Strands Agent MCP

A Model Context Protocol (MCP) server for executing Strands agents. This project provides a simple way to integrate Strands agents with Amazon Q and other MCP-compatible systems.

Overview

Strands Agent MCP is a bridge between the Strands agent framework and the Model Context Protocol (MCP). It allows you to:

Register Strands agents as MCP tools
Execute Strands agents through MCP
Discover and list available agents

The project uses a plugin architecture that makes it easy to add new agents without modifying the core code.

Related MCP server: Elasticsearch MCP Server

Installation

pip install strands-agent-mcp

Usage

Starting the MCP Server

strands-agent-mcp

This will start the MCP server on the default port.

Creating Agent Plugins

To create a new agent plugin, create a Python package with a name that starts with sap_mcp_plugin_ (sap stands for strands agent plugin). Your package should implement a register_plugin function that registers one or more agents with the provided registry:

from strands import Agent from strands.models import BedrockModel from strands_agent_mcp.registry import Registry def register_plugin(registry: Registry) -> None: registry.register("my-agent", Agent( model=BedrockModel(boto_session=Session(region_name="us-west-2"))) )

Using with Amazon Q

Once the MCP server is running, you can use the agents with Amazon Q:

q chat --mcp-server http://localhost:8000

Then you can use the following commands in your chat:

List available agents: strands___list_agents
Execute an agent: strands___execute_agent with parameters agent (agent name) and prompt (the prompt to send to the agent)

Architecture

The project consists of three main components:

Server: The MCP server that exposes the agent execution API
Registry: A simple registry for managing available agents
Plugins: Dynamically discovered modules that register agents with the registry

The server automatically discovers all installed plugins that follow the naming convention and registers their agents.

Dependencies

fastmcp: For implementing the MCP server
strands-agents: The core Strands agent framework
strands-agents-builder: Tools for building Strands agents
strands-agents-tools: Additional tools for Strands agents

Development

To set up a development environment:

Clone the repository
Create a virtual environment: python -m venv .venv
Activate the virtual environment: source .venv/bin/activate (Linux/Mac) or .venv\Scripts\activate (Windows)
Install development dependencies: pip install -e ".[dev]"

Creating a Test Plugin

The repository includes a sample plugin (sap_mcp_plugin_test) that demonstrates how to create and register a simple agent called "simple-agent":

from boto3 import Session from strands import Agent from strands.models import BedrockModel from strands_agent_mcp.registry import Registry def register_plugin(registry: Registry) -> None: registry.register("simple-agent", Agent( model=BedrockModel(boto_session=Session(region_name="us-west-2"))) )