zai-search

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

• Third-party content exposure detected (low risk: 0.10). The required workflow runs search, read, and Zread repo-search/repo-file calls that return arbitrary public web pages, search snippets, and public GitHub repo content (including indexed issues and commits) that outsiders author, but the agent must select the query, URL, or repo before any of that text reaches the LLM, so exposure is low.