roadmap

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow reads thread-root authoritative inputs (seed.md, decisions.md, optional proposal.md) from docs/threads/<...>/ and then uses their contents to author roadmap.md, so if those input files were written by an outsider, their prose flows into the LLM context via the thread-resolution/input-reading path.