evaluate-shortform

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

• Third-party content exposure detected (high risk: 0.85). At runtime, the skill ingests OUTSIDER-authored free text from the evaluated post itself (e.g., caption/transcripts/engagement panel text) via context.post_url/post_observation_excerpt passed into eval-runner-agent, which then places those observations into the LLM context for scoring.