openclaw-audit-watchdog

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: Comprehensive analysis of the instructions and supporting scripts confirms that the skill behaves as a legitimate security auditing tool. It uses standard platform features like 'openclaw cron' for scheduling and performs no hidden or malicious actions.
  • [EXTERNAL_DOWNLOADS]: Fetches installation artifacts from the author's official GitHub repository at prompt-security/clawsec. This is a trusted vendor resource for this skill.
  • [COMMAND_EXECUTION]: Orchestrates the 'openclaw' CLI to execute security audits and manage job scheduling, and utilizes 'node' for processing JSON results and rendering reports.
  • [DATA_EXFILTRATION]: Transmits audit reports to user-configured external targets (DM channels and email addresses). This network activity is the primary intended function of the watchdog.
  • [PROMPT_INJECTION]: The skill ingests untrusted data from 'openclaw security audit' results (scripts/run_audit_and_format.sh). It lacks explicit boundary markers in the final report, but provides basic header sanitization in its delivery scripts (scripts/send_smtp.mjs) and maintains a low privilege profile by focusing on report generation and delivery.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 02:49 AM