mcp -rabbitmq

The RabbitMQ MCP Server wraps the admin APIs of a RabbitMQ broker and exposes them as MCP tools that can be orchestrated by an MCP client. It supports connecting to multiple RabbitMQ brokers within a single session, enabling centralized management of several clusters from one interface. With OAuth support and FastMCP Bearer Auth integration, you can connect using tokens and authorization providers for secure, token-based authentication. The server is designed to work with Streamable HTTP transport via FastMCP, giving you scalable, transport-flexible tool access if you configure your IdP and a compatible Bearer Auth provider. You’ll interact with the MCP server through an MCP client (e.g., via uvx), which discovers, authenticates, and executes the broker administration tools exposed by this server. Tools are available to manage common RabbitMQ administration tasks, such as monitoring brokers, queues, exchanges, and user permissions, depending on the server’s configured capabilities. To start, run the server through your MCP client configuration, then list the available tools in your session and invoke them as needed. The README provides example configurations showing how to start the server via uvx and how a Strands agent can connect and issue commands against the MCP tools.

Prerequisites

• Python 3.8+ (or a compatible environment)
• Access to PyPI or the ability to build from source
• If using from source: Git installed and network access to install dependencies

Option A: Install from PyPI (recommended)

1. Install the MCP server package from PyPI using uvx as the launcher:

uvx amq-mcp-server-rabbitmq@latest --allow-mutative-tools

Note: The command above downloads and runs the RabbitMQ MCP server package from PyPI. The --allow-mutative-tools flag enables tools that can mutate broker state.

Option B: Install from source

1. Clone the repository:

git clone https://github.com/amazon-mq/mcp-server-rabbitmq.git
cd mcp-server-rabbitmq

2. Install dependencies (using your preferred Python environment):

pip install -e .

3. Run the server via uv (or uvx if you have a wrapper ready):

uv run amq-mcp-server-rabbitmq --allow-mutative-tools

Configuration tips

• If you plan to run multiple brokers, reuse a single uvx invocation and connect to each broker via separate tool configurations or sessions.
• For HTTP-based transport, ensure you supply --http and the Bearer Auth provider settings (JWKS URI, issuer, audience, and required scopes) as described in the CLI arguments.

Optional verification

• Run tests if provided in the repository (e.g., pytest) to verify environment and dependencies.

Tips and common considerations:

• The server supports multiple RabbitMQ brokers in one session, so you can manage clusters in a single MCP client context.
• OAuth and FastMCP Bearer Auth integration requires proper configuration of JWKS URI, issuer, audience, and scopes. Ensure your IdP is correctly set up and reachable from the MCP server.
• If you enable --http (streamable HTTP), you’ll switch from stdio to HTTP transport; verify firewall and TLS settings as needed.
• The default MCP server port is 8888 via --server-port; adjust as needed for your environment.
• The CLI flag --allow-mutative-tools controls whether tools can mutate broker state. Disable in read-only environments for safety.
• When running from source, paths such as /path/to/repo/mcp-server-rabbitmq should be replaced with your actual checkout path.