investigate-run
Fail
Audited by Snyk on Jun 15, 2026
Risk Level: HIGH
Full Analysis
HIGH W007: Insecure credential handling detected in skill instructions.
- Insecure credential handling detected (high risk: 0.90). The audit instructions repeatedly require quoting lines and embedding exact file/JSONL snippets and "current text" when documenting findings and proposed edits, which forces the model to reproduce any secrets that appear verbatim in those files (API keys, tokens, passwords), creating an exfiltration risk.
Issues (1)
W007
HIGHInsecure credential handling detected in skill instructions.
Audit Metadata