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

> ## Agent Instructions
> To query the Token Terminal data catalog, read https://tokenterminal.com/docs/catalog/agents-manual.md first. It is the whole catalog as one page: table naming grammar, key columns, partition and cluster rules, units, additivity, and the tables that are documented but not served yet.
> Never query a catalog table on a time bound alone. Also filter its cluster key, which you read from INFORMATION_SCHEMA.COLUMNS; an empty result means the object is a view, whose pruning contract is on its page. Compute is billed to the caller's own Google Cloud project.

# MCP Server

> Query Token Terminal data from any MCP-compatible AI client.

The Token Terminal MCP server gives AI clients structured access to our onchain dataset — projects, products, chains, market sectors, financial metrics, and methodologies — through the [Model Context Protocol](https://modelcontextprotocol.io/).

It is a centrally hosted remote server compatible with Claude, ChatGPT, Cursor, and any other MCP-compatible client.

## Availability

The MCP server is available on every Token Terminal plan. The free tier includes access to the `research` tool with limited usage; paid plans raise the usage limits.

For higher limits or enterprise options, contact our team at [sales@tokenterminal.xyz](mailto:sales@tokenterminal.xyz).

## Setup guides

<CardGroup cols={2}>
  <Card title="Claude Code" icon="terminal" href="/docs/mcp/claude-code">
    Set up MCP in Claude Code via the CLI.
  </Card>

  <Card title="Claude Desktop" icon="message-bot" href="/docs/mcp/claude-desktop">
    Connect Token Terminal from the Claude directory, or add it as a custom connector.
  </Card>

  <Card title="ChatGPT" icon="comments" href="/docs/mcp/chatgpt">
    Install the Token Terminal app or add the MCP server as a custom connector.
  </Card>

  <Card title="Codex CLI" icon="terminal" href="/docs/mcp/codex">
    Set up MCP in Codex CLI.
  </Card>

  <Card title="Cursor" icon="i-cursor" href="/docs/mcp/cursor">
    Set up MCP in the Cursor IDE.
  </Card>

  <Card title="Gemini CLI" icon="terminal" href="/docs/mcp/gemini-cli">
    Set up MCP in Gemini CLI.
  </Card>

  <Card title="OpenCode" icon="terminal" href="/docs/mcp/opencode">
    Set up MCP in OpenCode.
  </Card>

  <Card title="Microsoft Copilot Studio" icon="microsoft" href="/docs/mcp/copilot-studio">
    Set up MCP in Microsoft Copilot Studio.
  </Card>

  <Card title="Other clients" icon="plug" href="/docs/mcp/other-clients">
    Use MCP with any compatible client.
  </Card>
</CardGroup>

## How it works

The server is hosted at `https://mcp.tokenterminal.com/mcp` and implements the [authenticated remote MCP specification](https://modelcontextprotocol.io/specification) over the Streamable HTTP transport. Clients connect directly — there is no local proxy or companion process to run.

Authentication is handled by **OAuth 2.1 with PKCE**, so no API keys or long-lived secrets are exchanged with your client. The first connection opens a browser window in which you sign in to Token Terminal and select the user account and organization the session will operate under. Your client stores the resulting access token and refreshes it transparently.

## Available tool

The server exposes a single tool to connected clients.

<ResponseField name="research">
  AI-powered analyst for Token Terminal data across projects, chains, ecosystems, assets, products, metrics, methodologies, and mixed analytical questions.
</ResponseField>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.