| name | multi-device-agentic-mesh |
| description | Coordinate OpenClaw tasks across multiple devices/nodes with deterministic routing, failover, and handoff rules. Use when users run OpenClaw on laptop/desktop/VPS combinations and need resilient cross-device execution. |
Multi-Device Agentic Mesh
Coordinate work across multiple OpenClaw nodes safely.
Baseline discovery
Start with node and session visibility checks:
openclaw status
openclaw nodes list
openclaw sessions list
If a command is unavailable in a deployment, use the nearest supported status/list command.
Build a routing table
Classify each node/device by:
- trust level (high/medium/low)
- latency/availability
- tool access and capabilities
- data sensitivity suitability
Use this table to route tasks to the best node.
Routing policy
Apply these defaults:
- Send sensitive tasks to highest-trust local node.
- Send long-running/background tasks to stable always-on node.
- Keep user-facing replies near the active user channel/session.
Failover
If a primary node fails:
- retry once
- reroute to secondary node
- report failover event and degraded mode
Never hide failover from the user.
Session handoff
When moving work across devices:
- summarize current state
- include pending actions
- include exact next command/message
- confirm handoff destination
Guardrails
- Do not route sensitive data to untrusted nodes.
- Ask before changing long-lived routing defaults.
- Keep decisions auditable (what was routed where and why).
