autoresearch

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

• Third-party content exposure detected (high risk: 0.75). The required runtime workflow includes “Search literature” via web-based tools (e.g., web_search_exa / arXiv / CrossRef) and then “Save everything to literature/”; any fetched public web pages or scraped content would be outsider-authored free text ingested into the agent’s LLM context through the literature-search/scraping path.