deep-deep-research

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

• Third-party content exposure detected (high risk: 0.95). Outsider free text is ingested via Stage 3/5: WebFetch retrieves arbitrary public pages (URLs from Stage 2 search results), the fetched body text is extracted into claims, and those extracted quotes/evidence are then placed into the LLM context for Stage 5 adversarial verification prompts (prompts.py fetch → schemas.py extract → prompts.py verify).