cloudflare-images

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly ingests arbitrary public URLs (e.g., "Any publicly accessible URL can be transformed" in references/transformations.md, fetch examples like fetch("https://example.com/image.png") in SKILL.md and references/binding.md, and upload-via-URL in references/upload.md), so it clearly fetches untrusted third-party content as part of its runtime workflow which can influence processing decisions.