Agent MCP Studio → Integrations → Goose
CLI Agent · Block (Square)

Supercharge Goose CLI with Browser-Based MCP Tools

Goose (by Block/Square) is one of the most MCP-forward CLI AI agents available. It's built specifically for extending via MCP extensions (called "toolkits"). Connect Agent MCP Studio to give your Goose sessions access to browser-built Python, SQL, and API tools directly from your terminal.

Share:

What is Goose?

Goose is an open-source, terminal-based AI agent built by Block (formerly Square). It was designed from the ground up to work with MCP extensions — you install "toolkits" (MCP servers) and Goose can call them during autonomous task execution. Unlike IDE-based agents, Goose runs entirely in your terminal and can be scripted and automated.

Prerequisites

Step-by-Step Setup

  1. Get your relay URL and download bridge.js

    Open the studio → Settings → MCP Relay Bridge → copy URL → click Connect. Then:

    curl -O https://agentmcp.studio/bridge.js && npm install ws

  2. Add the MCP server to Goose's config

    Edit Goose's config file at ~/.config/goose/config.yaml:

    extensions:
  agent-mcp-studio:
    name: Agent MCP Studio
    type: stdio
    enabled: true
    cmd: node
    args:
      - /path/to/bridge.js
      - wss://agentmcp.studio/api/relay/YOUR-UUID

  3. Start a Goose session

    goose session

    Goose will launch bridge.js automatically. In the session, your browser tools appear alongside Goose's built-in tools. Ask Goose to use them: "Use the fetch_webpage tool to check the current status of our API."

Goose is great for automation scripts

Because Goose runs in the terminal, you can script it with goose run --task "..." and pipe it into CI/CD pipelines, cron jobs, or shell scripts — all with your browser MCP tools available.

Frequently Asked Questions

Yes — Goose (by Block/Square) is built around MCP. You configure toolkits in your Goose config, and any MCP-compatible server (like Agent MCP Studio via bridge.js) becomes a full-featured Goose toolkit.

Goose is an open-source, extensible AI agent by Block (the company behind Square). It runs as a CLI or desktop app and uses MCP as its primary tool extension protocol — making it one of the most MCP-native agents available.

Edit your Goose config file (usually at ~/.config/goose/profiles.yaml). Add a toolkit entry with type mcp and the bridge.js command + relay URL as arguments. Run goose session start to start with your tools active.

Goose supports multiple LLM providers including Anthropic Claude, OpenAI, and local models. MCP tool calling works with models that support function calling — Claude 3.5 Sonnet is the recommended model for Goose.

Yes — Goose runs in CLI mode (goose run) for scripting and automation. Your Agent MCP Studio tools are available in both interactive sessions and headless/scripted Goose runs.

Related Integrations

Claude DesktopDesktop client RaycastmacOS launcher All integrations →40+ platforms