Chapter 9: Node Network
A "node" in OpenClaw is any device paired with your Gateway—phone, tablet, laptop, remote server—that can execute commands or provide data. The node network extends OpenClaw's reach beyond a single machine, enabling distributed monitoring, remote command execution, camera access, location tracking, and system notifications across all your devices. The nodes tool is the interface: status lists all paired nodes with connection status and capabilities; describe returns a specific node's capabilities, OS, and available actions. Pairing requires installing the OpenClaw node agent on the device and registering it with the Gateway.
The most powerful node capability is running shell commands on remote machines. The run action executes commands on a named node: nodes: run: node: "home-server", command: ["df", "-h"] or multi-command via ["bash", "-c", "df -h && free -h && uptime"]. Use cases include distributed monitoring (cron job checking disk and memory across all server nodes hourly), remote deployment (git pull + npm install + pm2 restart on a production server), and distributed data collection (running a metrics script across multiple nodes and combining outputs). Camera access on paired mobiles enables programmatic photo capture: camera_snap with facing: "front" or "back", useful for visual monitoring (phone aimed at a server room, cron checks every 30 minutes, alerts on red/amber lights), document capture (photo a receipt, extract text via vision, parse amount and merchant, log to expenses JSON), and video clip recording (camera_clip with duration in milliseconds).
Screen recording captures the current display for visual testing, documentation, and bug reports. Location services (location_get with accuracy: coarse/balanced/precise) enable context-aware automation: polling every 10 minutes, comparing coordinates to home location, triggering "arrived home" or "left home" routines. The notification dispatcher sends native push notifications to devices via nodes: notify: node: "my-iphone", title: "...", body: "...", delivery: "system"—bypassing WhatsApp for system-native alerts. The real-world home office node setup coordinates mac-studio, macbook-pro, iphone-15, and home-server: morning startup checks all machines are running and backup completed, location determines travel vs. home briefing, with remote work monitoring hourly checking uptime and disk on the home server.
Key Patterns
- **Distributed monitoring cron:** Every hour, run `df -h && free -h` across all server nodes; alert on disk > 80% or memory > 90%
- **Camera as sensor:** Paired phone aimed at something to monitor; cron checks analyze images for anomalies
- **Location-based context:** Polling node location every 10 min; compare to home coordinates; trigger routines on geofence events
- **Native push notifications:** `nodes: notify` delivers system notifications directly, bypassing WhatsApp
- **Remote deployment via nodes:** git pull + install + restart commands on remote server nodes from the router
Related Concepts
- [[kelly-handbook-ch6-cron]] for the scheduling that drives node monitoring tasks
- [[kelly-handbook-ch5-communication-automation]] for alert routing and WhatsApp vs. native notifications
- [[kelly-handbook-ch10-browser]] for screen recording as a debugging tool
- [[kelly-gas-town-gap-analysis]] for comparing distributed agent architectures across factories