Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when polishing the language and rhetoric of a 《中国农村经济》 manuscript — eliminating empty-significance phrases, fixing the "加强完善推进" policy formula, replacing method-flexing with household mechanism, and aligning tense / person / hedging with CSSCI house style. 本技能服务于《中国农村经济》(China Rural Economy, CRE)。
.claude/skills/brycewang-stanford-cre-style/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 46% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 54% | 0% |
| 问题类型 | 黑名单（删/改） | 白名单（替换为） | |----------|-----------------------------------|---------------------------------------------------| | 空洞价值 | "具有重要的理论价值" | "为理解 X 提供了 Y 视角下的农户行为框架" | | 空洞意义 | "对农村经济发展具有重要参考价值" | "本文结果意味着 X 政策应将 Y 类农户纳入瞄准范围" | | 四件套建议 | "应加强 / 完善 / 推进 / 深化……" | "X 部门应在 Y 环节针对 Z 类农户建立……机制" | | 方法炫耀 | "本文使用复杂的因果识别方法" | "本文利用确权试点的分批推行构造交叠 DID，识别 X" | | 贡献模糊 | "丰富了三农研究" | "首次利用农户面板识别了 X 对 Y 的因果效应" | | 文献堆砌 | "（张三，2020；李四，2021；……）" | 按贡献分述：理论文献一段、三农实证一段、本文位置一段 | | 表格无解读 | "表 3 报告了基准回归结果。" | "表 3 显示，确权使农户农业投资提高约 X%，约相当于样本均值的 Y%" | | 机制空泛 | "促进了城乡融合" | "通过缓解农户信贷约束、提高土地经营规模实现" |
每个段落第一句必须是概括句，后续句子是支撑。审稿人快速读完每段首句应能复原文章逻辑。
错误： > 张三（2020）研究发现土地流转……。李四（2021）发现合作社……。王五（2022）发现外出务工……。
正确： > 既有研究就要素流动对农户收入的影响存在分歧（张三，2020；李四，2021）。一类研究强调……，另一类研究强调……。两者的分歧在于对农户自选择的不同处理。
【黑名单命中】X 处，分别在：[...]
【四件套套话命中】X 处
【段首概括率】X / 总段数
【政策建议落点】明确到农户 / 笼统
【经济含义阐述】到位 / 缺失 [...]
【时态/人称一致性】一致 / 待统一
【下一步】cre-submission先锁定农村问题、政策/制度场景、识别链条、机制证据和可执行含义，再判断稿件是否回应农村经济审稿人通常同时追问“三农”问题意识、政策场景、识别可信度和农村制度机制。
resources/official-source-map.md，列出仍可能改变建议的一个未核实事实。| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 27,351 | 25,455 | -7% | 1 | 1 | 0% | 3,445 | 4,555 | +32% | 0 | 0 | — |
case-02 | fail→fail | 22,035 | 18,290 | -17% | 1 | 1 | 0% | 2,768 | 3,767 | +36% | 0 | 0 | — |
case-03 | fail→pass | 26,293 | 13,512 | -49% | 1 | 1 | 0% | 2,864 | 2,939 | +3% | 0 | 0 | — |
case-04 | pass→fail | 12,596 | 16,524 | +31% | 1 | 1 | 0% | 2,475 | 4,145 | +67% | 0 | 0 | — |
case-05 | pass→fail | 16,598 | 19,400 | +17% | 1 | 1 | 0% | 1,933 | 4,540 | +135% | 0 | 0 | — |
case-06 | pass→pass | 22,555 | 22,303 | -1% | 1 | 1 | 0% | 2,517 | 4,027 | +60% | 0 | 0 | — |
case-20 | fail→pass | 15,850 | 2,629 | -83% | 1 | 1 | 0% | 2,323 | 1,776 | -24% | 0 | 0 | — |
case-07 | pass→pass | 25,150 | 30,165 | +20% | 1 | 1 | 0% | 3,181 | 5,310 | +67% | 0 | 0 | — |
case-08 | pass→pass | 19,212 | 25,650 | +34% | 1 | 1 | 0% | 3,057 | 4,729 | +55% | 0 | 0 | — |
case-09 | pass→pass | 21,623 | 26,475 | +22% | 1 | 1 | 0% | 2,639 | 4,080 | +55% | 0 | 0 | — |
case-10 | pass→pass | 20,970 | 15,960 | -24% | 1 | 1 | 0% | 2,204 | 3,141 | +43% | 0 | 0 | — |
case-11 | fail→pass | 26,070 | 26,930 | +3% | 1 | 1 | 0% | 3,031 | 4,439 | +46% | 0 | 0 | — |
case-12 | pass→pass | 24,819 | 27,715 | +12% | 1 | 1 | 0% | 2,624 | 4,006 | +53% | 0 | 0 | — |
case-13 | pass→pass | 25,356 | 23,015 | -9% | 1 | 1 | 0% | 2,830 | 4,207 | +49% | 0 | 0 | — |
case-14 | pass→pass | 24,068 | 21,118 | -12% | 1 | 1 | 0% | 3,464 | 4,284 | +24% | 0 | 0 | — |
case-15 | fail→pass | 20,998 | 16,688 | -21% | 1 | 1 | 0% | 2,507 | 3,858 | +54% | 0 | 0 | — |
case-16 | pass→pass | 18,327 | 20,926 | +14% | 1 | 1 | 0% | 2,548 | 3,607 | +42% | 0 | 0 | — |
case-17 | pass→pass | 23,771 | 20,866 | -12% | 1 | 1 | 0% | 2,909 | 3,900 | +34% | 0 | 0 | — |
case-18 | fail→pass | 12,677 | 19,350 | +53% | 1 | 1 | 0% | 2,038 | 3,256 | +60% | 0 | 0 | — |
case-19 | fail→pass | 20,243 | 20,388 | +1% | 1 | 1 | 0% | 2,728 | 3,616 | +33% | 0 | 0 | — |
case-21 | pass→pass | 21,108 | 18,606 | -12% | 1 | 1 | 0% | 2,302 | 3,962 | +72% | 0 | 0 | — |
case-22 | pass→pass | 18,841 | 12,675 | -33% | 1 | 1 | 0% | 2,523 | 3,013 | +19% | 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 +23 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are 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.