Python Testing Tools MCP Server

What is Python Testing Tools MCP Server?

It is an MCP server that generates unit tests, fuzz inputs, and coverage-focused tests for Python files using Google Gemini AI. It is designed for developers who want automated, AI-powered test generation integrated with Claude Code.

How to use Python Testing Tools MCP Server?

Install dependencies from requirements.txt, set the GEMINI_API_KEY environment variable, and run python mcp_server.py. For integration with Claude Code, use FastMCP to install the server, then configure it in .claude.json. After starting Claude Code, use commands like "create unit tests for @file.py", "generate comprehensive coverage tests for @file.py", or "fuzz test the function_name in @file.py".

Key features of Python Testing Tools MCP Server

• AI-powered unit test generation with 4–6 test cases per function
• Intelligent fuzz testing with 20 diverse inputs including edge cases
• Advanced AST analysis for maximum code coverage
• Built-in coverage measurement using coverage.py
• Proper unittest framework with exception testing and safe input parsing
• Automatic import resolution and formatted test file creation

Use cases of Python Testing Tools MCP Server

• Automatically generate unit tests for all functions in a Python file
• Perform intelligent fuzz testing to find crashes and edge cases
• Create comprehensive coverage tests to achieve high branch and path coverage
• Integrate AI-driven test generation into a Claude Code workflow

FAQ from Python Testing Tools MCP Server

What AI model does the server use?

It uses Google Gemini, defaulting to the gemini-2.5-flash model. You can set the GEMINI_MODEL environment variable to choose a different Gemini model.

How do I set up authentication?

Set the GEMINI_API_KEY environment variable with your Google Gemini API key. This is required for AI-powered test generation.

What are the runtime dependencies?

Python 3 with a virtual environment, dependencies from requirements.txt, and the BAML configuration in baml_src/main.baml. The server also requires coverage.py for coverage reports.

What transport protocol does the server use?

It uses the Model Context Protocol (MCP) over stdio. It can be run standalone with python mcp_server.py or integrated as an MCP server in Claude Code using FastMCP.

Where does the server output test files?

Test files are created in the same folder as the input Python file (the demo folder in the examples). The server reads files from the local filesystem based on the provided file path.