event-prospecting

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

• Third-party content exposure detected (high risk: 0.85). Yes—Step 5/7/8 subagents call node {SKILL_DIR}/scripts/extract_page.mjs "<url>" and browse cloud search, which fetch outsider-authored web page text at runtime and then the subagent LLM uses that extracted text to generate markdown (indirect prompt injection risk).