search

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

• Third-party content exposure detected (high risk: 0.85). This skill’s required workflow is to search across connected MCP sources (e.g., chat/email/cloud/project tracker/knowledge base) and then synthesize the returned results into the agent’s response, which implies the LLM will ingest outsider-authored free text from those sources at runtime via the MCP search/read results.