feature-intake

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

• Third-party content exposure detected (high risk: 0.75). The required workflow explicitly reads existing GitHub issues/bodies/comments and project state (treated as untrusted context) and then includes that textual content in the agent’s LLM context for duplicate detection and drafting (e.g., via “Search For Duplicates And Nearby Work” and “Gather Context”).