CLI tool that generates a complete, buildable MCP server project from a single YAML config file. Outputs Python project with tool stubs, pyproject.toml, tests, and init.sh. Supports stdio transport.

Quick Start • Features • Examples • Contributing

What is this?

MCPick is a command-line utility that transforms a simple YAML description into a ready-to-build MCP server project. It eliminates boilerplate so solo AI developers can focus on implementing tools instead of setting up projects.

Example usage:

$ mcpick generate examples/calculator.yaml
✔ Generated project in ./calculator-server

Problem

Setting up a new MCP server requires significant boilerplate: project structure, SDK wiring, transport configuration, and test scaffolding. This repetitive setup slows down developers who want to ship MCP tools quickly.

Features

Feature	Description
YAML Configuration Parsing and Validation	Reads a YAML file, validates required fields, and provides clear error messages for missing or malformed data.
MCP Server Code Generation	Produces a full Python server with stdio transport, tool handler stubs, and proper async/await patterns.
Project Structure and Dependency Management	Creates pyproject.toml, test suite, README, init.sh, and .gitignore following Python packaging best practices.
Interactive Tool Definition Wizard	Launches a prompt‑driven interface to build a YAML config when no file is supplied.
Template Customization and Extension	Allows users to override Jinja2 templates or add custom ones for specialized project layouts.
Built‑in Testing Support	Generates pytest fixtures and test cases for each tool so the generated project can be validated immediately.

Quick Start

1. Clone the repository:

   git clone https://github.com/m2ai-portfolio/mcpick.git
   cd mcpick
   

2. Install the tool in editable mode:

   pip install -e .
   

3. Generate a server from the example config:

   mcpick generate examples/calculator.yaml
   

4. Enter the generated directory and set up the environment:

   cd calculator-server
   chmod +x init.sh
   ./init.sh
   source .venv/bin/activate
   

5. Run the test suite to verify everything works:

   pytest
   

Examples

Basic generation
Create a simple calculator MCP server:

$ mcpick generate examples/calculator.yaml
✔ Generated project in ./calculator-server
$ ls calculator-server
README.md  init.sh  pyproject.toml  src  tests

Interactive wizard
Build a config via prompts and immediately generate the project:

$ mcpick generate --interactive
? Server name: weather-tools
? Description: A set of tools for fetching weather data
? Author name: Jane Doe
? Author email: jane@example.com
? Add a tool? (y/n) y
? Tool name: get_current
? Tool description: Returns current temperature for a location
? Parameter name: location
? Parameter type: string
? Required? (y/n) y
? Add another parameter? (y/n) n
? Add another tool? (y/n) n
✔ Config written to weather-tools.yaml
? Generate project now? (y/n) y
✔ Generated project in ./weather-tools

Custom template usage
Generate a server using a user‑defined template directory:

$ mcpick generate examples/database_tools.yaml --template-dir ./examples/custom_templates/database
✔ Generated project in ./database-server
$ grep -r "custom_logic" database-server/
src/database-server/server.py:    # Custom logic from template

File Structure

MCPick/
  src/                    # Core source code
    mcpick/               # MCPick package
      __init__.py
      cli.py              # Click‑based command line interface
      config.py           # YAML parsing and validation models
      generator.py        # Jinja2 template rendering and file writing
      templates/          # Default Jinja2 templates for server, tests, etc.
  tests/                  # Test suite for MCPick itself
    test_cli.py
    test_config.py
    test_generator.py
  examples/               # Sample configs and custom templates
    calculator.yaml
    database_tools.yaml
    weather_api.yaml
    custom_templates/
      api_wrapper/
        server.py.j2
      database/
        server.py.j2
      README.md
  screenshots/            # Demonstration output files (omitted for brevity)
  pyproject.toml
  README.md
  LICENSE

Tech Stack

Technology	Purpose
Python 3.11+	Core language for the CLI and generated projects
Click	Builds the command‑line interface with rich argument handling
Jinja2	Templating engine for generating Python files and project structure
PyYAML	Parses YAML configuration files
Pydantic	Validates configuration data and provides data models
pytest	Testing framework for both MCPick and generated projects
mcp	Model Context Protocol SDK used in generated servers
pathlib	Modern, cross‑platform file system operations
rich	Enhanced CLI output with progress indicators and formatting

Contributing

Fork the repo, create a feature branch, make changes, run pytest, and submit a pull request. Please keep commits atomic and update tests when adding new functionality.

License

MIT

Author

Matthew Snow -- M2AI | @m2ai-portfolio