Built an open source local first Kanban workflow for running AI coding agents without babysitting every step

I’ve been building BatonBot, a local first app for running AI coding workflows with less babysitting. The problem I kept running into, especially with local models, is that coding agents can be useful but the workflow gets slow: start task → wait → check output → fix next issue → run another step → wait again. BatonBot is my attempt to make that more hands off. You set up coding tasks, hand them off to agents, track progress visually in a Kanban-style board, and come back later to see what finished, failed, or needs review. It’s aimed at people using local or semi-local AI coding workflows with tools like Aider, Cline, Roo, Codex CLI, Claude Code, local LLMs, or mixed providers. I would mean a lot to me if the members from this community would pitch in/give me feedback. GitHub: [https://github.com/mdoty4/batonbot]() Website: [https://batonbot.io]()