search

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

• Third-party content exposure detected (high risk: 0.75). This workflow’s Layer 2 External Research explicitly instructs the agent to use runtime web/code search tools (e.g., Exa for web content and grep.app/GitHub code search) to fetch outside free-text sources, which would be ingested into the LLM context via the librarian/external-research paths.