autoresearch

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

• Third-party content exposure detected (high risk: 0.90). The skill's Phase 1.4b "Web search" option and the THINK step explicitly instruct the agent to search, read, and log web results (e.g., library docs, Stack Overflow, benchmarking guides) and to use those third‑party pages to form hypotheses and decide experiments, so untrusted public web content can directly influence tool use and actions.