agent-reproduce-feature

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s HTTP capture path runs a nested reference agent under local mitmproxy (scripts/run_with_mitm.sh using scripts/llm_dump.py), which ingests free-text request/response bodies (e.g., model messages/content) into http.jsonl for later LLM extraction.