paoding

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md:30-39 describes a runtime path where the agent runs scripts/collect.sh on user-supplied URLs, and that script downloads/reads outsider-authored web content (video/audio transcripts and info.json metadata) into local text files that are then “喂回庖丁” for LLM processing.