菜单
Align parsed baseline and candidate Darshan metrics before ranking deltas.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | align_hpc_trace_metrics |
| title | Align HPC Trace Metrics |
| description | Align parsed baseline and candidate Darshan metrics before ranking deltas. |
Compare only metrics that are present on both sides unless the absence itself is the evidence. Normalize field names before comparing, and keep units visible in the response.
Important aligned metrics include runtime, POSIX read/write bytes, POSIX read/write time, POSIX metadata time, MPI-IO collective writes, MPI-IO independent writes, and transfer-size hints.
Compare scientific variables with units, ranges, and quality caveats.
Inspect dataset structure before analysis or visualization.
Evaluate data quality from observed structure and summary evidence.
Choose visual checks that match the inspected data.
Route dataset questions to structure, analysis, or visualization experts.
Choose a plot that confirms numeric conversion integrity.
