Goose

Runs per-instance MCP servers for Goose using Python virtual environments via uv, enabling seamless MCP integration.

other

11

GitHub Stars

other

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "jay4242-goose_mcp": {
      "command": "uv",
      "args": [
        "run",
        "/foo/bar/path/to/mcp_server/.venv/bin/mcp_server_name"
      ]
    }
  }
}

You can run per-instance MCP servers for Goose by using Python virtual environments created inside each MCP directory and launching them through the uv tool. This setup lets Goose discover and start your custom MCP endpoints, enabling you to integrate data sources and actions directly into Goose workflows.

How to use

In Goose, you enable an MCP server by starting the MCP process in its own virtual environment and registering the start command with Goose. First, create and activate the Python virtual environment for your MCP, then run the MCP server through the uv tool so Goose can manage and expose it to your Goose configuration. Once registered, Goose will show the MCP server as an available option you can connect to or route data through.

How to install

Prerequisites you need on your system before you begin are an available Node.js and Python installation, plus the uv tool used to manage MCP servers. Follow these concrete steps in order.

Ensure Node.js and Python are installed on your machine.
Install or prepare the uv tool according to your environment so you can manage MCP virtual environments and run MCP servers.
In each MCP directory, create the Python virtual environment and install required dependencies, then synchronize the environment with uv.
Register the MCP server with Goose by adding the startup command to Goose configuration. The command you register is the one used to start the MCP server from its virtual environment, as shown here.

uv sync
goose configure add uv run /foo/bar/path/to/mcp_server/.venv/bin/mcp_server_name

Additional sections

Notes and recommendations for using this Goose MCP setup: the uv sync step creates a per-MCP Python virtual environment, ensuring that each MCP server has its own isolated runtime. The startup command you register with Goose should point to the binary inside that virtual environment. If you need to adjust paths for different machines, replace the placeholder path with the actual location of your MCP server in your environment.