diagnose

Pass

Audited by Gen Agent Trust Hub on Sep 13, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements safety-first protocols for debugging, including mandatory redaction of secrets, tokens, and authentication headers from command outputs and captured artifacts.
  • [SAFE]: The instructions provide a proactive defense against indirect prompt injection by explicitly telling the agent to treat error messages, stack traces, and CI logs as untrusted data and to ignore any commands or URLs contained within them.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 13, 2026, 09:31 AM
Security Audit — agent-trust-hub — diagnose