skill-creator

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

• Third-party content exposure detected (medium risk: 0.30). The eval-viewer reads and embeds arbitrary free text from the workspace’s execution artifacts—specifically transcript.md and files under each run’s outputs/ (e.g., .txt/.md/.json/.csv/.py/.html/.css)—into the served HTML via eval-viewer/generate_review.py and therefore exposes an outsider-authored prompt-injection surface whenever those artifacts contain outsider text.