Hi! What’s the procedure for agent-canvas hardenin...
# proj-agent-canvas
l
Hi! What’s the procedure for agent-canvas hardening now? I’ve been creating PRs targeting main for both features and bug fixes, but should we be cherry-picking the bug fixes against a beta branch or something? Sorry if this has been covered already but I couldn’t find it easily
h
Good question. We don't have a beta branch release yet or anything. AFAIK we are still making PRs against main and trying to get it to a beta release ready point
l
Oh, got it. So we didn’t actually implement a feature freeze branch?
I guess that might be OK as long as we’re focusing on squashing bugs
p
No feature freeze branch that I'm aware of (cc: @ancient-car-4311 or maybe @green-greece-46378 - is that accurate?) I think we do want to cut our beta branch soon? I'm still a little fuzzy on the plan here and would love to get that plan clearly defined.
c
@nutritious-balloon-65098 find all issues tagged MVP and on roadmap for MVP and tell us where we are in the quest to it.
ty 2
n
I'm on it! can track my progress here.
I answered the question by checking the
OpenHands/agent-canvas
GitHub issue tracker for the
canvas-mvp
label and the
OpenHands Agent Canvas Roadmap
project status. Summary: • Found 30 total
canvas-mvp
issues
. • Found 23 closed and 7 still open, so MVP-tagged work is about 77% complete by issue count. • Found 7 MVP-tagged issues on the roadmap project, with 5 still open and 2 closed. • Identified the remaining open MVP work: ◦ Secure auth for canvas ◦ Automations regressions ◦ Docker/local-backend conversation persistence after restart ◦ Event-based automation UX/detail visibility ◦ Linear issue triage prebuilt automation ◦ Default model mismatch ◦ Unnecessary local git polling without attached workspace • Noted that active PRs exist for 3 open MVP issues: ◦ Auth modes ◦ Default model mismatch ◦ Local git polling • Concluded that agent-canvas appears to be in late MVP hardening: most tagged issues are closed, but MVP/beta does not look fully ready until the auth story, automations reliability/visibility, and Docker conversation persistence are resolved or explicitly de-scoped. No code changes were made.