My MCP Server

My MCP Server

My MCP Server is a Python template for building a Model Context Protocol (MCP) server. It uses FastMCP to expose tools via JSON-RPC over stdio (and can be extended to HTTP), enabling secure and scalable context and tool invocation for language models.

Project Structure

├── src/ Source code (package: my_mcp_server) │ ├── server.py Entry point and tool registrations │ ├── main.py Console script entry │ ├── config/ Configuration modules │ ├── context/ Context providers for MCP │ ├── models/ Pydantic models and schemas │ ├── tools/ Custom tool implementations │ └── utils/ Helper utilities ├── tests/ Unit, integration, and performance tests ├── Taskfile.yaml Common development and CI tasks ├── pyproject.toml Project metadata and build configuration ├── requirements.txt Runtime dependencies └── docs/ Contribution & architecture guides

Prerequisites

• Python >=3.13
• Git
• (Optional) Taskfile task runner

Setup

Create and activate virtual environment

task create-venv source .venv/bin/activate

Install runtime dependencies

task install-requirements

(Optional) Install development dependencies

task install-dev

Taskfile Tasks

Run task --list to see all available tasks. Common tasks include:

• default: List available tasks
• create-venv: Create a Python virtual environment in .venv
• activate-venv: Show activation commands for the venv
• clean-venv: Remove the .venv directory
• install-requirements: Install runtime dependencies (pip install -r requirements.txt)
• install-dev: Install the project in editable mode with dev dependencies (pip install -e ".[dev]")
• flake8: Run flake8 over src/ and tests/
• mypy: Run mypy over src/ and tests/
• black: Format code with black
• format: Alias for black (formats all Python files)
• lint: Run both flake8 and mypy
• test: Run all tests (unit, integration, performance) with pytest
• test-unit: Run unit tests only
• test-integration: Run integration tests only
• test-performance: Run performance tests only
• coverage: Generate coverage report (HTML + terminal) with pytest-cov
• check: Run format, lint, and test sequentially
• run: Start the console-based MCP server (python -m my_mcp_server)
• dev: Alias for run (start server in development mode)
• clean: Clean Python build artifacts and caches
• build: Build distribution packages (sdist & wheel) using Hatchling

Development & Usage

After setup, you can start the server in stdio mode:

Or directly:

Testing & Linting

• Run all tests: task test
• Run linters: task lint
• Format code: task format

Build & Release

• Build distributions: task build
• Upload to PyPI: twine upload dist/*

Configuration

Environment variables can be managed via a .env file (loaded by python-dotenv/pydantic-settings).

Contributing

See docs/CONTRIBUTING.md for guidelines on contributing and architecture details.

License

MIT License