Use Cases

Use Cases

Use the Wiki when you already understand the project from the README and need deeper operating detail.

Best-Fit Scenarios

Scenario	What the skill should produce
Research paper planning	Research question, contribution boundary, novelty risks, experiment plan
Literature review	Search strategy, inclusion rules, literature matrix, synthesis outline
Related work revision	Claim audit, missing baselines, theme grouping, overclaim reduction
Rebuttal	Reviewer-comment classification, response strategy, evidence gaps, revision actions
Experiment planning	Dataset, metric, baseline, ablation, status, reproducibility notes
Integrity review	Fabricated-citation checks, SOTA qualification, result-status checks

Compatible Secondary Scenarios

Undergraduate thesis, graduation project, capstone, proposal, midterm report, and defense preparation can use the same evidence-first workflow, but they are not the main positioning.

For those cases, start from school or advisor requirements, then map each required chapter, figure, experiment, and deliverable to evidence.

Poor-Fit Scenarios

Do not use the skill to:

• invent citations or papers;
• make a thesis look like a completed research paper when experiments are only planned;
• guarantee acceptance, grades, or advisor approval;
• rewrite claims so they sound stronger than the evidence;
• replace human authorship, ethics, or final submission review.

Selection Rule

If the task is about academic reasoning, literature, rebuttal, experiments, or evidence, use this skill. If the task is only formatting, exporting, or document conversion, use a document or export workflow after the evidence is settled.

中文说明

这个 Skill 最适合处理需要学术判断和证据约束的任务，而不是单纯排版或文件转换。

适合场景：

场景	应产出什么
研究论文规划	研究问题、贡献边界、novelty 风险、实验计划。
文献综述	检索策略、纳入规则、文献矩阵、综述结构。
Related work 修改	Claim 审计、遗漏 baseline、主题分组、降低过强表述。
Rebuttal	审稿意见分类、回应策略、证据缺口、修改动作。
实验规划	数据集、指标、baseline、消融实验、结果状态和可复现说明。
完整性检查	虚构引用、SOTA 夸大、实验状态不实等风险。

本科论文、毕设、开题、中期、答辩可以用，但要先满足学校和导师的真实要求，再把每个章节、图表、实验和交付物映射到证据。

不适合用来虚构引用、保证录用、代替作者伦理判断，或把没做完的实验写成已经完成。