ClawHub

Gateway Watchdog

v1.0.0

Production-grade bash watchdog for the OpenClaw gateway. Runs via launchd every 5 minutes. Handles boot grace periods, progressive retry with backoff, port-l...

⭐ 0· 94·0 current·0 all-time
byBryan Tegomoh, MD, MPH@bryantegomoh
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
ℹ
Purpose & Capability
The script's actions (HTTP health check, port check, PID/age checks, and restart via launchctl) match the described purpose of a gateway watchdog. Minor inconsistency: registry metadata lists 'Required binaries: none' while the SKILL.md/script rely on macOS tools (launchctl, lsof, date -j) and curl; those runtime dependencies are expected but not declared in metadata.
✓
Instruction Scope
SKILL.md and the script confine activity to local checks and local restarts: curl against http://127.0.0.1:18789, lsof for port binding, ps/date for process age, and launchctl to restart the ai.openclaw.gateway job. Logs are written to ~/.openclaw/logs/watchdog.log and a cooldown file in /tmp. There are no external network endpoints or data exfiltration behaviors.
✓
Install Mechanism
This is an instruction-only skill with an included shell script; there is no install spec that downloads or writes arbitrary code to disk beyond the provided script. Risk from installation is low (user-controlled plist creation + launchctl load).
✓
Credentials
The skill requests no credentials or environment variables. It uses $HOME and standard system tools; no secrets are accessed or transmitted. The script acts only on a local service label (ai.openclaw.gateway), which is appropriate for a watchdog.
ℹ
Persistence & Privilege
The skill is not 'always:true'. It can be invoked by the agent autonomously (default), and the script calls launchctl kickstart to restart the gateway job. Restarting another launchd job is a legitimate high-impact action for a watchdog, so users should be aware that installing and enabling this skill grants the agent the ability to restart the gateway process in the user's session.
Assessment
This skill appears to do what it says: local health checks and restart of the OpenClaw gateway via launchd. Before installing: (1) Verify the script path in the provided plist points to the actual location you intend to run; (2) ensure required binaries exist on your machine (launchctl, curl, lsof, and macOS date -j); (3) run the script manually once and inspect ~/.openclaw/logs/watchdog.log to confirm behavior; (4) confirm the launchd label ai.openclaw.gateway matches your gateway job and you want an automated restart capability; (5) because the source/homepage is unknown, review the script contents yourself and store it from a trusted location or repository before enabling it. If you do not want the agent to autonomously restart the gateway, avoid enabling autonomous invocation or run the plist under a controlled user account.

Like a lobster shell, security has layers — review code before you run it.

latestvk974xr3jrjbm1ew3983kg7x5f983bb9j

License

MIT-0
Free to use, modify, and redistribute. No attribution required.
Termshttps://spdx.org/licenses/MIT-0.html

Comments