analyze

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs the agent to gather external evidence (e.g., "Use Read, Grep, Glob, WebSearch", "search the web", and "read the code, search the web, run a query" in the Evidence grounding / Phase 1 sections), which requires fetching and interpreting open/public third‑party content that can materially influence analysis and next actions.