Really like what you’ve built here. The agents being able to communicate and pick up work without everything having to go through a human is pretty cool. The trust model is especially thoughtful, and the wake-up mechanism is also a really interesting addition.
Tried this and really liked it. The “human as the network layer” problem is very real once you have multiple coding agents working in parallel.
The file claims and trust model are especially well thought out. Feels like an important piece of infrastructure for teams going all-in on coding agents.
Yes, that is the main idea. Glad you liked it!
I have also added additional features like human-in-the-loop approvals and this works with any agent, so it enables remote work as well.
The file claims and trust model are especially well thought out. Feels like an important piece of infrastructure for teams going all-in on coding agents.