clarify-requirements

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

• Third-party content exposure detected (high risk: 0.85). Yes—Step 2/3 collects the operating user’s free-form answers to the agent’s questions (“Reply in free form”) and then Step 4 compiles “Clarifications obtained” (including the user’s text) into the markdown context that is passed to requirements-generator, so the LLM ingests outsider-authored free text from the user’s responses.