assistant

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

• Third-party content exposure detected (high risk: 0.80). This skill's SKILL.md explicitly shows runtime passthrough and "shadow mode" behavior that routes inputs to a third-party teacher model (scillm / "DeepSeek V3.2 via Chutes") and returns/uses that external output as the result (e.g., "Passthrough mode" and "Shadow mode" in SKILL.md), so the agent ingests untrusted third-party responses that can materially influence decisions.