DevCoffee MCP Tools

DevCoffee MCP Tools is a Production-Grade Model Context Protocol (MCP) server implementation and an automated injector for the devcoffee ecosystem. It connects AI agents seamlessly to DevCoffee's local infrastructure, enabling them to explore databases, run queries, and utilize specialized skills.

🚀 Features

• MCP Server (mcp): A robust, standard-compliant MCP Server running over stdio. It exposes tools to introspect and query the devcoffee-product PostgreSQL database safely.
• Automated Setup (setup): An interactive CLI tool to inject MCP server configurations into various AI agents (Claude Code, Codex CLI, Gemini, and Antigravity).
• Skill Injection (skills): Installs bundled, production-grade instructions (SKILL.md) directly into your AI agents' working directories to guide them on how to effectively use DevCoffee's infrastructure.
• Strictly Typed: Built with TypeScript, validated by Zod, and configured with aggressive ESLint rules to guarantee stability and type safety.

📦 Installation

Ensure you have Node.js installed, then clone the repository and install dependencies:

npm install
npm run build

🛠️ Usage

This project exposes an executable CLI. You can run it locally using npx or npm run start.

1. Start the MCP Server

Typically, this command is executed automatically by AI agents, but you can test it directly:

npx @devcoffee/mcp-tools mcp

2. Setup MCP Configuration for Agents

Interactively configure the Database credentials and inject the server path into your AI assistants:

npx @devcoffee/mcp-tools setup

You can also bypass the prompt by specifying an agent: npx @devcoffee/mcp-tools setup claude

3. Inject DevCoffee Skills

Inject authoritative guidelines (mcp-guide, database-exploring) to teach AI agents how to interact with the system:

npx @devcoffee/mcp-tools skills

Supported targets include: Claude Code (.claude), Codex CLI (.codex), Gemini/Antigravity (.agents).

🗄️ Database Tools Available

The MCP Server exposes the following tools to the AI agents:

• list_tables: Retrieves a complete list of tables in the public schema.
• describe_table: Analyzes column data types, nullability, and extracts foreign-key relationships.
• run_sql_query: Executes read-only SQL queries (SELECT or WITH) securely against the PostgreSQL database.

⚙️ Environment Configuration

The server relies on the following environment variables to connect to the PostgreSQL database (these are automatically collected and injected into the target config during setup):

• DATABASES_DEFAULT_HOST
• DATABASES_DEFAULT_PORT
• DATABASES_DEFAULT_NAME
• DATABASES_DEFAULT_USER
• DATABASES_DEFAULT_PASSWORD

👨‍💻 Development

Start the development server with the MCP inspector:

npm run dev

Run Code Quality checks (Typecheck & Linting):

npm run check

We utilize path aliasing (@/*) via tsc-alias to ensure clean, readable imports without relative path nesting.

📄 License

ISC