autoresearch

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md's mandatory Bootstrap step instructs the agent to search and ingest literature from open/public sources (e.g., Exa MCP/web_search_exa, Semantic Scholar, arXiv, CrossRef) and to save and act on those findings as part of the inner/outer loop decision-making, so the agent will read third‑party web content that can materially influence its actions.