code-refinement
Fail
Audited by Snyk on Mar 4, 2026
Risk Level: HIGH
Full Analysis
HIGH W007: Insecure credential handling detected in skill instructions.
- Insecure credential handling detected (high risk: 1.00). The skill requires capturing and outputting "current code snippet" and inline evidence from the codebase (file paths/line ranges), which forces the agent to reproduce any secrets present in source files verbatim and thus creates an exfiltration risk.
Audit Metadata