Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when finalizing regression tables and figures for a 《中国农村经济》 manuscript. Enforces three-line table style, footnote conventions, column-count discipline, and figure aesthetics. 本技能服务于《中国农村经济》(China Rural Economy, CRE)。
.claude/skills/brycewang-stanford-cre-tables-figures/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 42% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 72% | 0% |
表 3 土地确权对农户农业投资的影响
═══════════════════════════════════════════
(1) (2) (3)
变量 农业投资 农业投资 农业投资
───────────────────────────────────────────
确权×确权后 0.062*** 0.051*** 0.048***
(0.018) (0.017) (0.019)
......
───────────────────────────────────────────
控制变量 否 是 是
农户固定效应 是 是 是
年份固定效应 是 是 是
观测值 12,486 12,486 12,486
R² 0.184 0.226 0.231
═══════════════════════════════════════════
注：括号内为聚类于村级的稳健标准误；***、**、* 分别表示 1%、5%、10% 显著水平。*** ** * 统一用三档主回归表之前必须有变量定义表，三农研究的格式约定：
| 变量类型 | 变量名称 | 变量定义 | 数据来源 | |----------|----------|----------|----------| | 被解释变量 | farm_inv | 农户农业投资（取对数，元/年） | CFPS | | 核心解释变量 | titled | 土地确权完成虚拟变量 | 农村固定观察点 | | 控制变量 | edu | 户主受教育年限 | CFPS | | 控制变量 | landsize | 家庭经营耕地面积（亩） | CFPS | | 工具变量 | iv_pilot | 所在县纳入确权试点的时间 | 政策文件整理 |
约束：
变量定义表前必须有一段数据说明（≈ 200 字），结构固定：
> 本文使用 时间跨度] 的 数据库 A，如 CFPS] 和 数据库 B]，研究样本包含 地区 / 群体] 的 N] 个农户 / 村庄观测。 > 样本筛选标准：(1) 保留有农业经营的农村家庭；(2) 剔除关键变量缺失样本；(3) 剔除明显异常值。 > 为缓解极端值影响，对所有连续变量进行了 1% 和 99% 分位的 winsorize 处理。 > 数据合并方式：以 家庭 ID / 村码 / 年份] 为匹配键，跨轮次追踪。
禁忌：不要写"数据来源于公开渠道"或"调研所得"而不点名——必须点名调查 / 数据库，自行调研的需说明抽样框、年份、样本量。
表格图形从拟合结果生成，不要手抄数字。完整映射见 execution-with-mcp。《中国农村经济》是三农实证刊，政策评估与微观面板为主；突出识别与选择性偏误处理。
etable（多列）或 did_summary_to_latex 直接从 result_id 生成。plot_from_result / enhanced_event_study_plot / event_study_table，坐标单位与标准误/聚类注记自带。
完整“拟合结果 → 图表”链见 JF 执行 walkthrough。
【主表列数】X 列
【三线表合规】是 / 否
【标准误格式】统一（聚类层次：...）/ 不统一
【数据来源点名】是 / 否
【图形 DPI】X
【缺漏注释】[...]
【下一步】cre-policy-implication| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 40,466 | 31,846 | -21% | 1 | 1 | 0% | 4,566 | 6,006 | +32% | 0 | 0 | — |
case-02 | fail→pass | 42,045 | 37,668 | -10% | 1 | 1 | 0% | 4,617 | 6,215 | +35% | 0 | 0 | — |
case-03 | fail→pass | 32,289 | 25,269 | -22% | 1 | 1 | 0% | 3,713 | 5,046 | +36% | 0 | 0 | — |
case-04 | pass→pass | 25,192 | 22,238 | -12% | 1 | 1 | 0% | 3,143 | 4,170 | +33% | 0 | 0 | — |
case-05 | pass→pass | 20,025 | 23,410 | +17% | 1 | 1 | 0% | 2,424 | 4,514 | +86% | 0 | 0 | — |
case-06 | pass→pass | 24,451 | 21,873 | -11% | 1 | 1 | 0% | 2,506 | 4,105 | +64% | 0 | 0 | — |
case-07 | pass→pass | 21,926 | 22,880 | +4% | 1 | 1 | 0% | 2,539 | 4,155 | +64% | 0 | 0 | — |
case-08 | pass→pass | 22,106 | 19,570 | -11% | 1 | 1 | 0% | 2,132 | 3,643 | +71% | 0 | 0 | — |
case-09 | pass→pass | 27,971 | 25,265 | -10% | 1 | 1 | 0% | 2,780 | 4,455 | +60% | 0 | 0 | — |
case-10 | pass→pass | 20,074 | 17,126 | -15% | 1 | 1 | 0% | 2,426 | 3,903 | +61% | 0 | 0 | — |
case-11 | pass→pass | 20,604 | 15,634 | -24% | 1 | 1 | 0% | 1,951 | 3,194 | +64% | 0 | 0 | — |
case-12 | fail→pass | 22,422 | 18,210 | -19% | 1 | 1 | 0% | 2,708 | 3,834 | +42% | 0 | 0 | — |
case-13 | pass→pass | 15,909 | 17,596 | +11% | 1 | 1 | 0% | 2,209 | 3,428 | +55% | 0 | 0 | — |
case-14 | pass→pass | 25,741 | 19,990 | -22% | 1 | 1 | 0% | 2,653 | 3,860 | +45% | 0 | 0 | — |
case-15 | fail→pass | 16,379 | 21,248 | +30% | 1 | 1 | 0% | 2,165 | 3,731 | +72% | 0 | 0 | — |
case-16 | fail→pass | 26,695 | 29,097 | +9% | 1 | 1 | 0% | 3,301 | 4,953 | +50% | 0 | 0 | — |
case-17 | pass→pass | 25,051 | 18,753 | -25% | 1 | 1 | 0% | 2,238 | 3,314 | +48% | 0 | 0 | — |
case-18 | fail→pass | 42,722 | 22,107 | -48% | 1 | 1 | 0% | 3,025 | 4,183 | +38% | 0 | 0 | — |
case-19 | pass→pass | 28,067 | 25,155 | -10% | 1 | 1 | 0% | 2,879 | 4,249 | +48% | 0 | 0 | — |
case-20 | pass→pass | 22,091 | 26,885 | +22% | 1 | 1 | 0% | 2,119 | 4,223 | +99% | 0 | 0 | — |
case-21 | pass→pass | 40,427 | 46,362 | +15% | 1 | 1 | 0% | 4,914 | 6,482 | +32% | 0 | 0 | — |
case-22 | pass→pass | 43,156 | 28,589 | -34% | 1 | 1 | 0% | 4,015 | 4,945 | +23% | 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 +32 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.