evaluating-llms-harness

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

• Third-party content exposure detected (high risk: 0.85). The required runtime workflow is evaluating benchmark tasks (e.g., MMLU/GSM8K/HumanEval) whose dataset examples are loaded/ingested as prompt text; these benchmark documents are authored by third parties (outsider) and are fed into the LLM via the harness’s task prompt construction (doc_to_text/doc_to_target) during evaluation.