Your docs as a tool for every AI agent
Switch on the MCP server of a published site, and agents in editors and chat apps can search and read your docs directly — always the version you published.
Three tools, no setup
The server lives at your site’s address followed by /mcp, and speaks the Streamable HTTP transport.
search
The same full-text search readers use, returning the best-matching pages with a snippet and their URL.
read_page
Any page as Markdown, straight from your published content, with its source URL to cite.
list_pages
Every page of the site by section, with descriptions — your llms.txt.
Connect a client
Add the URL as a remote MCP server in any client that supports Streamable HTTP. For example, in Claude Code:
claude mcp add --transport http acme-docs https://docs.example.com/mcpThe same content, the same rules
Only what you published
Agents see what llms.txt shows: published sections, without hidden pages or pages marked noindex.
Private sites stay private
An unlisted site’s server needs its share token in the URL, like every other link to the site.
Usage in Insights
See how many searches and page reads agents make, and which searches found nothing.
Learn moreFrequently asked questions
What is MCP?
The Model Context Protocol is an open standard that lets AI assistants call tools and read data from outside sources in a consistent way.
How do I turn it on?
In the site’s settings, under Customize → MCP server. Copy the URL shown there into your MCP client. The MCP server is part of the Business plan and above.
Which clients work with it?
Clients that support remote MCP servers over Streamable HTTP. The server needs no sign-in for public sites.
Does it use AI credits?
No. The server only searches and returns your pages; the agent that calls it does the thinking.
How can agents use my docs without it?
Through Markdown versions of every page, llms.txt and llms-full.txt, or the REST API with a personal token.
Give agents docs worth reading
Publish accurate, reviewed docs, and the MCP server serves them to every agent.