A curated chaos of AI agents: 200+ projects, one repo

What it does

This is a hand-maintained directory of AI autonomous agents, split cleanly between open-source projects and closed-source products. Each entry gets a category tag—general purpose, coding, multi-agent, research—and a terse description of what it actually does. The maintainers also run a web UI version with filtering, because a flat markdown list at this scale becomes essentially unreadable.

The interesting bit

The curation itself is the product. The repo includes a landscape diagram (landscape-latest.png) that tries to map the whole space visually, plus an explicit backlog of agents waiting to be added—some held up because they lack English documentation or because the maintainers haven’t verified they qualify as “agents” versus mere wrappers. That transparency about editorial judgment is rarer than it should be.

Key highlights

• Coverage spans from research toys (Agent4Rec’s 1,000-agent recommender simulator) to production-ish tools (Aider for CLI pair programming, SWE-agent for GitHub issue resolution)
• Each entry links out to papers, Discord servers, and live demos where they exist
• Explicit scope boundary: SDKs and frameworks get exiled to a sibling repo, keeping this list strictly for end-user agents
• Submit via PR or Google Form; alphabetical order within categories is enforced
• Maintained by E2B, a code-interpreting infrastructure company, which explains the occasional cross-promotion

Caveats

• The README admits the list is “definitely not comprehensive” and the TBD backlog is substantial
• Some entries are thin—AgentPilot’s entire description is “Integrated into Open Interpreter and MemGPT” plus “Group chats feature”
• Closed-source section is mostly logos and landing pages; less vetting depth than the open-source side

Verdict

Worth bookmarking if you’re building in this space and need to know what already exists. Skip it if you want deep analysis or comparative benchmarks—the value is breadth, not depth.