azure-ai-contentsafety-py

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

• Third-party content exposure detected (high risk: 0.70). The SKILL.md explicitly shows ingesting external content (e.g., the "Image from URL" example using ImageData(blob_url="https://example.com/image.jpg")) and the skill analyzes user/third-party text and images whose analysis results (e.g., blocklist hits, severity) are used to drive decisions, so untrusted public content could indirectly influence agent behavior.