exploring-replay-vision-observations

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime path reads outsider-authored free text when it calls vision-scanners-observations-list / vision-observations-list / vision-scanners-observations-get to ingest scanner_result fields (e.g., summarizer free-text summaries and monitor “open-ended observation”), which originate from the scanners’ prior LLM runs over session recordings and may include arbitrary text extracted from those recordings/events.