propositions

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

• Third-party content exposure detected (high risk: 0.75). The required deep-dive workflow performs “Web research (optional)” and then feeds the resulting findings (including URLs and extracted buyer/competitive language) from the research agent outputs (research/deep-dive-*-*-*.json) into the LLM context during the briefing/dialogue phases, which is outsider-authored free text from public web sources.