Community Edition
AI agent teams you can run locally
Clawborate is a platform for creating and managing collaborative AI agent teams. Define roles, assign tasks through a coordinator, and watch specialists work together — each with its own identity, memory, and tools inside a disposable Docker environment.
How it works
Coordinator-led teams
A master agent receives your requests, breaks work into tasks, and routes them to the right specialists. You talk to one coordinator; the team handles the rest.
Specialized agents
Each agent has its own role, identity, memory, and workspace. The default ACE template ships with coordinator, developer, designer, and tester agents ready to collaborate.
Sandboxed runtime
Agents run in isolated Docker containers spawned by the backend. Each container connects to your LLM, loads its personality from agent-arch, and communicates over a shared notification bus.
Chat through Open WebUI
Conversations happen in Open WebUI, embedded in the dashboard. Messages flow through the backend so agents can reply, update task status, and coordinate with each other.
Getting started
The community edition runs entirely with Docker Compose. After docker compose up -d, you can:
- 1Create a team from a template (e.g. ACE Development Team)
- 2Wait for agent containers to pass health checks
- 3Configure your LLM provider in Settings
- 4Chat with the coordinator and assign work to the team
Customize your agents
Agent definitions live in agent-arch/openclaw/. Edit each agent's identity, personality (SOUL.md), and memory, or add new team templates. Changes take effect when you create a new team. The directory is mounted read-only into the backend container.
Ready to spin up a team?
Sign in, pick a template, and start coordinating your first multi-agent workflow in minutes.