Hello from Redeep, your personal coding agent with a builtin Kanban board to help you manage your work in parallel.
Use Redeep to write code, analyze data, and automate your work using the latest open source models from OpenRouter & more.
The fix: one outcome per Kanban card, one isolated git worktree per agent, and column actions that automate handoffs β while keeping a human decision gate on what ships. The board moves work forward. You decide what merges. 3/4
The core problem is coordination, not capability. Two agents edit the same file. A third finishes unnoticed. Context scatters across chat tabs. More autonomy only makes the mess move faster. 2/4
A few weeks ago we shared redeep's field guide for working with AI agents. The question that came back most: how do you run several at once without chaos? Our new guide answers it. π§΅ 1/4
As more of us work with AI agents, the difficult part will not be opening another chat window. It will be giving several agents enough context to do useful work, then keeping track of what each one is doing. That's the problem Redeep is trying to solve. π§΅ 1/5
Redeep ships with a built-in Kanban board that helps you keep track of your work and coordinate with your agents. A board card gives one worker a brief, the files it needs, and one task to finish. 2/5
Move a card to "Working" and it starts. Several cards can run at once, each in its own isolated copy of the project, so their changes don't collide. 3/5
Workers improve over time: give them reusable skills, ask focused agents to explore or review, and connect tools through MCP. Send a hard card to Codex or Claude for another opinion. They can browse the web and use your computer, too. 4/5
Our goal is to explore how people can work with AI to get real work done β with your own keys, files, and Mac, running open source models like Kimi K3 and DeepSeek. Start with the 3-part field guide β https://t.co/6LNYrCkBGi Free for macOS β 5/5