Skip to content

Connect your AI client

Every client runs the same command, uvx mcp-trino-python, and passes the connection settings as environment variables. Install uv first.

Plugin (recommended). Claude Code asks for the connection settings when you install it:

/plugin marketplace add alaturqua/mcp-trino-python
/plugin install trino@mcp-trino-python

Or add the server directly:

claude mcp add trino -e TRINO_HOST=localhost -e TRINO_USER=trino -- uvx mcp-trino-python

One-click extension (recommended).

  1. Download mcp-trino-python.mcpb from the latest release.
  2. Double-click it, or drag it into Settings → Extensions.
  3. Fill in your Trino host and user when asked.

Or edit claude_desktop_config.json (Settings → Developer → Edit Config):

{
  "mcpServers": {
    "trino": {
      "command": "uvx",
      "args": ["mcp-trino-python"],
      "env": { "TRINO_HOST": "localhost", "TRINO_PORT": "8080", "TRINO_USER": "trino" }
    }
  }
}

Create .vscode/mcp.json in your workspace, or run MCP: Add Server from the command palette:

{
  "servers": {
    "trino": {
      "type": "stdio",
      "command": "uvx",
      "args": ["mcp-trino-python"],
      "env": { "TRINO_HOST": "localhost", "TRINO_PORT": "8080", "TRINO_USER": "trino" }
    }
  }
}

Open Copilot Chat in Agent mode; the Trino tools appear in the tools list.

For the GitHub Copilot CLI, put the same entry under mcpServers in ~/.copilot/mcp-config.json.

Most clients use the mcpServers format, for example ~/.cursor/mcp.json:

{
  "mcpServers": {
    "trino": {
      "command": "uvx",
      "args": ["mcp-trino-python"],
      "env": { "TRINO_HOST": "localhost", "TRINO_PORT": "8080", "TRINO_USER": "trino" }
    }
  }
}

The server is also in the MCP Registry as io.github.alaturqua/mcp-trino-python, so clients that browse the registry can install it from there.

Clients that start the server themselves can run the container instead of uvx:

{
  "mcpServers": {
    "trino": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "-e", "TRINO_HOST", "-e", "TRINO_USER", "ghcr.io/alaturqua/mcp-trino-python"],
      "env": { "TRINO_HOST": "host.docker.internal", "TRINO_USER": "trino" }
    }
  }
}

host.docker.internal reaches a Trino running on your own machine.

Troubleshooting

uvx: command not found
Install uv, then restart your client so it picks up the new PATH.
The tools don't show up
Restart the client after changing its config. In Claude Code, /mcp shows the server's status and errors.
Connection refused or timeouts
Check TRINO_HOST and TRINO_PORT. Inside Docker, localhost is the container itself; use host.docker.internal or the Trino container's name.
Both catalog and schema must be specified
Mention the catalog and schema in your question, or set TRINO_CATALOG and TRINO_SCHEMA.