datahub-evals

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

• Third-party content exposure detected (medium risk: 0.30). At runtime, the skill uses acryl-datahub-cloud evals list/get to fetch eval questions/conditions from DataHub and then passes that untrusted eval “question” free text into the answering agent via claude --print "<the eval's question>", before reporting the agent’s output for judging.