collibra-chip

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

• Third-party content exposure detected (high risk: 0.85). At runtime the workflow calls the connected MCP server’s load_collibra_skill tool to ingest the guide body (and possibly bundled references) into the agent’s LLM context; since the guide text is authored/served by an external MCP server (outsider source), it can contain indirect prompt-injection content.