AgentIndex icon
AgentIndex
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Submit Tool
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Home/
Multi-Agent/
mcp-agent
mcp-agent logo

mcp-agent

Active·★ 8.6k·Apache-2.0·Updated 2026-01-25
★ Essential

mcp-agent, from LastMile AI, is a Python framework built on the idea that MCP is all you need to build agents. It manages the lifecycle of MCP server connections for you, and an agent is declared with a name, an instruction and the MCP servers it may use, such as filesystem and fetch in the README's minimal example. You then attach an LLM (OpenAI in the example) and ask it to do the work.

The README lists three pillars. It implements every pattern from Anthropic's Building Effective Agents article in a composable way (map-reduce, orchestrator, evaluator-optimizer, router and more), so patterns can be chained. It claims full MCP support, including tools, resources, prompts, notifications, OAuth, sampling, elicitation and roots. And it can run the same agent code on Temporal for durable workflows that pause, resume and recover without API changes.

It can also expose agents as MCP servers with a FastMCP-compatible API, and offers a beta cloud deployment path (mcp-c). Install with `pip install mcp-agent`; the full docs, an llms.txt and a docs MCP server are at docs.mcp-agent.com. Our data shows the last push on 2026-01-25.

This write-up is based on the project's README and public repository data. We have not hand-tested the tool. Data as of 2026-10-23.

License
Apache-2.0
Latest release
v0.0.21 · 2025-05-09
Last commit
2026-01-25
Contributors
60
Forks
893
Open issues and PRs
149
First published
2024-12-18

Source: GitHub, npm and PyPI public APIs.

#Agent Framework#Model Context Protocol#LLM#Python#Temporal#Coding
$ Install
$ pip install mcp-agent
↗ Visit site★ GitHub
01

Features

01Full MCP support, abstracting server connection lifecycle management.
02Composable implementation of effective agent patterns (e.g., map-reduce, orchestrator).
03Durable agents with Temporal for production-scale workflows, enabling pause, resume, and recovery.
04Ability to create and expose custom MCP servers, including agents as servers.
05Production-ready features like structured logging, token accounting, and first-class cloud deployment.
02

Why choose it

+Implements Anthropic's agent patterns (orchestrator, router, evaluator-optimizer, map-reduce) as composable building blocks.
+Handles MCP server lifecycles and claims full MCP support, including OAuth, sampling and elicitation.
+The same agent code can run on Temporal for durable, resumable workflows.
+Agents can be exposed as MCP servers; Apache-2.0 with docs, llms.txt and a docs MCP server.
03

Trade-offs

!Python only; teams on TypeScript need a different framework.
!Temporal durability adds infrastructure you must run, and the cloud deployment path is marked beta.
!Our data shows the last push on 2026-01-25, so check recent activity before relying on it.
04

Maintenance check

SlowingChecked: 2026-10-25
Last commit
2026-01-25
Commits, last 90 days
0
Releases, last 90 days
0
Latest release
v0.0.21 (2025-05-09)
Open issues
148
Recent issues with at least one comment
2/10
Commits per week, last 12 weeks
Commits per week, last 12 weeks: 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0

Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

Most discussed open issues

  • #40 Support LLMs through Cloud Vendors18 comments · 2025-03-05
  • #25 content in OpenAI ChatCompletion messages is a list, not string16 comments · 2025-02-13
  • #151 [mcp_agent.mcp.mcp_agent_client_session] send_request failed: EndOfStream() (EndOfStream)13 comments · 2025-04-23
  • #11 Surfacing intermediate events (e.g. tool calls, LLM generations, etc.) in a long-running flow.9 comments · 2025-01-29
  • #33 Use of agents in threads8 comments · 2025-02-28

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

05

Compatibility

Model Context Protocol
Native
Verified via docs
Temporal
Supported
Verified via docs
OpenAI
Supported
Verified via docs
Anthropic
Supported
Verified via docs
Google Gemini
Supported
Verified via docs
Azure OpenAI
Supported
Verified via docs
06

Quick start

1
$ pip install mcp-agent
07

Use cases

↳Building robust LLM agents that integrate with diverse MCP servers (e.g., filesystem, web fetch).
↳Developing and deploying custom MCP servers, including exposing intelligent agents as services.
↳Scaling agent applications to production environments with durable execution and cloud deployment capabilities.
08

Alternatives

⚡Top VS:
VSPydantic AI★20.9k→VSchatarena★1.6k→VSconcordia★1.8k→VSdeer-flow★84.5k→
Pydantic AI logo
Pydantic AI★ 20.9k
GenAI Agent Framework, the Pydantic way. A Python agent framework designed for production with type safety at its core.
VSCompare→
chatarena logo
chatarena★ 1.6k
ChatArena (or Chat Arena) is a Multi-Agent Language Game Environments for LLMs. The goal is to develop communication and collaboration capabilities of AIs.
VSCompare→
concordia logo
concordia★ 1.8k
Concordia, from Google DeepMind, is a Python library for generative social simulation: building generative agent-based models in which a Game Master simulates the environment that LLM-driven entities act in.
VSCompare→
deer-flow logo
deer-flow★ 84.5k
DeerFlow is a community-driven Deep Research framework, combining language models with tools like web search, crawling, and Python execution, while contributing back to the open-source community.
VSCompare→
Microsoft AutoGen logo
Microsoft AutoGen★ 61.6k
A framework that enables the development of LLM applications using multiple agents that can converse with each other to solve tasks.
VSCompare→
edsl logo
edsl★ 510
Design, conduct and analyze results of AI-powered surveys and experiments. Simulate social science and market research with large numbers of AI agents and LLMs.
VSCompare→
CrewAI logo
CrewAI★ 60.2k
Framework for orchestrating role-playing, autonomous AI agents. By working together, your Crew can tackle complex tasks.
VSCompare→
skyagi logo
skyagi★ 775
SkyAGI: Emerging human-behavior simulation capability in LLM
VSCompare→
See all alternatives →

Comments

Log in to leave a comment

No comments yet. Be the first!

On this page
01Features02Why choose it03Trade-offs04Maintenance check05Compatibility06Quick start07Use cases08Alternatives
lastmile-ai /
mcp-agent
↗

Apache-2.0 Python framework for building agents on MCP with Anthropic's composable workflow patterns and optional Temporal durability

8.6k893Apache-2.0
Last commit9mo ago
StatusActive
CategoryMulti-Agent
Contributors60
Latest releasev0.0.21
Links
Documentation↗Discussion↗Issues↗Releases↗
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit
© 2026 AgentIndex.app|Built by a 10-year iOS Developer.
Best by use caseAboutContactAdvertisePrivacy PolicyTerms of UseQYSGitHubBuy me a coffee ☕

Browse by Category

MCP ServersCode AssistantWorkflow AutomationRAG / Knowledge BaseMulti-AgentBrowser AutomationLLM InfraDev ToolingObservability

Not affiliated with Anthropic, OpenAI or Microsoft.