hume

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly shows runtime ingestion of arbitrary public URLs in "Start Inference Job from URLs" (e.g., posting "urls": ["https://example.com/media-file.mp4"]) and retrieving user-generated conversation data via EVI chat endpoints ("List Chat Events"), meaning untrusted third-party content is fetched and analyzed and could influence subsequent behavior.