🚨 We're open-sourcing Druids, a library for coordinating and deploying coding agents across machines. Our beta users have used Druids to work on open math problems, conduct ML "autoresearch," and make software faster.

We believe Druids is the right way to write agent software. We'd love to hear what you think — try it today: Website: GitHub: Launch Post:

Research and software development advance through repeated cycles of execution, evaluation, and refinement. But human cognitive labor makes this process costly. Druids provides the language and infrastructure to automate these cycles with agents.

How does Druids work? Users write Python programs that specify agents, their goals, and how they talk to each other. Druids handles provisioning VMs and coordinating the agents.

Druids programs are written as state machines. You define what your program tracks, the events that move it forward, and how agents handle errors and branching. Because every dependency is explicit, Druids makes it easy to parallelize your agents.

What does this look like in practice? → Force a model to iterate against hard tests until it passes → Build a verification chain where agents check and challenge each other's outputs → Control how multiple agents write to shared resources without conflicts

You focus on specifying the shape of your agent team while Druids handles the infrastructure. The result is agent software that is predictable, reproducible, and testable.

Join us also on discord for questions or feedback:

agents coordinating is wild

This is a big unlock. Coordinating agents across machines is the hard part most teams underestimate — excited to see Druids tackle it in the open. Curious how beta users are handling state management across long-running agent sessions.