Build, run and scale AI agents like API and microservices
Review agent-authored changes in hunk from herdr and send inline comments back to the responsible agent.
A sandbox service for running untrusted, agent-authored code that reports the isolation boundary each run executed behind. Lessons, no clone needed: https://plimsollmark.github.io/plimsoll/trainers/