Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when deciding which cre-* sub-skill to invoke next, or when sequencing manuscript work from topic selection through rebuttal for 《中国农村经济》 (China Rural Economy). Routes — does not replace — the specialized skills.
.claude/skills/brycewang-stanford-cre-workflow/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -7% | 0% |
这是路由器，它不替代任何专用 Skill。它告诉你当前阶段应该用哪一个 cre- skill。
默认假设：除非用户明确说明目标期刊不是《中国农村经济》，否则按 CSSCI 一区、中国社会科学院农村发展研究所主办的《中国农村经济》编委口味处理。该刊聚焦"三农"（农业、农村、农民），以微观农户 / 村级数据的因果识别实证为主，姊妹刊为《中国农村观察》。
| 当前症状 | 下一个 Skill | |----------------------------------------------|---------------------------| | 选题想法模糊，不确定是否契合"三农"场景 / 边际贡献写不出 | cre-topic-selection | | 文献综述缺三农领域中文经典 / 缺理论文献 | cre-literature-review | | 实证只有 OLS / 描述统计，担心识别策略不过关 | cre-identification | | 主结果有了，但没有农户微观机制 | cre-mechanism | | 没切异质性 / 切分维度脱离农村场景 | cre-heterogeneity | | 表格列数过多 / 注释不规范 / 不像《中国农村经济》风格 | cre-tables-figures | | 文末政策含义脱离乡村振兴 / 是"加强完善推进"四件套 | cre-policy-implication | | 摘要写成套话，没量化结果 / 中英文不对齐 | cre-abstract | | 通篇空话套话 / 政策建议是"加强完善推进"四件套 | cre-style | | 准备投稿，需要 checklist | cre-submission | | 收到 R&R，需要写回复信 | cre-rebuttal |
cre-topic-selection — 先把"三农场景 + 理论贡献 + 中国农村政策现实"定下来cre-literature-review — 中外并重，三农经典与权威文献必引cre-identification — 农户 / 村级数据的准实验识别cre-mechanism — 农户微观机制路径cre-heterogeneity — 异质性切分（落到农村场景）cre-tables-figures — 主表 / 主图最终化（三线表）cre-policy-implication — 乡村振兴 / 农村政策含义（意义层但落到三农场景）cre-abstract — 摘要五句法 + 黑名单短语清除cre-style — 全文语言风格 polishcre-submission — 投稿前 preflightcre-rebuttal — 外审后> cre-abstract 与 cre-style 是后段 polish 阶段触发，不要在识别策略未立住时去做。
把下面的快照复制到工作笔记，每完成一阶段更新对应行；第一处空白即当前瓶颈，对应右侧的 cre- 入口：
text《中国农村经济》稿件阶段快照 选题定位：三农场景=____（农户/村级/县域） 边际贡献一句话=____ 文献对话：三农中文经典已引 __ 篇；理论/英文对话文献已引 __ 篇 识别策略：数据=____ 方法=____ 内生性处理=已/未 机制检验：农户微观路径共 __ 条，已实证检验 __ 条 异质性 ：切分维度=____（须落回农村场景，忌只切东中西） 表格规范：主表为三线表？是/否 注释含标准误聚类层级？是/否 政策含义：对应的农村政策抓手=____（禁"加强完善推进"四件套） 摘要风格：量化结果已写入摘要？是/否 "具有重要意义"命中 __ 处（目标 0） 投稿状态：submission checklist 已过 / 外审意见 __ 份待回复
cre-topic-selectioncre-literature-reviewcre-identificationcre-identificationcre-mechanismcre-heterogeneitycre-tables-figurescre-policy-implicationcre-abstractcre-stylecre-submissioncre-rebuttal两刊均由中国社会科学院农村发展研究所主办，但侧重不同：
如果稿子偏案例 / 质性 / 政策观察，《中国农村观察》可能更契合（具体以两刊当年定位为准，建议到官网"投稿须知"核对）。
cre-literature-review 直接进识别——审稿人首先看三农领域的理论定位与文献对话cre-tables-figures 在识别策略未立住时就美化表格cre-rebuttal 在你修订正文之前生成回复信| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 22,122 | 20,619 | -7% | 1 | 1 | 0% | 2,617 | 4,007 | +53% | 0 | 0 | — |
