trim-language

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

• Third-party content exposure detected (medium risk: 0.65). This skill feeds repo-authored free prose (including outsider-authored docs/issues content if present in the user’s project) into the LLM via the Phase 2 worker prompts (“Reads its batch files — issue the Read/Grep calls for a batch…”) and via the Phase 3 reviewer receiving the full findings draft plus cited source context; nothing in the workflow restricts that prose to only the operating user’s authorship.