openclaw-multi-agent - Configure multiple agents with routing and isolation policies
Set up multiple agents with message routing, sandboxing, and per-agent tool policies
Tags
Updated: 2026-03-09Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Create agent workspaces
- Configure message routing bindings
- Set per-agent sandbox policies
- Manage agent tool permissions
- Isolate agent sessions and state
Inputs
- Agent workspace directories
- Message routing bindings configuration
- Sandbox mode settings
- Tool allow/deny lists
- Model registry configurations
Outputs
- Multiple agent configurations
- Routing binding rules
- Agent isolation policies
- Per-agent tool policies
- Agent workspace structure
Requirements
- OpenClaw runtime environment
- Workspace directory permissions
- Agent state directory access
- Model API access credentials