case-02 | fail→pass | 26,409 | 26,098 | -1% | 1 | 1 | 0% | 3,085 | 4,365 | +41% | 0 | 0 | — |
case-03 | fail→pass | 28,639 | 20,040 | -30% | 1 | 1 | 0% | 3,343 | 4,715 | +41% | 0 | 0 | — |
case-04 | fail→pass | 25,766 | 19,981 | -22% | 1 | 1 | 0% | 2,980 | 4,003 | +34% | 0 | 0 | — |
case-05 | fail→pass | 28,261 | 8,226 | -71% | 1 | 1 | 0% | 2,939 | 2,726 | -7% | 0 | 0 | — |
case-06 | fail→pass | 23,269 | 15,213 | -35% | 1 | 1 | 0% | 2,605 | 3,062 | +18% | 0 | 0 | — |
case-07 | fail→pass | 18,365 | 14,549 | -21% | 1 | 1 | 0% | 2,787 | 2,952 | +6% | 0 | 0 | — |
case-08 | fail→pass | 24,808 | 12,524 | -50% | 1 | 1 | 0% | 3,045 | 2,651 | -13% | 0 | 0 | — |
case-09 | fail→pass | 23,644 | 8,057 | -66% | 1 | 1 | 0% | 2,502 | 2,771 | +11% | 0 | 0 | — |
case-10 | fail→pass | 20,454 | 12,034 | -41% | 1 | 1 | 0% | 2,267 | 2,546 | +12% | 0 | 0 | — |
case-11 | fail→pass | 21,664 | 12,324 | -43% | 1 | 1 | 0% | 2,501 | 2,547 | +2% | 0 | 0 | — |
case-12 | pass→pass | 20,430 | 7,509 | -63% | 1 | 1 | 0% | 2,024 | 2,519 | +24% | 0 | 0 | — |
case-13 | fail→pass | 31,686 | 10,135 | -68% | 1 | 1 | 0% | 2,339 | 2,294 | -2% | 0 | 0 | — |
case-14 | pass→pass | 23,451 | 21,881 | -7% | 1 | 1 | 0% | 2,421 | 3,705 | +53% | 0 | 0 | — |
case-15 | pass→pass | 16,327 | 20,241 | +24% | 1 | 1 | 0% | 2,324 | 3,657 | +57% | 0 | 0 | — |
case-16 | pass→pass | 25,870 | 19,985 | -23% | 1 | 1 | 0% | 2,565 | 3,718 | +45% | 0 | 0 | — |
case-17 | pass→pass | 26,350 | 15,822 | -40% | 1 | 1 | 0% | 2,636 | 3,977 | +51% | 0 | 0 | — |
case-18 | fail→pass | 28,462 | 19,840 | -30% | 1 | 1 | 0% | 3,001 | 3,784 | +26% | 0 | 0 | — |
case-19 | fail→pass | 24,855 | 22,007 | -11% | 1 | 1 | 0% | 3,233 | 4,067 | +26% | 0 | 0 | — |
case-20 | pass→pass | 28,849 | 23,803 | -17% | 1 | 1 | 0% | 3,802 | 4,963 | +31% | 0 | 0 | — |
case-21 | fail→pass | 19,930 | 20,782 | +4% | 1 | 1 | 0% | 2,899 | 4,284 | +48% | 0 | 0 | — |
case-22 | fail→pass | 19,418 | 16,205 | -17% | 1 | 1 | 0% | 2,895 | 3,846 | +33% | 0 | 0 | — |
case-23 | pass→pass | 22,615 | 13,789 | -39% | 1 | 1 | 0% | 2,987 | 3,313 | +11% | 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 +70 percentage points is the difference between those two pass rates over the 23 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.