uncertainty-verification

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs a research subagent to fetch and ingest public web content via Context7/web fetch (mcp_fetch_fetch or fetch_webpage), citing sources such as developer.mozilla.org, github.com, and wikipedia.org, so the agent will read untrusted, third‑party (including user‑generated) pages as part of its workflow.