Introducing the MDN MCP server
5 points by HugoDaniel
5 points by HugoDaniel
I wonder how much different this is to just having the AI curl the MDN doc URL. I guess the main benefit would be that it doesn't have to load in all of the HTML bloat?
I wonder if it would just be better if MDN just made a good API with search functionality and the ability to get the raw text of the page in markdown format for easy consumption by AI.
I've run git clone against https://github.com/mdn/browser-compat-data in the past so my coding agents can consult those docs directly
Add a clone of https://github.com/mdn/content and I think you get everything that might otherwise be available thorough this MCP.
I really like git clone for this, I think it may be the single most efficient way to grab a sizable chunk of context for a coding agent - especially since you can then grab updates with git pull.