Peter Steinberger adds a core auto-updater with dry-run preview to OpenClaw — addressing the operational challenge of keeping always-on AI assistants running the latest code without breaking active sessions.
AI assistants that run as persistent daemons face a unique operational challenge: they're always on, often in the middle of conversations, and can't simply be stopped and restarted like typical software. Yet they need to stay current — security patches, bug fixes, and new features arrive constantly.
The f442a35 commit introduces a core auto-updater that addresses this with two key features:
The auto-updater integrates with OpenClaw's gateway restart mechanism, which already handles graceful restarts via SIGUSR1 signals. The update flow:
1. Check for available updates (npm/git)
2. Show dry-run preview of changes
3. User confirms or auto-applies based on config
4. Download and stage updates
5. Trigger graceful restart
6. New version picks up where old left off
This is particularly elegant because it builds on existing infrastructure. The gateway already knows how to restart without dropping connections — the auto-updater just needs to orchestrate when and what.
Based on the implementation, updates can be triggered via:
update.run action in gateway toolThis commit is part of a broader operational maturity push. Related commits from the same day include network consolidation (333fbb8, f14ebd7) and security hardening for local-host and tailnet CIDR checks. Together, they signal OpenClaw is moving from "developer tool" to "production infrastructure."
Expect future iterations to add: