tech-advisor

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow explicitly performs web searching and then fetches arbitrary selected URLs (Step 3 “Fetch key sources” and Step 4 “Evaluate/Analyze”), which means outsider-authored text from search results/review sites (including user-submitted content like Reddit) can be ingested via URL fetching.