parallel-literature-search

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly fetches and ingests third‑party content from PubMed (NCBI E‑utilities) and from the open web via Perplexity (web URLs/snippets) and then has the agent synthesize and interpret those results (see search_pubmed/search_perplexity_simulation and the parallel_search → synthesize_results flow), exposing it to untrusted public web content that could carry indirect prompt injections.