Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when targeting 《中国工业经济》(China Industrial Economics) — a top journal for industrial organization, firm behavior, regional/innovation/digital economy with empirical rigor and strong policy relevance. Apply when a paper needs exhaustive robustness, clean China-policy quasi-experiments, mechanism + heterogeneity, and actionable industrial-policy implications. Famous for methodological rigor norms.
.claude/skills/brycewang-stanford-china-industrial-economics/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 48% | 0% |
国内产业经济 / 企业与创新实证顶刊。以"规范实证"著称——它甚至发表过被高频引用的方法操作指南（中介效应、双重差分等的规范流程）。编委口味：识别要干净、稳健性要做满、机制+异质性要齐、政策含义要落地。 比《经济研究》更"实证工程化"，比《管理世界》更偏经济学计量。
china-economic-quarterly（《经济学（季刊）》） / china-economic-studies（《中国经济问题》） / china-public-administration-review（《公共管理评论》） / china-rural-economy（《中国农村经济》）。若这些刊物的读者对象更贴近，不要因为名称、地区或变量相似而强行投本刊。../../resources/source-basis.md 和 ../../resources/official-source-map.md，从其中的官方来源锚点或同一主办/出版体系入口开始核验，并在回答中说明核验日期。【匹配度】高 / 中 / 低
【识别】DID/event-study 合规性：平行趋势 □ 安慰剂 □ 异质性稳健 □
【机制】分渠道/调节 √ | 仅三步中介（需改）
【异质性维度】<已切维度>（建议补 <维度>）
【稳健性清单】已做 <…> / 建议补 <…>
【政策可操作性】高 / 中 / 低| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 39,972 | 35,541 | -11% | 1 | 1 | 0% | 4,650 | 5,495 | +18% | 0 | 0 | — |
case-02 | fail→pass | 33,075 | 37,272 | +13% | 1 | 1 | 0% | 4,005 | 5,694 | +42% | 0 | 0 | — |
case-09 | pass→pass | 31,919 | 33,328 | +4% | 1 | 1 | 0% | 3,365 | 5,109 | +52% | 0 | 0 | — |
case-20 | pass→pass | 24,765 | 25,795 | +4% | 1 | 1 | 0% | 3,390 | 4,672 | +38% | 0 | 0 | — |
case-03 | fail→fail | 41,878 | 34,859 | -17% | 1 | 1 | 0% | 4,191 | 5,259 | +25% | 0 | 0 | — |
case-04 | fail→pass | 19,223 | 15,974 | -17% | 1 | 1 | 0% | 2,532 | 3,449 | +36% | 0 | 0 | — |
case-05 | pass→pass | 27,165 | 20,463 | -25% | 1 | 1 | 0% | 2,936 | 3,542 | +21% | 0 | 0 | — |
case-06 | pass→pass | 27,550 | 27,688 | +1% | 1 | 1 | 0% | 2,904 | 4,237 | +46% | 0 | 0 | — |
case-07 | pass→pass | 28,622 | 21,547 | -25% | 1 | 1 | 0% | 3,030 | 4,266 | +41% | 0 | 0 | — |
case-08 | pass→pass | 29,942 | 27,227 | -9% | 1 | 1 | 0% | 3,403 | 4,550 | +34% | 0 | 0 | — |
case-10 | pass→pass | 36,335 | 28,941 | -20% | 1 | 1 | 0% | 4,056 | 4,604 | +14% | 0 | 0 | — |
case-11 | pass→pass | 26,946 | 28,080 | +4% | 1 | 1 | 0% | 3,034 | 4,413 | +45% | 0 | 0 | — |
case-12 | fail→pass | 31,008 | 31,017 | +0% | 1 | 1 | 0% | 3,604 | 5,096 | +41% | 0 | 0 | — |
case-13 | pass→pass | 28,872 | 23,975 | -17% | 1 | 1 | 0% | 3,108 | 3,973 | +28% | 0 | 0 | — |
case-14 | fail→pass | 34,557 | 20,884 | -40% | 1 | 1 | 0% | 3,595 | 3,663 | +2% | 0 | 0 | — |
case-15 | pass→pass | 27,611 | 28,063 | +2% | 1 | 1 | 0% | 2,696 | 4,166 | +55% | 0 | 0 | — |
case-16 | pass→pass | 26,221 | 22,232 | -15% | 1 | 1 | 0% | 2,549 | 3,772 | +48% | 0 | 0 | — |
case-17 | fail→pass | 29,681 | 28,561 | -4% | 1 | 1 | 0% | 3,092 | 4,582 | +48% | 0 | 0 | — |
case-18 | pass→fail | 23,647 | 24,771 | +5% | 1 | 1 | 0% | 3,310 | 4,659 | +41% | 0 | 0 | — |
case-19 | pass→pass | 11,271 | 18,927 | +68% | 1 | 1 | 0% | 1,902 | 3,637 | +91% | 0 | 0 | — |
case-21 | fail→fail | 32,676 | 39,268 | +20% | 1 | 1 | 0% | 4,085 | 6,247 | +53% | 0 | 0 | — |
case-22 | fail→pass | 30,294 | 26,205 | -13% | 1 | 1 | 0% | 3,406 | 5,042 | +48% | 0 | 0 | — |
case-23 | pass→pass | 29,104 | 24,549 | -16% | 1 | 1 | 0% | 3,186 | 4,826 | +51% | 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. 23 cases were attempted. The headline lift of +22 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.