MCP Server (mcp-tools)

What is MCP Server (mcp-tools)?

MCP Server (mcp-tools) is a general MCP server that provides tools for AI agents, communicating via stdin/stdout using the Model Context Protocol. It currently offers a set of markdown link checking tools.

How to use MCP Server (mcp-tools)?

Navigate to the project directory and run make install-dev to install dependencies. Use make run to start the server manually. For Cursor integration, configure ~/.cursor/mcp.json with the absolute path to the project’s virtual environment and use the python -m mcp_server.server command.

Key features of MCP Server (mcp-tools)

• Checks HTTP/HTTPS links in Markdown files
• Supports single file, file list, directory, and project-wide scans
• Respects .gitignore when scanning the entire project
• Generates consolidated text reports of link status
• Built with Python and managed via uv and a Makefile

Use cases of MCP Server (mcp-tools)

• Validate documentation links before publishing
• Ensure all markdown files in a repository have no broken links
• Automate link checking in development workflows
• Integrate link checking into AI agent toolchains
• Quickly check links across multiple files or directories

FAQ from MCP Server (mcp-tools)

What tools does MCP Server (mcp-tools) provide?

It provides four tools: check_markdown_link_file, check_markdown_link_files, check_markdown_link_directory, and check_markdown_links_project. All check HTTP/HTTPS links in Markdown files.

How do I install and run the server?

Run make install-dev to create a virtual environment and install dependencies. Run make run to start the server. The server listens on stdin/stdout and expects properly framed JSON-RPC messages.

Can I integrate this server with Cursor?

Yes. Add a configuration entry in ~/.cursor/mcp.json using the absolute path to the project’s .venv/bin/python and the -m mcp_server.server arguments. Restart Cursor after making changes.

How do I add new tools?

Define the tool’s schema in handle_list_tools in src/mcp_server/server.py and implement its logic under a new elif block in handle_call_tool. Add helper functions in src/mcp_server/tools/ and unit tests in tests/.

What are the runtime requirements?

Python 3.x (managed via uv), with dependencies installed via make install-dev. The server runs as a script using the project’s virtual environment.