginhub xiaolai/init-workspaceAutomatically configure Claude Code best practices for new projects: directory skeleton, AGENTS.md, hooks, and interactive placeholders fill. Invoke via `/init-project`.
Generates Claude Code project setups including CLAUDE.md, hooks, permissions, commands, and agents. Analyzes stack (TypeScript, JavaScript, Python, Go, Rust, etc.) to create minimal/standard/full configs.
Initializes Claude Code project configuration by reading global protocol, detecting project type, and generating project-specific CLAUDE.md, AGENTS.md, and docs/. Use for new project setup or when configuring Claude Code.