Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use to articulate the theoretical gap and contribution for 《南开管理评论》 (Nankai Business Review) — turning a set of construct relationships into a statement of what existing management theory cannot explain and what this paper adds (concept / proposition / framework / boundary condition). Use when the contribution still reads as "we test whether X affects Y".
.claude/skills/bilal140202-nbr-theory-gap/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 34% | 0% |
> "前人研究较少"不是缺口。缺口必须落到某个理论"欠"了什么。
| 级别 | 表述形态 | 本刊命运 | |------|----------|----------| | G0 | "X 与 Y 的研究较少" | 大概率被批"数空白不是缺口" | | G1 | "X→Y 的结论存在分歧" | 起步可用，但须解释分歧为何存在 | | G2 | "理论 T 无法解释现象 P 的黑箱/反常" | 合格，机制或边界缺口成形 | | G3 | "中国管理实践中的现象 P 迫使理论 T 修正前提" | 最契合本刊"实践问题理论化"的取向 |
向 G2/G3 升级的杠杆：先找到理论 T 的一个明确预期，再找一个与之相抵的现象或条件。
设想初稿贡献写"检验断裂带对双元创新的影响"：
就 <议题>，<理论T> 主张 <判断A>；但 <反常现象/分歧/黑箱> 难以由 T 解释（缺口）。
本文引入 <概念/机制C>，提出 <命题/路径>，论证 <核心机理>。
这对 <理论T> 构成 <扩展 / 边界条件 / 整合>，改变了我们对 <X> 的理解。本刊已刊文章引言普遍在开篇即呈现现象张力或理论分歧，并于引言后段以一段话预告贡献（常与摘要的贡献句同源）；文献回顾多并入"理论基础与研究假设"而非独立长节。以上为经验观察，以近期刊文与期刊最新投稿指南为准。
【缺口类型】解释 / 机制 / 整合-边界
【缺口一句话】<理论T 欠了什么>
【贡献形态】新概念 / 新命题 / 新框架 / 边界条件
【一句话命题】<…>
【对话理论】<T1 / T2>：扩展 / 边界 / 整合
【下一步】nbr-hypothesis-development| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 17,867 | 14,275 | -20% | 1 | 1 | 0% | 2,645 | 3,492 | +32% | 0 | 0 | — |
case-02 | fail→pass | 20,530 | 15,167 | -26% | 1 | 1 | 0% | 3,103 | 3,629 | +17% | 0 | 0 | — |
case-03 | fail→pass | 17,573 | 16,146 | -8% | 1 | 1 | 0% | 2,754 | 3,638 | +32% | 0 | 0 | — |
case-04 | pass→pass | 9,543 | 6,697 | -30% | 1 | 1 | 0% | 1,621 | 2,366 | +46% | 0 | 0 | — |
case-05 | pass→pass | 14,459 | 11,779 | -19% | 1 | 1 | 0% | 2,189 | 3,115 | +42% | 0 | 0 | — |
case-06 | pass→pass | 15,185 | 18,290 | +20% | 1 | 1 | 0% | 2,795 | 4,607 | +65% | 0 | 0 | — |
case-07 | fail→pass | 15,615 | 10,371 | -34% | 1 | 1 | 0% | 2,269 | 2,801 | +23% | 0 | 0 | — |
case-08 | fail→pass | 21,757 | 18,066 | -17% | 1 | 1 | 0% | 2,958 | 3,967 | +34% | 0 | 0 | — |
case-09 | pass→pass | 16,296 | 10,600 | -35% | 1 | 1 | 0% | 2,370 | 2,743 | +16% | 0 | 0 | — |
case-10 | pass→pass | 14,893 | 15,738 | +6% | 1 | 1 | 0% | 2,199 | 3,672 | +67% | 0 | 0 | — |
case-11 | fail→pass | 15,288 | 10,428 | -32% | 1 | 1 | 0% | 2,231 | 2,977 | +33% | 0 | 0 | — |
case-12 | fail→pass | 14,337 | 8,691 | -39% | 1 | 1 | 0% | 2,195 | 2,663 | +21% | 0 | 0 | — |
case-13 | fail→pass | 15,083 | 10,498 | -30% | 1 | 1 | 0% | 2,154 | 2,942 | +37% | 0 | 0 | — |
case-14 | fail→pass | 16,695 | 16,780 | +1% | 1 | 1 | 0% | 2,769 | 3,972 | +43% | 0 | 0 | — |
case-15 | fail→fail | 17,694 | 17,768 | +0% | 1 | 1 | 0% | 2,872 | 4,069 | +42% | 0 | 0 | — |
case-16 | fail→pass | 20,955 | 9,598 | -54% | 1 | 1 | 0% | 3,406 | 2,867 | -16% | 0 | 0 | — |
case-17 | fail→pass | 11,321 | 4,909 | -57% | 1 | 1 | 0% | 2,067 | 2,165 | +5% | 0 | 0 | — |
case-18 | fail→pass | 13,932 | 11,498 | -17% | 1 | 1 | 0% | 2,294 | 3,113 | +36% | 0 | 0 | — |
case-19 | fail→pass | 6,265 | 5,031 | -20% | 1 | 1 | 0% | 1,143 | 2,098 | +84% | 0 | 0 | — |
case-20 | fail→pass | 14,642 | 8,631 | -41% | 1 | 1 | 0% | 2,400 | 2,860 | +19% | 0 | 0 | — |
case-21 | fail→pass | 17,583 | 15,485 | -12% | 1 | 1 | 0% | 2,815 | 3,853 | +37% | 0 | 0 | — |
case-22 | fail→pass | 8,931 | 2,083 | -77% | 1 | 1 | 0% | 1,443 | 1,569 | +9% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +73 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.