Docs for AI agents
Every public Revdoku docs site is also made for AI agents. An AI agent can search your docs, read each page as Markdown and list all pages. It needs no sign-in and no web scraper.
What each public docs site gives AI agents
| Resource | Address | What it does |
|---|---|---|
| MCP server | https://<site>/mcp |
Read-only tools search_docs, read_page and list_pages. Streamable HTTP, no sign-in. |
| MCP discovery | https://<site>/.well-known/mcp |
Tells AI apps where the MCP server is. |
llms.txt |
https://<site>/llms.txt |
Every page in the order of the navigation, with a summary and a link to its Markdown copy. |
llms-full.txt |
https://<site>/llms-full.txt |
Every page in one file. |
| Markdown pages | https://<site>/<page>/index.md |
One page as Markdown. A request with the header Accept: text/markdown also gets Markdown. |
| Search | https://<site>/_revdoku/search?q=webhooks |
JSON results with the page, the matching section and a snippet. |
| Sitemap | https://<site>/sitemap.xml |
Every page, for a site that allows search indexing. |
MCP is a standard way for AI apps to use outside tools. Each page of a docs site also names these addresses in its HTML head and in its “For AI agents” footer.
Revdoku makes these files each time it publishes the site. A site that you protect with a password or an email code has none of them.
Connect an AI app to a docs site
Give the AI app the MCP address of the site. The examples use https://docs.example.com/mcp.
| AI app | How to add the docs site |
|---|---|
| Claude Code | claude mcp add --transport http acme-docs https://docs.example.com/mcp |
| Cursor | In .cursor/mcp.json: {"mcpServers": {"acme-docs": {"url": "https://docs.example.com/mcp"}}} |
| Claude | Add a custom connector with the URL and No sign-in. See Claude. |
| ChatGPT, Codex, Gemini and others | See the guide for your app on Connect AI apps. |
Then ask a question that the docs answer:
Use the Acme docs. How do I list orders page by page?
Name the page that you used.The AI app calls search_docs, reads the best page with read_page, and answers from that page.
Why AI agents can read the docs
- Cloudflare does not block or challenge AI agents and crawlers on Revdoku docs sites.
- The MCP server and the search answer from the published files of the site, so they always match the pages that people see.
- Each network can send a limited number of agent requests to a site in a minute. This keeps the site fast for everyone.
See which AI agents read your docs
The analytics of each docs site show:
- The AI agents and assistants that read the docs, and their requests: MCP calls, searches,
llms.txtand Markdown reads. - Bots and crawlers, and the countries of visitors.
- The searches in the site’s search box and the questions that people ask with Ask AI.
Use the searches and the questions to find pages that are missing or hard to find.
Let AI agents manage your docs
The MCP server of a docs site only reads it. To let an AI app publish your sites and read your support email, connect it to your Revdoku account at https://mcp.revdoku.com. That server uses sign-in, and you choose the buckets and the access level. See the MCP guide, the CLI and the API.
Docs from GitHub,
for people and AI agents
Connect a repository. Revdoku publishes it as a docs site with search, Ask AI, an API console and an MCP server.
- Publishes from GitHub on every push
- Works with Mintlify repositories
- API console from your OpenAPI file
- MCP server and llms.txt for AI agents
- Ask AI and a support mailbox
- API, MCP and CLI access