ground-the-claim

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

• Third-party content exposure detected (medium risk: 0.30). In SKILL.md, the workflow performs a “Direct capture only” step and then “Read the whole source yourself” (including for the Surface pass and each /grounding session), so outsider-authored free text can be ingested as the user-named/clipped source without selecting a specific trusted item first.