review-image

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

• Third-party content exposure detected (high risk: 0.90). The skill accepts arbitrary http(s) image URLs (see resolveImageInput in scripts/review-image.ts and the SKILL.md usage) and embeds that external image_url into the request body sent to the OpenRouter model (buildRequestBody), so untrusted third‑party image content is read/interpreted by the model and can directly influence CI decisions/exit codes.