wtf is mcp-radar?

MCP Radar is a directory website that automatically tracks new and trending MCP servers. MCP stands for Model Context Protocol, a standard that lets AI assistants connect to external tools and data sources. This project scans GitHub every day looking for newly created repositories that implement MCP servers, then ranks them by momentum, meaning how quickly they are gaining stars and forks relative to how new they are. The whole thing runs without any servers to maintain and without any ongoing cost. A scheduled GitHub Actions workflow runs once a day, queries the GitHub Search API for repositories tagged with MCP related topics, filters out forks, archived repos, and anything with too few stars, then calculates a momentum score for what remains. The results are saved as a JSON data file and a readable markdown digest listing the top twenty new servers of the week, and both files are committed back into the repository automatically so the git history becomes an archive of every past snapshot. The output also powers a live website at mcp.liqiwa.com, which includes a full directory, a weekly archive, and an RSS feed, all rebuilt automatically whenever the data updates. Someone wanting to see what MCP servers are gaining traction can simply visit the site rather than running anything themselves. If you want to run the scanner yourself, it only needs the Python standard library, no extra installations required. It works without a GitHub token, though requests will be rate limited, or you can set a token for higher limits. Anyone forking the project who wants the automated daily workflow to keep working will need to turn on write permissions for GitHub Actions in the repository settings. The project is released under the MIT license, a permissive license that allows free use, modification, and distribution as long as the original copyright notice is kept.