research-powerpack

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

• Third-party content exposure detected (high risk: 0.95). The SKILL.md explicitly instructs the agent to run search_google, scrape_pages, search_reddit and fetch_reddit (see "The Five Tools" and the Research Workflow Patterns / Steps in Sections 2–3 and the Category Router) to fetch and ingest content from public sites (Google results, StackOverflow/GitHub pages, Reddit threads and other public web pages) and to read/synthesize that untrusted user-generated content into decisions and follow-up actions.