Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when drafting or strengthening the policy-implications section of a 《中国农村经济》 manuscript. Frames implications at the significance / institutional level but anchored in concrete 三农 scenarios — not the empty "加强完善推进" formula.
.claude/skills/brycewang-stanford-cre-policy-implication/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 59% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 69% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 48% | 0% |
《中国农村经济》的政策含义偏"意义层"——是研究结论对乡村振兴 / 农村政策的制度含义，而不是给某部委的执行备忘录。但"意义层"不等于空泛：必须落到具体的三农场景与对象（哪类农户、哪个环节、哪项制度），与本文实证证据严格挂钩。
| 维度 | 合格（本刊偏好） | 不合格 | |------|--------------------------------|----------------------------| | 层面 | 制度 / 原理层，但点明三农对象 | 笼统"政府应重视" | | 落点 | 哪类农户 / 哪个环节 / 哪项政策 | 不指对象的"加强完善推进" | | 依据 | 与本文机制 / 异质性结论呼应 | 与本文证据无关的泛泛建议 | | 收尾 | 与理论贡献 + 乡村振兴议题呼应 | 复述政府工作报告 |
本文的研究具有以下政策含义。
第一，[原理层 + 三农落点]——本文揭示了[机制]，提示在[土地 / 补贴 / 金融 / 组织等]制度设计中应重视……。
结合本文异质性发现，对[某类农户 / 某类地区]而言尤其如此。
第二，[宏观 / 战略层]——本文发现……，这意味着乡村振兴 / 粮食安全 / 共同富裕在……方面仍有改进空间。
第三，[微观主体层]（如适用）——对农户 / 合作社 / 家庭农场而言，本文提示……。本文为理解中国农村[领域]的[现象 / 制度]提供了新的微观证据。
本文发现……，这一结果意味着[制度意义]，对[乡村振兴 / 农业政策 / 相关理论]具有……启示。【段数】X
【层级覆盖】原理 / 宏观战略 / 微观主体
【三农落点】明确（哪类农户 / 环节 / 制度）/ 笼统
【与本文证据挂钩度】高 / 中 / 低
【四件套套话命中】X 处
【下一步】cre-abstract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 | 29,062 | 40,497 | +39% | 1 | 1 | 0% | 3,643 | 4,395 | +21% | 0 | 0 | — |
case-02 | fail→fail | 25,351 | 20,307 | -20% | 1 | 1 | 0% | 2,964 | 3,351 | +13% | 0 | 0 | — |
case-03 | fail→pass | 19,813 | 21,435 | +8% | 1 | 1 | 0% | 2,289 | 3,630 | +59% | 0 | 0 | — |
case-04 | pass→pass | 26,375 | 34,215 | +30% | 1 | 1 | 0% | 3,466 | 4,843 | +40% | 0 | 0 | — |
case-05 | pass→pass | 26,781 | 29,257 | +9% | 1 | 1 | 0% | 3,943 | 4,770 | +21% | 0 | 0 | — |
case-06 | pass→fail | 21,766 | 25,491 | +17% | 1 | 1 | 0% | 2,417 | 4,903 | +103% | 0 | 0 | — |
case-07 | pass→pass | 16,401 | 24,179 | +47% | 1 | 1 | 0% | 2,192 | 3,883 | +77% | 0 | 0 | — |
case-08 | fail→pass | 19,493 | 29,775 | +53% | 1 | 1 | 0% | 2,919 | 4,172 | +43% | 0 | 0 | — |
case-09 | fail→fail | 19,818 | 25,723 | +30% | 1 | 1 | 0% | 2,932 | 5,000 | +71% | 0 | 0 | — |
case-10 | pass→pass | 21,317 | 28,321 | +33% | 1 | 1 | 0% | 3,516 | 4,386 | +25% | 0 | 0 | — |
case-11 | fail→fail | 27,639 | 29,542 | +7% | 1 | 1 | 0% | 3,174 | 4,453 | +40% | 0 | 0 | — |
case-12 | fail→pass | 23,231 | 22,973 | -1% | 1 | 1 | 0% | 2,562 | 4,336 | +69% | 0 | 0 | — |
case-13 | pass→pass | 22,011 | 28,937 | +31% | 1 | 1 | 0% | 3,208 | 5,519 | +72% | 0 | 0 | — |
case-14 | pass→pass | 26,752 | 21,482 | -20% | 1 | 1 | 0% | 2,836 | 4,191 | +48% | 0 | 0 | — |
case-15 | pass→pass | 20,684 | 21,325 | +3% | 1 | 1 | 0% | 2,502 | 3,914 | +56% | 0 | 0 | — |
case-16 | fail→pass | 23,705 | 28,147 | +19% | 1 | 1 | 0% | 3,179 | 4,706 | +48% | 0 | 0 | — |
case-17 | fail→pass | 20,012 | 28,998 | +45% | 1 | 1 | 0% | 2,614 | 4,636 | +77% | 0 | 0 | — |
case-18 | fail→pass | 21,530 | 26,603 | +24% | 1 | 1 | 0% | 2,927 | 4,334 | +48% | 0 | 0 | — |
case-19 | fail→pass | 11,790 | 25,191 | +114% | 1 | 1 | 0% | 1,750 | 4,153 | +137% | 0 | 0 | — |
case-20 | pass→pass | 18,947 | 20,695 | +9% | 1 | 1 | 0% | 2,701 | 4,115 | +52% | 0 | 0 | — |
case-21 | fail→pass | 13,980 | 15,265 | +9% | 1 | 1 | 0% | 1,740 | 3,364 | +93% | 0 | 0 | — |
case-22 | pass→pass | 17,570 | 17,330 | -1% | 1 | 1 | 0% | 2,560 | 3,351 | +31% | 0 | 0 | — |
case-23 | fail→pass | 26,258 | 19,021 | -28% | 1 | 1 | 0% | 2,281 | 3,567 | +56% | 0 | 0 | — |
case-24 | pass→pass | 25,410 | 25,082 | -1% | 1 | 1 | 0% | 3,320 | 4,332 | +30% | 0 | 0 | — |
case-25 | fail→pass | 12,984 | 27,217 | +110% | 1 | 1 | 0% | 1,946 | 4,514 | +132% | 0 | 0 | — |
case-26 | fail→pass | 12,457 | 25,241 | +103% | 1 | 1 | 0% | 1,786 | 4,040 | +126% | 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. 26 cases were attempted. The headline lift of +42 percentage points is the difference between those two pass rates over the 26 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.