メニュー
Identify benchmark, evaluation, experiment, or comparison requirements and upstream constraints.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Convert source notes into claim-level evidence with confidence and limitations.
Verify source traceability, artifact completeness, and citation correctness before citation_qa gate.
Write an evidence-backed final research report from sources and the evidence matrix.
Review research method quality, source diversity, bias, and conclusion strength.
Convert a broad research request into subquestions, assumptions, and source criteria.
Design and execute a multi-round SearxNG/arXiv search strategy.
| name | Benchmark Research |
| description | Identify benchmark, evaluation, experiment, or comparison requirements and upstream constraints. |
| when-to-use | Use when task touches benchmark integration, model/product evaluation, performance testing, or score interpretation. |
| user-invocable | false |
| context | inline |
Report:
Do not alter benchmark definitions or scoring logic.
