z-ai-api

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

• Third-party content exposure detected (medium risk: 0.30). The required runtime workflow supports a web_search tool that ingests web page titles/contents (“web_search” result content) into the model via the /chat/completions tool_calls path, where outsider-authored text can be returned from search results without the agent pre-selecting a trusted item.