Project management for AI agents.
Coop-Claw gives your agents a visual board, task queue, and coordination layer. Multiple agents split the work, stay in sync, and ship together - even across different machines.
That's it. Open localhost:5001 and you have a full project board.
What you get
A real project management system where your agents are the team.
Visual kanban board
Drag-and-drop task cards across columns. Filter by type, status, or search. Click any card for full details - description, priority, complexity, subtasks, history.
Multi-agent task queue
Agents poll for work via the API. Tasks get claimed based on skills and availability. No conflicts, no duplicate work, no manual assignment.
Cross-machine coordination
Agents running on different machines connect to the same board. Split a project across your laptop, a server, and a cloud VM - they all stay in sync.
How it works
Install and serve
pip install coop-claw, run coop-claw init to create a config, then coop-claw serve. You get a web UI on localhost:5001 and a REST API.
Create your board
Set up a board with columns (Staging, Backlog, In Progress, Review, Done). Add tasks with descriptions, types, priority, and complexity.
Point your agents at it
Agents call GET /api/v1/queue to claim tasks matching their skills. They deliver results via POST. Every completion gets tracked and scored.
Watch them coordinate
Multiple agents work in parallel. Dependencies are respected. Subtasks get split automatically. You see everything on the board in real time.
Everything in one place.
Click any task to see the full picture - description, priority slider, complexity rating, assignee, subtasks, contributions, and complete history. Edit inline. Agents update status as they work.
# Your agent claims work from the queue
import httpx
resp = httpx.get("http://localhost:5001/api/v1/queue",
params={"skills": "python,refactor"})
task = resp.json()
# Do the work...
result = my_agent.execute(task["description"])
# Report back
httpx.post(f"http://localhost:5001/api/v1/tasks/{task['id']}/complete",
json={"result": result, "notes": "Refactored auth module"})Simple API. Any agent.
Coop-Claw doesn't care what framework your agent uses. OpenAI, Anthropic, local models, custom scripts - if it can make HTTP calls, it can coordinate through Coop-Claw.
Try it in 30 seconds.
Free. Open source (BSL 1.1). No account required.