code-audit

Fail

Audited by Snyk on Mar 2, 2026

Risk Level: HIGH
Full Analysis

HIGH W007: Insecure credential handling detected in skill instructions.

  • Insecure credential handling detected (high risk: 1.00). The prompt mandates "MUST quote actual code from Read tool output" and produce PoCs/reports based on repo contents, which would force the LLM to reproduce any embedded secrets (API keys, tokens, or passwords) verbatim if they exist in the codebase.
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 2, 2026, 07:07 AM