MCP Code Checker

What is MCP Code Checker?

MCP Code Checker is a Model Context Protocol (MCP) server that provides code quality checking operations for Python projects. It enables AI assistants like Claude, VSCode with GitHub Copilot, or other MCP-compatible clients to run pylint, pytest, and mypy checks within a specified project directory.

How to use MCP Code Checker?

Install the server via pip (pip install git+https://github.com/MarcusJellinghaus/mcp-tools-py.git), then run it with the required --project-dir argument. Configure it in your MCP client (e.g., Claude Desktop) using the mcpServers JSON block, pointing either --python-executable or --venv-path to the environment where pytest, pylint, and mypy are installed.

Key features of MCP Code Checker

• Run pylint checks for code quality issues
• Run pytest to identify failing tests
• Run mypy type checking with strict mode
• Auto-detection of source and test directories from pyproject.toml
• Structured JSON logging with full call tracking
• Security: only defined tools can be executed, scoped to project directory

Use cases of MCP Code Checker

• Automatically check Python code quality during AI-assisted development
• Run tests and type checks without giving the AI assistant direct shell access
• Integrate code quality checks into MCP-compatible IDEs like VSCode or Cursor
• Obtain structured, context-aware feedback for LLMs to act on

FAQ from MCP Code Checker

What is the difference between MCP Code Checker and giving an AI assistant bash access?

MCP Code Checker provides a more focused, secure approach: only pylint, pytest, and mypy can be executed, all operations are scoped to the specified project_dir, and results are formatted and size-limited to reduce context load. The server also logs every tool call with parameters, timing, and results.

What dependencies does MCP Code Checker require?

The server requires pytest, pylint, and mypy to be installed in the same Python environment. The --python-executable or --venv-path must point to that environment (typically the tool’s own venv, not the project’s runtime venv).

Does MCP Code Checker support non-Python projects?

No. This server covers Python projects only. Support for other languages is planned for separate, dedicated MCP servers.

Where are logs stored?

Logs can be written to the console by default or to a structured JSON file using --log-file. If not specified, a timestamped log file is created in project_dir/logs/mcp_tools_py_{timestamp}.log. Use --console-only to disable file logging.

How do I configure the server for my project?

The only required argument is --project-dir, pointing to the base directory of your project. Optional arguments include --python-executable, --venv-path, --test-folder, --log-level, and --log-file. Tool-specific parameters (like target_directories) can be passed at runtime.