internal-narrative

Pass

Audited by Gen Agent Trust Hub on Apr 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No signs of prompt injection or attempts to override system safety guidelines were detected in the instructions or examples.
  • [DATA_EXFILTRATION]: The skill does not access sensitive local files (e.g., SSH keys, environment variables) and contains no network-capable code for exfiltration.
  • [REMOTE_CODE_EXECUTION]: There are no remote script downloads, package installations, or dynamic code execution patterns (eval/exec).
  • [COMMAND_EXECUTION]: The skill does not invoke shell commands or interact with the operating system beyond providing text-based communication guidance.
  • [CREDENTIALS_UNSAFE]: No hardcoded API keys, tokens, or credentials were found in the metadata or instructional content.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 5, 2026, 08:05 PM
Security Audit — agent-trust-hub — internal-narrative