Mcp Agent
@lastmile-ai
Build effective agents using Model Context Protocol and simple workflow patterns
Overview
What is Mcp Agent?
Mcp Agent is a simple, composable framework for building effective agents using the Model Context Protocol (MCP). It fully implements MCP, manages the lifecycle of MCP server connections, and implements every pattern from Anthropic's "Building Effective Agents" in a composable way. Mcp Agent is for developers who want to build agents that can pause, resume, and recover without API changes, scaling from simple agents to sophisticated workflows on Temporal.
How to use Mcp Agent?
Install with uv add mcp-agent or pip install mcp-agent and optionally add LLM provider extras (e.g., openai, anthropic). Configure MCP servers and secrets in YAML files, then use the Python SDK to create Agent instances, attach an Augmented LLM, and generate responses. The CLI is available via uvx mcp-agent for scaffolding and deployment.
Key features of Mcp Agent
- Full MCP support: Tools, Resources, Prompts, Notifications, OAuth, and more
- Composable agent patterns: map-reduce, orchestrator, evaluator-optimizer, router
- Durable execution via Temporal without
