skill-creator

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

• Third-party content exposure detected (medium risk: 0.30). The skill-creator’s eval viewer workflow (eval-viewer/generate_review.py) reads prompt text from run directories (eval_metadata.json or transcript.md) and embeds it into an HTML/JSON payload that is then consumed/visible during review, where the text originates from the outsider-authored eval prompts fed into the runs.