I ended up building this for myself as a fairly heavy duty workflow skill and a plugin in Paseo - it's not a kanban view yet, just a filtered list, but turning it into kanban would be relatively trivial.
Everyone saying that app x, y and z already do this may be missing the point - I tried them all within the last week and none of them actually do this out of the box, at least not yet. Not properly.
BB is probably the one to build this in, I just used Paseo because that's what I already had set up.
The important bit is the orchestration - you need PR reviewer threads in the same context as the worker thread, and a coordinator when either get blocked, with model pool, state machine, database, etc. It's much harder to set up than just having one thread talk to another with a handoff.
I'm happy to share if you want to mess with what I've built, though the skill requires a significant documentation commitment in your repo which might not fit for you. DM me if interested
@Shpigford I'm using Fable or Sol as orchestrator.. Grok 4.6 is a *great* worker model. All in t3code (talking to native harnesses) + a skill to write worker handoffs - spin up a new worktree for each worker (easy in t3) and have the orchestrator review and merge the PR. Works great
@linode yeah sorry, but 3 blackouts in a month is too many. This looks like exactly the same problem as last time. There should be redundancy in place by now.
Oh looky, it wasn't a fluke; I've written another thing.
This time it's about how creative freedom is a fallacy and we crave rules and constraints.
Go on, go give it a read. But really quickly, or standing on your head or somethingโฆ
https://t.co/hqzm25NnZT