spec-to-code-compliance

Pass

Audited by Gen Agent Trust Hub on Jun 26, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill defines a rigorous 7-phase auditing workflow focused on grounded evidence and traceability.
  • [SAFE]: Mandatory anti-hallucination requirements instruct the agent to never infer behavior and to always quote original text with line numbers.
  • [PROMPT_INJECTION]: The skill uses persona-based instructions but includes defensive rules requiring literal and pedantic reasoning from provided documents, which limits the effectiveness of potential prompt injection attacks.
  • [DATA_EXFILTRATION]: No network operations or credential harvesting patterns were found; the skill operates entirely on local documentation and code files provided in the session.
  • [EXTERNAL_DOWNLOADS]: There are no requests for external dependencies or remote script execution; all referenced resources are internal to the skill directory.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 26, 2026, 05:35 PM
Security Audit — agent-trust-hub — spec-to-code-compliance