prompt-engineering-refine

Warn

Audited by Snyk on Aug 25, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (medium risk: 0.30). The runtime workflow described in SKILL.md (“Write → Test → Analyze → Improve”) uses a user-provided prompt/fixture set as inputs for evaluation, meaning outsider-authored free text can be ingested into the agent’s test loop without selecting a specific item first.

Issues (1)

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 25, 2026, 06:54 PM
Issues
1
Security Audit — snyk — prompt-engineering-refine