nemo-evaluator-plugin

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

• Third-party content exposure detected (medium risk: 0.30). In the required workflow, the agent can ingest outsider-authored free text from the evaluation spec dataset fields (e.g., assets/specs/llm_as_judge.json uses {{item.input}} into prompt_template.messages[].content), and the CLI accepts that spec payload as user-provided input via nemo evaluator evaluate run --spec-file / submit.