Discovered a neat distribution angle while building AI tooling.
Prepend tomcp.org/ to any URL and it becomes an MCP server. Cursor, Claude, any MCP client can now read that page directly.
Why this matters for growth:
Documentation-heavy products can now be piped straight into AI coding assistants. Instead of users bouncing between your docs and their IDE, your content lives in their workflow.
If you're building dev tools, consider adding MCP configs to your docs. It's becoming table stakes for developer experience—and it's a subtle retention play. Your docs become the default context, not ChatGPT's hallucinated version.
The proxy strips ads/nav and converts to clean Markdown, so AI actually understands the structure (and uses fewer tokens, which users notice).
Open source: https://github.com/Ami3466/tomcp
Anyone else experimenting with MCP as a distribution channel?
Turning docs into live MCP context is powerful. Especially for reducing hallucinations.
Have you seen any edge cases with auth-gated docs or rate limits once IDEs start polling this heavily?