voice-extraction

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

• Third-party content exposure detected (high risk: 0.90). SKILL.md:23-28 and 35-82 describe the operating user pasting “posts, articles, or tweets” (outsider-authored free text) which is then inserted as “[FULL PASTED CONTENT]” into the prompts for 10 parallel agents at runtime, so the LLM ingests outsider text.