Open-Source AI Agents Directory
Find open-source AI agents you can run yourself: coding, research, browser and workflow agents, plus frameworks for building your own. Compare them by GitHub stars, licence and recent activity, see what each agent can touch on your machine, and read the dated scan report where one has run.
Coding agents, research agents, browser agents, workflow builders and agent frameworks, all open source.
Try a different search or category, or submit the repo you were looking for.
List refreshed November 15, 2026. Data comes from public GitHub repositories.
Compare open-source AI agents by job
Pick the job first, then compare. Coding agents edit and run code in your repository. Research agents search, read and summarise sources. Browser agents click and type in a real browser. Workflow builders chain steps in a visual editor. Frameworks are libraries for building your own agent.
Open-source coding agents
Coding agents read your repository, edit files and run commands in a terminal, so they usually touch files and the shell. Compare them by licence, how they ask before writing, and which models they support. Browse coding agents →
Research and analysis agents
Research agents search the web, read documents and write summaries or reports. Most need internet access and an API key for a model. Check what they store and where. Browse research agents →
Browser agents
Browser agents control a real web browser to fill forms, click and scrape pages. A logged-in browser gives the agent your sessions, so pages it reads can try to steer it. Browse browser agents →
Workflow and no-code agent builders
Workflow builders connect models, tools and triggers in a visual editor, then run the flow on a schedule or event. They often hold many API keys at once. Browse workflow agents →
Open-source AI agent frameworks
Frameworks are code libraries for building your own agent: they handle the model loop, tool calls and memory. You write the agent, so the risk is in what you wire up. Browse developer tools →
How to choose the best open-source AI agent
There is no single best open-source AI agent. The right one depends on the job, where it runs and what it can reach. Start from the task, then compare licence, recent commits, model support and the what-it-can-touch label, and try it on a small task in a project with no secrets first.
Licence: open source or source-available?
MIT, Apache-2.0 and BSD let you use and change the code commercially. Some well-known agents have moved parts to source-available licences, which limit commercial use, so read the LICENSE file, not just the repo badge. Every listing shows the licence from the repository.
Activity and models
Look at the last code change and open issues; this list hides projects with no push in 12 months by default. Check whether the agent works with your model provider or with local models.
AI agent vs agent framework vs chatbot
An AI agent decides its own next steps to finish a task and calls tools such as a shell, browser or API along the way. An agent framework is the library developers use to build one. A chatbot answers in conversation but does not act on its own. This directory lists agents and frameworks.
Running open-source AI agents with less risk
Open-source AI agents run with whatever access you give them. Use a separate project or container, give the agent its own API keys with the smallest scope, keep approval prompts on for anything that writes, sends, pays or deletes, and read what the listing says the agent can touch before you start it.
Most-starred open-source AI agents right now
Ranked by GitHub stars, these open-source agents and frameworks lead the list today. Stars show popularity, not quality, so read the licence, recent activity and what each one can touch too.
- 1ECC by affaan-mThe agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.294k stars · MIT
- 2Deepseek Harness by deepseek-aiDeepSeek Harness: Everything is a Plugin.268k stars · MIT
- 3Hermes Agent by NousResearchThe agent that grows with you259k stars · MIT
- 4Firecrawl by firecrawlSupercharge your AI agents with data from the web and beyond. Building the library for superintelligence. 🔥198k stars · AGPL-3.0
- 5AutoGPT by Significant-GravitasAutoGPT 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.188k stars
- 6Ponytail by DietrichGebertMakes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.182k stars · MIT
Search this AI agent directory from your editor
The free AgenticMart MCP server and CLI let Claude Code, Cursor or VS Code look up agents, MCP servers and agent skills, with the same stars, licence and scan data shown here. Setup takes one command per client.
Set up the MCP server and CLI →MCP servers →Agent protocols →
List your open-source agent
Listing an open-source agent is free, and so is the automated scan. Submit the GitHub repository and the listing appears with stars, licence, install steps and what the agent can touch. Claim an existing listing to edit it. Paid selling is not open yet: list free now, paid selling opens soon.
How this list is ranked
This list ranks open-source AI agents by GitHub stars, and by default shows only projects with code pushed in the last 12 months. Switch to All to include older projects, or sort by newest or recently updated. Data comes from public GitHub metadata and each project's README. Scan results are provided under our scan policy.
AI Agents FAQ
What is the best open source AI agent?
There is no single best one; it depends on the job and where the agent runs. Filter this directory by coding, research, browser or workflow agents, then compare licence, recent activity and what each one can touch.
Are open source AI agents safe to use?
They are only as contained as the access you give them. Run new agents in a container or throwaway project, use scoped API keys, keep approval prompts on, and read the listing's what-it-can-touch rows and dated scan report first. A scan is not a guarantee.
Can I run an AI agent completely locally?
Yes, many open-source agents can run on your own machine with a local model. Check the project's README for local model support; some features, such as hosted sandboxes or web search, still call outside services.
What is the difference between an AI agent and an agent framework?
An agent is a ready program that plans and acts on a task; a framework is a library you use to build one. This directory lists both, and the category filters separate coding agents, workflow builders and developer frameworks.
What is an AI agent marketplace?
An AI agent marketplace is a place to find, compare and get AI agents, sometimes paid. AgenticMart lists open-source agents, MCP servers and agent skills for free today; paid selling opens soon.
How is this list of AI agents ranked?
By GitHub stars, limited by default to projects with code pushed in the last 12 months. Switch to All for older projects, or sort by newest or recently updated.
Can I list my own agent?
Yes. Listing is free and so is the automated scan: submit the GitHub repository, or claim an existing listing to edit it. Paid selling opens soon.