菜单
Use when the research task is in material — atomistic simulation, interatomic potentials and materials property prediction — at study design, analysis or writing. Run the protocol as specified as the foreground result, and leave every stage's default diagnostic panel
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
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。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/tangxiangru/AutoR --skill material-as-specified-run-and-stage-diagnostics命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Use when writing results and discussion, and when a task or a reviewer asks why an effect happens rather than whether it does. Covers the difference between reporting an effect and accounting for it, and what a mechanism claim needs behind it.
Use when the supplied data looks synthetic, degraded, incomplete or wrong, and whenever you are tempted to reframe the study around what you found about the inputs, the harness or the evaluation. Covers what to do with a real data problem without losing the study.
Use when planning figures, at study design and again before writing. Covers the figures a paper in this field is expected to contain, why an original figure does not substitute for a standard one, and how to decide what to draw first.
基于 SOC 职业分类
正在显示 SKILL.md
| name | material-as-specified-run-and-stage-diagnostics |
| description | Use when the research task is in material — atomistic simulation, interatomic potentials and materials property prediction — at study design, analysis or writing. Run the protocol as specified as the foreground result, and leave every stage's default diagnostic panel |
Materials pipelines are graded stage by stage, so run the protocol exactly as specified before you improve it: the given parameter values, cell sizes, sample counts, budgets and convergence thresholds. That run is the foreground result, reported in the protocol's own units. Nearly every supplied specification has a defect you will find; the audit belongs in a clearly separated second analysis with the delta attributed, and must not take the title, the abstract's first sentence, or panel (a). A lead figure captioned with what is wrong with the spec is read as a declined reproduction even when the correct number sits in a table two sections later.
Emit each stage's default diagnostic panel even when a deeper analysis supersedes it: input characterisation (N per split, class balance, target range, replicate noise), training and validation objective versus epoch, held-out metric versus step, parity plot against the reference with the identity line, best-so-far versus evaluation count, generated-versus-reference scatter in the physical parameter space. These panels are cheap, conventional, and their absence is unrecoverable.
When the specification names a method family -- a particular surrogate, optimiser, simulator or architecture -- run that one and report it, then place your alternative beside it rather than instead of it. When the final validation needs an instrument you lack (synthesis, a measurement, a high-fidelity simulation), do not drop the stage: report the best available proxy, label it a proxy, quantify its uncertainty, and give it its own panel.
Targets the four Material mechanisms that produce 0-35 scores with no absence: task substitution (all four bare runs detected a defective spec and made the audit the headline, so the demanded number existed nowhere), framing inversion (the correct result relegated behind a 'the spec is broken' lead panel scored 25), the never-plotted standard diagnostic (a weight-0.5 criterion scored 5 and 0 because loss- and accuracy-versus-epoch traces were judged too boring to draw), and the silently dropped wet-lab/expensive-simulation stage.