The Pantavisor documentation is now available as a Model Context Protocol (MCP) server at https://mcp.docs.pantavisor.io/mcp. Add it to Claude or Claude Code and the model can search docs.pantavisor.io and read the pages it finds, so answers about state.json, pv-xconnect, Yocto builds or pvr commands come from the current docs instead of from memory. It is public, needs no account or sign-in, and is listed in the Claude connector directory under Pantavisor Docs.
What it does
The server exposes two tools:
- Search docs takes a query like
xconnect networkingorstate.json schemaand returns matching page titles, URLs and a snippet of the matching content. - Fetch docs page returns the full content of one page as Markdown, so Claude can read a guide end to end before answering.
The index is rebuilt from the published site in the background, so whatever lands on docs.pantavisor.io shows up in Claude shortly after. Claude also cites the page URLs it used, so you can jump straight to the source.
Add it to Claude (web and desktop)
The quickest route is the connector directory. Open Customize > Connectors at https://claude.ai/customize/connectors, search for Pantavisor Docs and click Connect. That is it. There is no sign-in step.
If you would rather add it by URL, or are on a plan where the directory entry is not shown:
- Go to Customize > Connectors and click Add custom connector.
- Name it
Pantavisor Docsand enter the URLhttps://mcp.docs.pantavisor.io/mcp. - Leave authentication on No sign-in and click Add.
On Team and Enterprise plans an organization owner adds the connector once under Organization settings > Connectors (https://claude.ai/admin-settings/connectors), and members then enable it from their own Customize > Connectors page.
Once connected, make sure the connector is toggled on in the tools menu of a chat, then ask something like “How do I configure a device to use pv-xconnect?” and watch Claude search the docs before it answers.
Add it to Claude Code
One command:
claude mcp add --transport http pantavisor-docs https://mcp.docs.pantavisor.io/mcp
By default that registers the server for the current project only. Add --scope user to have it available in every project on your machine, or --scope project to write it into a .mcp.json in the repository so your whole team gets it when they check out the code:
claude mcp add --transport http --scope project pantavisor-docs https://mcp.docs.pantavisor.io/mcp
The resulting .mcp.json looks like this, and you can also just commit it by hand:
{
"mcpServers": {
"pantavisor-docs": {
"type": "http",
"url": "https://mcp.docs.pantavisor.io/mcp"
}
}
}
Check that it is connected with claude mcp list, or type /mcp inside a Claude Code session. From then on Claude Code will search the Pantavisor docs on its own when a task touches Pantavisor, pvr, meta-pantavisor or Pantacor Hub, for example while it writes a state.json or a container run.json for you.
Other MCP clients
Anything that speaks MCP over Streamable HTTP can use the same endpoint. Point your client at https://mcp.docs.pantavisor.io/mcp with no authentication.
Under the hood
The server is small and open source. It lives in the mcp-server/ directory of the docs repository at GitHub - pantavisor/docs.pantavisor: main for pantavisor released docs · GitHub, built on the official TypeScript MCP SDK with a MiniSearch index of the site, and runs stateless behind the same infrastructure as the docs. Rate limiting is per client and generous for interactive use.
Try it, and tell us below what you asked and whether the answers held up. If the search misses a page it should have found, or a doc is missing entirely, that is a bug in the docs, and issues and pull requests on the repository are very welcome.