MCP Catalogs
Homemcp-agent screenshot

mcp-agent

by lastmile-ai·8,325·Score 59

A comprehensive Python framework for building effective agents with Model Context Protocol and simple workflow patterns.

ai-llmdeveloper-toolsproductivity
837
Forks
127
Open issues
4 mo ago
Last commit
2d ago
Indexed

Overview

mcp-agent is a powerful Python framework that fully implements the Model Context Protocol (MCP) to build robust agents. It provides simple, composable patterns for agent development, supports both basic and sophisticated workflows with Temporal for durable execution, and allows agents to leverage MCP servers for various capabilities. The framework includes LLM integration, tools/resources/prompts support, and cloud deployment options, making it suitable for production-grade agent applications.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

you:Building complex multi-step agents that use MCP servers to interact with filesystems, web APIs, and other systems
you:Creating durable, stateful agents with Temporal for production workloads
you:Deploying agents as MCP servers that can be integrated into other systems
you:What MCP features are supported?
you:Can I use this with different LLM providers?

When to choose this

Choose mcp-agent when you need to build production-ready LLM agents with advanced patterns like durable execution, human-in-the-loop, and cloud deployment.

When NOT to choose this

Avoid if you need a minimal MCP implementation or are already deeply invested in another agent framework with different abstractions.

Tools this server exposes

3 tools extracted from the README
  • fetch

    Fetch content from URLs using the fetch MCP server

  • filesystem

    Read and write files on the local filesystem using the filesystem MCP server

  • list_tools

    List all MCP tools available to the agent

Note: Tool names were inferred from the example code showing the agent connecting to 'fetch' and 'filesystem' MCP servers, and the code demonstrating listing available tools.

Comparable tools

fastmcplangchain-mcpn8ntemporal

Installation

Installation

Install with pip:

pip install mcp-agent

Or using uv (recommended):

uv add mcp-agent

For additional LLM providers:

uv add "mcp-agent[openai, anthropic, google, azure, bedrock]"

Claude Desktop Integration

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "mcp-agent": {
      "command": "uvx",
      "args": ["mcp-agent", "run"]
    }
  }
}

FAQ

What MCP features are supported?
Full MCP support including tools, resources, prompts, notifications, OAuth, sampling, elicitation, and roots.
Can I use this with different LLM providers?
Yes, it supports OpenAI, Anthropic, Google, Azure, and Bedrock LLM providers with easy configuration.

On Hacker News

Recent discussion from the developer community.

Compare mcp-agent with

GitHub →

Last updated · Auto-generated from public README + GitHub signals.