Coaching Analysis

Fail

Audited by Snyk on Mar 3, 2026

Risk Level: HIGH
Full Analysis

HIGH W007: Insecure credential handling detected in skill instructions.

  • Insecure credential handling detected (high risk: 0.80). The prompt requires including full dialogue and exact verbatim quotes/timestamps from transcripts and deal metadata, so if those transcripts or context contain API keys, tokens, cookies, or passwords the model would be instructed to reproduce them verbatim, creating an exfiltration risk.
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 3, 2026, 11:37 AM