Skip to main content

3.10 OpenClaw

OpenClaw is an on-board agent that runs on RDK hardware. Enable AI assistant capabilities on the device by configuring it under AI Capabilities → On-device Agent.

On-device Agent page: after connecting RDK X5, view deployment status, model settings, and quick actions

Suggested workflow

StepWhat to do
1Confirm the current device is an RDK device and is online
2Open the On-device Agent page and check the header status
3If not installed yet, follow the page to check prerequisites, then deploy
4After deployment, verify model configuration works
5Once status is OK, use on-board chat, skills, or message channels

What you can do on this page

AreaPurpose
Header statusSee whether on-board Agent, models, and network are ready
Deploy & connectionDeploy, redeploy, or uninstall OpenClaw
Diagnose & fixReview warnings and follow suggested actions
ModelsConfigure models OpenClaw can use
Feishu integrationSet up Feishu bot app and pairing
On-board chatAfter OpenClaw is ready, talk to the on-board assistant directly

When you need OpenClaw

ScenarioRecommendation
Short-term debugging with your PC always onlineMoss plus SSH is enough
Need the on-board agent to handle device-centric tasksDeploy OpenClaw
Need to deploy skills to the deviceDeploy OpenClaw, then sync from Skill Workshop
Non-RDK Linux hostOpenClaw deployment not offered by default—prefer local Moss skills

Before deployment

  • The device must be added over SSH and be online.
  • The device network must reach required files during deployment.
  • Enough free space on the board for installation.
  • Any model OpenClaw uses must be reachable from the board; a PC-local Ollama URL is usually not reachable from the board.

If OpenClaw is missing, the page guides you step by step. Actions that affect device state (deploy, uninstall, redeploy) always ask you to confirm first.

Confirm the target device and scope of impact before any operation that writes files, runs commands, or restarts services.

Further reading