teach

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow runs entire search "<topic>" and then entire explain --checkpoint <id> --full/--raw-transcript, which ingests repository-authored checkpoint transcripts; since those transcripts are not guaranteed to be authored by the operating user (could be other contributors), this can feed outsider-authored free text into the LLM context via the checkpoint transcript/explain path.