emergency circuit
PassAudited by ClawScan on May 10, 2026.
Overview
This is a disclosed safety-monitor and kill-switch skill, but the reviewed package is documentation-only and its real behavior depends on an external tool the user installs separately.
Before installing, understand that this ClawHub submission contains documentation only. If you use the external npm/GitHub tool, review that code separately, test it in sandbox mode, configure conservative policies, and confirm where monitoring logs and incident data are stored.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If configured incorrectly, the tool could stop the wrong agent or interrupt production work.
The skill documents commands that can stop monitored agents. This is purpose-aligned for an emergency safety circuit, but it is a high-impact control users should apply only to intended agents.
emergency-circuit kill --agent-id <id> [--reason <text>]
Use explicit agent IDs, test policies in sandbox mode first, and require human review for production kill-switch policies where possible.
Installing the external package would run and trust code outside the reviewed artifacts.
The reviewed ClawHub artifact is instruction-only, while the documented implementation comes from an external npm package/GitHub repository that is not included in the scan.
npm install -g openclaw-emergency-circuit
Review the npm/GitHub source, pin versions, and install in a controlled environment before granting it authority over production agents.
Logs or monitoring records could reveal agent activity, usage patterns, costs, or other operational details.
The monitoring function may collect operational logs, usage statistics, costs, and incident data. That is expected for the purpose, but it can contain sensitive operational information.
Track API calls, token consumption, and execution time; Monitor costs across providers
Check where logs are stored, limit retention, and avoid monitoring sensitive workloads unless the external implementation’s data handling is understood.