I just released version 0.1.3 of Canopy github.com/ferranbt/canop…
The theme of this version is compatibility. It includes +70 integration test that run Canopy's runners against the real GH runner and validate input verbatim.
What is people using to coordinate agents? Are you rolling out your own coordination layer (launching, syncing, communication...) or is there some common framework to do it?
Just open sourced my latest summer project: canopy
github.com/ferranbt/canopy
it's a deconstruction of the Github Actions protocol. take the protocol, lower it into its independent modular components, then build the services back up from those.