idea-generation-and-ideation

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

• Third-party content exposure detected (high risk: 0.75). The required workflow explicitly instructs the agent to ingest “real audience signals” by having the user paste/cluster outsider-authored text (e.g., comments/DMs/FAQs/poll results and community/support conversation text) into the expansion prompt, which then becomes LLM context (indirect prompt-injection risk).