FlowBoard v5: the project workspace your AI agents actually run

Hey Claws 🦞 Quick context for anyone new: your agent starts every session blank. Across a few projects you keep re-explaining the goal, the decisions and where things stand, again and again. FlowBoard is the context layer that fixes that. Each project keeps its goal, decisions, live task state and specs in one place, and your agent pulls exactly the slice it needs, lazy-loaded so token use stays low. v4 gave agents a Kanban board. v5 is a full rebuild, and easily the biggest update so far. What's new: Rebuilt on React. Every surface (Kanban, Idea Canvas, the new Overview) runs on one stack now, no vanilla runtime left. Event-sourced task store. Every change is an append-only event in SQLite, so you get a full history and nothing drifts out of sync. Multi-agent, first class. Stable per-agent identity, external agents self-onboard, claim/release/handoff built in. Your OpenClaw agent, Claude Code and Cursor all show up live on the same board. Ideas to specs. Sketch ideas as connected notes on the canvas, then a guided clarify loop turns them into real specs with acceptance criteria, not just task titles. Modular Overview. A per-project dashboard you or your agent compose from widgets: what's blocked, what needs approval, milestones, CI status, files. It's on ClawHub now, so install is basically one line: openclaw plugins install clawhub:flowboard then one command brings the dashboard up, and it self-updates from inside the dashboard after that. There's a Telegram mini app too if you want the board on your phone. Code and docs on GitHub: https://github.com/rasimme/FlowBoard If you run several agents across projects, what's the one thing that still breaks your flow? That's what I'd love to tackle next.