silent-failure-hunter
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- No Security Issues Detected (SAFE): The skill performs its stated function of hunting for silent failures in code using local search tools like
grep. No data exfiltration, remote code execution, or credential exposure patterns were found. - Indirect Prompt Injection Surface (LOW): As a code auditor, the skill reads external source files. While malicious code could theoretically attempt to influence the agent's report (Indirect Prompt Injection), this is an inherent property of auditing tools and is considered a low risk managed by model guardrails.
Audit Metadata