skills/softspark/ai-toolkit/panic/Gen Agent Trust Hub

panic

Pass

Audited by Gen Agent Trust Hub on Jun 21, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a project-local emergency stop mechanism by creating a lockfile at .claude/HALT. This is a documented pattern for agent control and does not involve any dangerous operations, data exfiltration, or malicious persistence.- [COMMAND_EXECUTION]: The skill uses standard shell commands (touch, rm) to manage the lockfile. These operations are restricted to the local filesystem and are consistent with the skill's stated purpose of providing a kill switch.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 21, 2026, 02:41 PM
Security Audit — agent-trust-hub — panic