feynman-perspective

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

• Third-party content exposure detected (high risk: 0.90). Step 2 ("费曼式研究") explicitly mandates using tools like "WebSearch等" to fetch real-world information and then base the agent's conclusions on that research, which requires reading and acting on open/public third‑party content and thus enables indirect prompt injection.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)