What are the best Python autonomous agents?

By maintained adoption — GitHub stars plus recent commit activity — Hermes Agent, AutoGPT and deer-flow lead the Python autonomous agents in this index of 9. The full ranking is below; it reflects what is genuinely used and still maintained, not what pays.

Are these Python autonomous agents open source and free?

Yes — every project on this page is open source and written primarily in Python. MIT is the most common licence here, on 5 of 9, across 3 licences in total. The software is free to run and self-host; you still pay for the infrastructure you run it on and any model or API usage it makes.

Why choose Python autonomous agents specifically?

Staying in your team's primary language — Python — makes self-hosting, extending, and debugging far easier, because you can read and modify the source directly instead of treating it as a black box.

Open source · Python

Best Python autonomous agents

The 9 open-source autonomous agents written in Python in this index, ranked by real maintained adoption (GitHub stars and recent commit activity), never by sponsorship. Hermes Agent leads the set at 249,707 GitHub stars, and 9 of the 9 have shipped a commit in the last 90 days. Every project here is Python-first, self-hostable, and free to run; the trade-off is you host and maintain it yourself. Right now it's led by Hermes Agent (250k stars), with AutoGPT and deer-flow close behind. Rankings shift as projects gain stars and ship commits.

9 open-source records · all autonomous agents · more best-of lists · how we rank

Hermes AgentAgentNous Research's self-improving AI agent that builds skills from experience and remembers you across sessions.Open sourcePython · MITVerified · Active 250kGitHub stars AutoGPTAgentAutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.Open sourcePythonVerified · Active 188kGitHub stars deer-flowAgentAn open-source long-horizon SuperAgent harness that researches, codes, and creates.Open sourcePython · MITVerified · Active 83kGitHub stars nanobotAgentLightweight, open-source AI agent for your tools, chats, and workflows.Open sourcePython · MITVerified · Active 49kGitHub stars CowAgentAgentOpen-source super AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-model, multi-channel. Lightweight, extensible, one-line install.Open sourcePython · MITVerified · Active 47kGitHub stars agnoAgentBuild, run, and manage agent platforms.Open sourcePython · Apache-2.0Verified · Active 42kGitHub stars wshobson/agentsPlatformMulti-harness agentic plugin marketplace for Claude Code, Codex CLI, Cursor, OpenCode, GitHub Copilot, and Gemini CLIOpen sourcePython · MITVerified · Active 40kGitHub stars agenticSeekAgentFully Local Manus AI. No APIs, No $200 monthly bills. Enjoy an autonomous agent that thinks, browses the web, and code for the sole cost of electricity.Open sourcePython · GPL-3.0Verified · Active 27kGitHub stars agentic_securityAgentAgentic LLM Vulnerability Scanner / AI red teaming kitOpen sourcePython · Apache-2.0Verified · Active 2.0kGitHub stars

Frequently asked

What are the best Python autonomous agents?: By maintained adoption — GitHub stars plus recent commit activity — Hermes Agent, AutoGPT and deer-flow lead the Python autonomous agents in this index of 9. The full ranking is below; it reflects what is genuinely used and still maintained, not what pays.
Are these Python autonomous agents open source and free?: Yes — every project on this page is open source and written primarily in Python. MIT is the most common licence here, on 5 of 9, across 3 licences in total. The software is free to run and self-host; you still pay for the infrastructure you run it on and any model or API usage it makes.
Why choose Python autonomous agents specifically?: Staying in your team's primary language — Python — makes self-hosting, extending, and debugging far easier, because you can read and modify the source directly instead of treating it as a black box.