assetseeker

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

• Third-party content exposure detected (medium risk: 0.30). In SKILL.md the runtime search workflow takes the user’s keyword string (outsider-authored free text) and uses it to query external asset sources’ APIs/scrapers (e.g., Pexels/Pixabay/Unsplash/Iconify/Freesound/Mixkit), causing the agent to ingest attacker-influenced result metadata/content during “Step 3 — Search” and “Step 4 — Present results”.