recipe-blueprint

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow’s “Context Reading” step ingests multiple repo documents (e.g., vision, design principles, personas, opportunities, journeys, learnings) that are outsider-authored relative to the operating user’s own contributions, so their prose can flow into the agent’s LLM context at runtime.