evalstate

fast-agent: Build and Run LLM Agents and Workflows

A Python framework and CLI for building, testing, and running LLM agents that use MCP tools and compose into workflows. It suits developers who want both interactive agent development and a programmable framework.

ActivePythonApache-2.0AI AgentsCLIMCP
View on GitHub Compare
Stars
3.9k
Forks
446
Stars, last 30 days
+30
Last commit
2 days ago
Contributors
78

Our take

Activeregular commits and releases

Health score

70/100

How it is scored
Activity
86
Community
55
Issues
56
Pull requests
84

+30 stars in the last 30 days

  • Merges more pull requests than 96% of the projects we track
  • Releases more often than 88% of the projects we track

fast-agent is a capable, actively maintained option for Python teams building MCP-connected agents, with adoption risk concentrated in its reliance on a dominant contributor.

Good fit if

  • You want a Python-first framework that combines agent development, CLI use, and workflow composition.
  • Your application needs MCP features such as tools, resources, sampling, elicitation, or OAuth-backed HTTP connections.
  • You can work with Python 3.12 through 3.14 and value a project with regular commits and releases.

Look elsewhere if

  • Your team needs a framework in a language other than Python.
  • You require a project whose development is broadly distributed, since most contributions come from one person.
  • Your use case does not need the MCP and agent-workflow capabilities that are central to this project.
All health signals
Last commit2026-10-04 (2 days ago)
Commits, last 90 days100+
Releases, last 12 months59 (latest v0.10.41, 2026-09-29)
Contributors78 (top contributor: 82% of commits)
Issues closed, last 90 days6+ (typically closed in 4 days)
Project age21 months

Checked on 2026-10-07 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

3,970

Forks

469

Change since Nov 16, 2026: 0 stars·0 forks

Overview

fast-agent is a Python framework and command-line tool for building and interacting with LLM agents. It addresses the gap between experimenting with one agent and assembling useful applications by providing a declarative way to connect agents, tools, and workflow steps.

It is designed for developers who need MCP integration, multiple model providers, and workflows such as routing, chaining, parallel execution, or evaluator-optimizer loops. It can also be used as an interactive coding and development agent, or run in server mode.

Key Features

  • Define agents in Python and connect them to MCP servers or Python function tools.
  • Compose agents into chains, parallel workflows, routers, orchestrators, and evaluator-optimizer loops.
  • Work with multiple model providers, including Anthropic, OpenAI, and Google, with additional providers supported through TensorZero.
  • Use MCP capabilities including sampling, elicitation, prompts, resources, and OAuth for HTTP and SSE servers.
  • Run an interactive CLI with shell access, model selection, skills management, and streaming terminal output.
  • Pass multimodal resources such as PDFs and images into agent prompts.
  • Expose an application as an MCP server or run it through an ACP client.

Use Cases

  • A Python developer can prototype an agent that calls MCP tools, then keep its prompts and configuration under version control.
  • A team can split a research or analysis task among specialist agents and aggregate their results with parallel or orchestrated workflows.
  • Developers can test how different models interact with the same MCP server while diagnosing transport behavior.
  • A coding-agent user can launch an interactive shell-enabled agent from the terminal and connect it to skills or MCP services.

What you need

Detected in the repository

  • Python >=3.12,<3.15 (from pyproject.toml)
  • A test suite and automated checks on GitHub Actions

License in plain words

Apache-2.0permissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the license, the NOTICE file and a note of your changes
  • Share your changes: no
  • Includes an explicit patent grant from the contributors:

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

Install with uv and launch the interactive CLI:

uv tool install -U fast-agent-mcp
fast-agent go

See the README and documentation for model configuration, MCP setup, and examples.

Alternatives

  • agent-factory: Agent Factory generates executable Python workflows from task descriptions, while fast-agent provides a framework and CLI for building and composing agents.
ProjectLanguageLicenseStarsStatus
fast-agentPythonApache-2.03.9kActive
agent-factoryPythonApache-2.0115Inactive

Considerations

The project is actively maintained, with frequent recent commits and releases, and issues are being closed; however, development is highly concentrated in one contributor. Teams should account for that dependency when evaluating long-term maintenance risk. The runtime requirement is Python 3.12 or later, but below 3.15. Model-provider credentials and any external MCP services needed by an application must be configured separately.

Found this useful?

Share it with someone who would like fast-agent.

Comparisons