research

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow sends the user-provided “research question” text into scripts/antigravity.py (via uv run ... research --query "<question>" in Phase 2), which then runs Gemini’s managed deep-research agent that can ingest outsider-authored web content during browsing.