short
Fail
Audited by Snyk on Aug 19, 2026
Risk Level: HIGH
Full Analysis
HIGH W007: Insecure credential handling detected in skill instructions.
- Insecure credential handling detected (high risk: 0.80). The instruction to "rewrite your last response" requires reproducing prior output verbatim (albeit shorter), which could cause the model to include any secret values present in that prior response and thus poses a direct exfiltration risk.
Issues (1)
W007
HIGHInsecure credential handling detected in skill instructions.
Audit Metadata