humanlayer

Why we included this project

Running an AI coding agent on a real codebase means deciding how much autonomy to grant it. HumanLayer's answer is a human-in-the-loop approval layer that intercepts risky actions, such as writing to a database or sending messages, and waits for a person to sign off before the agent proceeds. The maintainers are upfront that this repository is essentially a public issue tracker; the code here is deprecated in favor of the rebuilt product at humanlayer.com. That makes the repo a useful place to understand the design of the approval workflow and follow the discussion, rather than something to deploy. If you're shopping for a way to keep human oversight over your agents, the SDK docs and issue threads here are a solid starting point.