policy-starter

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

• Third-party content exposure detected (high risk: 0.85). The required workflow includes “Web search for the current published model policies and guidance” and then “read the model policies” from public sources at runtime, so outsider-authored free text from fetched web pages is ingested into the agent’s LLM context for drafting.

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

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