higgsfield-to-figma

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow ingests outsider-authored free text via the user-provided inline brief/angle brief inputs (Step 2: “This skill takes angles, not finished copy… If no brand files exist, collect a short inline brief”), which the agent then uses to generate the ad copy and variant grid (Step 7).