AI CLI Toolkit

Universal toolkit for AI CLI tools, featuring MCP server setup and command conversion.

Features

• MCP Server Setup: Interactively configure MCP servers for Claude Code, Claude Desktop, VS Code, and Gemini CLI.
• Dynamic Server Configuration: Add and configure MCP servers dynamically with client-specific optimizations.
• Universal Command Format: Write commands once in YAML, convert them for all supported platforms.
• Platform-Specific Optimization: Each platform receives an optimized format and structure.
• Bundled Commands: Comes with a set of pre-built commands for immediate use.
• Flexible Configuration: Customize source and output directories for commands.
• Dry Run Mode: Preview changes and conversions before they are executed.

Supported Platforms

MCP Server Setup

• Claude Code
• Claude Desktop
• VS Code (GitHub Copilot)
• Gemini CLI

Command Conversion

| Platform | Format | Structure | |----------|--------|-----------| | Claude Code | .md | Preserves subfolders | | GitHub Copilot | .prompt.md | Prefix naming (e.g., kiro_commit.prompt.md) | | Gemini CLI | .toml | Preserves subfolders |

Installation

npm install -g @quydoan/ai-cli-toolkit

Usage

The toolkit is available under two aliases: ai-cli-toolkit and ait.

setup Command

Interactively set up and configure MCP servers for your AI CLI clients.

# Start the interactive setup process
ait setup

# Preview the setup without making changes
ait setup --dry-run

convert Command

Convert YAML commands to formats compatible with different AI CLI platforms.

# Convert bundled commands
ait convert

# Preview the conversion process
ait convert --dry-run

# Use a custom source directory for your commands
ait convert --source ./my-commands

# Specify custom output directories
ait convert --claude-dir ./claude --copilot-dir ./copilot --gemini-dir ./gemini

Command Line Options

setup

Options:
  --dry-run  Preview changes without applying them
  --config   Path to a custom configuration file
  --help     Show help

convert

Options:
  --source       Source directory for YAML command files (defaults to bundled commands)
  --claude-dir   Output directory for Claude Code commands (default: ./.claude/commands)
  --copilot-dir  Output directory for GitHub Copilot prompts (default: ./.github/prompts)
  --gemini-dir   Output directory for Gemini CLI commands (default: ./.gemini/commands)
  --dry-run      Show what would be done without making changes
  --help         Show help

YAML Command Format

Create your commands using this universal YAML format:

name: commit
description: Create a git commit with a descriptive message
prompt: |
  Create a git commit with the following message: $ARGUMENTS

  First, run git status to see what files are staged, then create the commit.
arguments: "[message]"
tools:
  - "Bash(git add:*)"
  - "Bash(git commit:*)"
model: "claude-3-sonnet"  # optional

Subfolder Organization

Commands in subfolders are handled differently per platform during conversion:

commands/
├── commit.yaml
└── git/
    └── advanced-commit.yaml

Conversion Results:

• Claude: commit.md, git/advanced-commit.md
• Copilot: commit.prompt.md, git_advanced-commit.prompt.md
• Gemini: commit.toml, git/advanced-commit.toml

Platform Formats

Claude Code (.md)

---
allowed-tools: Bash(git add:*), Bash(git commit:*)
argument-hint: [message]
description: Create a git commit with a descriptive message
---

Create a git commit with the following message: $ARGUMENTS

GitHub Copilot (.prompt.md)

---
mode: 'agent'
tools: ['terminal', 'terminal']
description: 'Create a git commit with a descriptive message'
---

Create a git commit with the following message: ${args}

Gemini CLI (.toml)

description = "Create a git commit with a descriptive message"

prompt = """
Create a git commit with the following message: {{args}}
"""

Bundled Commands

The toolkit includes pre-built commands for various development tasks, such as code review, project specification, and implementation guidance.

Run ait convert --dry-run to see a list of all available bundled commands.

MCP Server Configuration

The toolkit includes pre-configured MCP servers for popular services:

• GitHub: Repository management and operations
• Linear: Issue tracking and project management
• Notion: Notion API integration
• Context7: Enhanced context management
• Figma: Design file management
• Filesystem: Local file and directory operations
• PostgreSQL: Database operations
• SQLite: SQLite database management
• Puppeteer: Browser automation and web scraping
• Slack: Workspace integration

Each server is automatically configured with the appropriate transport method (Docker, SSE, HTTP, or npm command) based on the target client.

License

MIT © Quy Doan