Openclaw Skill Observability

Provides tools to monitor OpenClaw health by reporting recent errors and estimating API usage costs over the last 24 hours.

This skill appears to implement exactly what it says: cost estimates from OpenClaw sessions and recent error lines from the OpenClaw system unit. Before installing: 1) Verify the host has the 'openclaw' CLI and systemd/journalctl available — the manifest did not declare these required binaries. 2) Be aware that returned system logs and session metadata may include sensitive data (IDs, stack traces, or secrets logged by other components); avoid sending these outputs to public channels or non-trusted chat endpoints. 3) If you deploy to production, test in a safe environment first and review the GitHub repository (package.json points to a repo) and the author (erain). 4) If you want stricter behavior, request the author add explicit required-binaries to the skill manifest and/or add sanitization of log lines to redact secrets before returning them.

Type: OpenClaw Skill Name: openclaw-skill-observability Version: 0.1.0 The skill is classified as suspicious due to its use of `child_process.exec` in `index.mjs` to execute the `journalctl` system command. While the command `journalctl --user -u openclaw-gateway -n 100 --no-pager` is restricted to the `openclaw-gateway` unit and filtered for errors, granting an AI agent the ability to execute system commands and access system logs is a high-risk capability, even if plausibly needed for observability. There is no clear evidence of intentional malicious behavior like data exfiltration to external endpoints or persistence mechanisms.