> ## Documentation Index
> Fetch the complete documentation index at: https://docs.avaluma.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect this documentation to your AI assistant via MCP

> Query the Avaluma documentation directly from Claude Code, OpenAI Codex, Gemini CLI, and Google Antigravity using the hosted Model Context Protocol (MCP) server.

This documentation is automatically available as a [Model Context Protocol (MCP)](https://modelcontextprotocol.io) server. Once you connect your AI coding assistant to it, the assistant can **search and read these docs on demand** — so its answers about Avaluma stay grounded in the current documentation instead of guessing.

## The MCP endpoint

The hosted MCP server is served from the documentation domain with `/mcp` appended:

```text theme={null}
https://docs.avaluma.ai/mcp
```

There is nothing to deploy or maintain — Mintlify hosts the server and keeps it in sync with the live docs. It exposes read-only **search** and **navigation** tools over every published page.

<Note>
  The hosted MCP server is included in Mintlify's free plan and is generated automatically for every documentation site — no extra configuration on our side.
</Note>

## Connect your assistant

All four clients connect over the same remote HTTP endpoint. Pick your tool below.

<Tabs>
  <Tab title="Claude Code">
    Add the server with the Claude Code CLI:

    ```bash theme={null}
    claude mcp add --transport http avaluma-docs https://docs.avaluma.ai/mcp
    ```

    Then check it is connected:

    ```bash theme={null}
    claude mcp list
    ```

    The `avaluma-docs` tools become available in your next Claude Code session.
  </Tab>

  <Tab title="OpenAI Codex">
    Add the server with the Codex CLI:

    ```bash theme={null}
    codex mcp add avaluma-docs --url https://docs.avaluma.ai/mcp
    ```

    This writes the server to `~/.codex/config.toml`. The equivalent manual entry is:

    ```toml ~/.codex/config.toml theme={null}
    [mcp_servers.avaluma-docs]
    url = "https://docs.avaluma.ai/mcp"
    ```

    <Note>
      The exact `codex mcp add` flags can change between Codex versions. If the command differs in your version, add the `[mcp_servers.avaluma-docs]` block to `~/.codex/config.toml` directly.
    </Note>
  </Tab>

  <Tab title="Gemini CLI">
    Add the server under `mcpServers` in `~/.gemini/settings.json`. For a remote HTTP server, use the `httpUrl` field:

    ```json ~/.gemini/settings.json theme={null}
    {
      "mcpServers": {
        "avaluma-docs": {
          "httpUrl": "https://docs.avaluma.ai/mcp"
        }
      }
    }
    ```

    Restart the Gemini CLI so it picks up the new server.
  </Tab>

  <Tab title="Google Antigravity">
    Add the server under `mcpServers` in `~/.gemini/config/mcp_config.json`. Antigravity uses the `serverUrl` field for remote servers, and the file is shared between the Antigravity CLI and IDE:

    ```json ~/.gemini/config/mcp_config.json theme={null}
    {
      "mcpServers": {
        "avaluma-docs": {
          "serverUrl": "https://docs.avaluma.ai/mcp"
        }
      }
    }
    ```

    <Note>
      Use `serverUrl` (not the older `httpUrl`), keep the JSON comment-free, and reload Antigravity afterwards.
    </Note>
  </Tab>
</Tabs>

## Verify it works

After connecting, ask your assistant a question that can only be answered from these docs, for example:

> Using the Avaluma docs MCP server, how do I mute the microphone of an embedded avatar?

If the connection works, the assistant calls the docs search tool and answers with content from the [Messaging](/en/agents/embedding#messaging) page.

## Troubleshooting

<AccordionGroup>
  <Accordion title="The assistant doesn't see the new tools">
    Restart the client after adding the server — most assistants only load MCP servers at startup. For Claude Code, run `claude mcp list` to confirm the server status is `connected`.
  </Accordion>

  <Accordion title="Connection or transport errors">
    Make sure you used the HTTP transport (`--transport http` / `httpUrl` / `serverUrl`) and the exact URL `https://docs.avaluma.ai/mcp`. SSE- or stdio-style entries will not work against the hosted endpoint.
  </Accordion>

  <Accordion title="JSON config won't load">
    Validate the JSON — a trailing comma or an inline comment will stop Gemini CLI and Antigravity from reading the file. Antigravity in particular does not allow comments in `mcp_config.json`.
  </Accordion>
</AccordionGroup>