Skip to content
Agent Search Engine.

Record · colab-mcpMCP serverCosts not checkedRecord checked Sep 23, 2026

Colab MCP

Google Colab's MCP server that bridges a local agent to a Colab notebook session in the browser.

Sources and review

Checked against public documentation on . Suitability guidance is our editorial judgment; this is not a performance benchmark or an endorsement.

Best for

Letting a local coding agent use Colab's hosted runtime, including its accelerators, from your own terminal.

Avoid if

Your MCP client runs remotely or lacks tools/list_changed support; both are requirements.

About Colab MCP

Colab MCP, from the Google Colab team, connects a locally running agent to a Colab session open in your browser, so the agent can work in that notebook. It requires a client that runs on your device and supports tool-list change notifications, such as Gemini CLI, Claude Code or Windsurf.

Colab MCP is listed in this index written primarily in Python, with 1.4k stars on GitHub. Its latest recorded repository push was in June 2026; that is not an ASE editorial review date.

Before connecting: access and data

This record's specific permissions have not been reviewed by ASE. Read the publisher's documentation before copying setup or connecting.

  • List the accounts, files and services the enabled tools can read, change or execute. Check your actual credentials and discovered tools.
  • Use narrowly scoped credentials and read-only mode where supported. Review proposed writes before approving them; a generated configuration does not apply these limits for you.
  • Local launch is not local-only data processing. External services receive requests, and your client may pass tool results to its model provider. Check both destinations before sharing private data.
  • Keep keys and browser session files out of shared config, repositories and screenshots. Know how to disconnect the server and separately revoke its credentials or account sessions.

Read the setup guide →

Add Colab MCP to your MCP client

Generated from the launch command in the project's README. Your client starts the server on demand.

Claude CodeRun in your terminal
claude mcp add colab-mcp -- uvx git+https://github.com/googlecolab/colab-mcp
Cursor~/.cursor/mcp.json or .cursor/mcp.json
{
  "mcpServers": {
    "colab-mcp": {
      "command": "uvx",
      "args": ["git+https://github.com/googlecolab/colab-mcp"]
    }
  }
}
VS Code.vscode/mcp.json
{
  "servers": {
    "colab-mcp": {
      "type": "stdio",
      "command": "uvx",
      "args": ["git+https://github.com/googlecolab/colab-mcp"]
    }
  }
}
Claude Desktopclaude_desktop_config.json
{
  "mcpServers": {
    "colab-mcp": {
      "command": "uvx",
      "args": ["git+https://github.com/googlecolab/colab-mcp"]
    }
  }
}
Codex~/.codex/config.toml
[mcp_servers.colab-mcp]
command = "uvx"
args = ["git+https://github.com/googlecolab/colab-mcp"]

Check the project's README for required accounts, keys and permissions before connecting. Setup guide for each client →

For agent buildersMake your product part of the discovery.Explore advertising