log-diagnosis

Pass

Audited by Gen Agent Trust Hub on Jun 25, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill operates entirely using local scripts and standard AWS command-line interfaces for log retrieval and analysis. No remote code execution or data exfiltration risks were identified.- [PROMPT_INJECTION]: The skill ingests and processes untrusted data from external CloudWatch logs. This constitutes an indirect prompt injection surface where malicious log entries could attempt to trick the analyzing agent into producing misleading reports or ignoring specific errors. However, the use of regex patterns and structured output schemas provides significant mitigation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 25, 2026, 05:41 AM
Security Audit — agent-trust-hub — log-diagnosis