visualisera

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

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md explicitly requires web research in Create mode Step 2—"Search for design context" and "3-5 targeted searches. Read promising results deeply"—which directs the agent to fetch and read open/public third-party content (web pages, design references) that the agent must interpret and use to make design decisions, creating a clear avenue for indirect prompt injection.