Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when user asks for a sequenced learning path within a Buddhist tradition — 学修次第, 先学什么, 从哪入门, 下一步读什么, curriculum, 学习计划, 路径推荐. Differs from /compare-masters (parallel opinion) and /master-debate (adversarial dialectic) by being 纵向 / 时序: stage-by-stage plan keyed on tradition × level (L0-L3) → foundation → intermediate → advanced + blind spots. Trigger is planning intent — "禅宗对比" goes to /compare-masters; "禅宗从哪开始学" goes here.
.claude/skills/xr843-master-curriculum/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -5% | 0% |
> 本路径依据历史佛教文献生成，仅供学习参考。如需正式修行指导，请亲近善知识。
| 用户说 | 加载 reference | |--------|---------------| | 禅宗 / 禅 / 见性 | references/chan.md | | 净土 / 念佛 / 弥陀 | references/jingtu.md | | 天台 / 止观 | references/tiantai.md | | 华严 / 一真法界 | references/huayan.md | | 唯识 / 法相 / 瑜伽行 | references/weishi.md | | 三论 / 中观 / 般若 | references/sanlun-zhongguan.md | | 格鲁 / 应成 / 道次第 | references/gelug-madhyamaka.md | | 上座部 / 内观 / vipassana | references/theravada-vipassana.md |
从用户输入抽取关键词，匹配每份 reference 顶部的 ## 触发关键词 列表，取最高分。若用户说"什么传统都行 / 综合理论"，按 keyword density 给一份默认推荐而非平均加载。
若全部 reference 关键词分数均为 0（典型例：噶举 / 米拉日巴 / 真言宗 / 黄檗 / 等暂未提供路径的传统），不要用任何 reference。明确告诉用户本传统尚无 curriculum 路径，并建议改用对应单 master skill（如 /master-milarepa）或先 /compare-masters 横向了解后再选定方向。禁止套用错误传统的路径。
markdown> 本路径依据历史佛教文献生成，仅供学习参考。如需正式修行指导，请亲近善知识。 ## 你的学修路径：<传统> · 从 L<n> 开始 ### 一、根基（入门，建议 N 周） - **主用经/论**：《<经名>》【<cbeta_id 或 sc_uid>】 - **推荐 master**：`/master-<slug>` — <此阶段教什么> - **目标**：能用自己的话讲清 <3 个核心概念> ### 二、深入（进阶，M 周） - 主用经/论 + 配合 master + 关键议题 ### 三、精研（专修，长期） - 主用经/论 + 配合 master + 验收标准 ### 四、可能的盲点 （本传统初学者最易踩 2-3 个陷阱 + 各祖师对此的提醒） ### 延伸 - 交叉对比 → `/compare-masters` - 了解争议 → `/master-debate`
meta.json.sources。CI 通过 scripts/validate-curriculum-sources.py 强制。/master-<slug> 必须指向已存在的 prebuilt/master-<slug>/。/compare-masters 和 /master-debate 的边界compare-masters = 横向并列（多家看一题，单轮）master-debate = 多轮交锋（看分歧）master-curriculum = 纵向时序（按月按季规划学什么）次第 / 先学 / 路径 / 计划 / 入门 → curriculum；不与 compare/debate 重叠。| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 32,340 | 23,970 | -26% | 1 | 1 | 0% | 4,817 | 4,706 | -2% | 0 | 0 | — |
case-02 | fail→pass | 38,134 | 28,497 | -25% | 1 | 1 | 0% | 6,244 | 5,840 | -6% | 0 | 0 | — |
case-03 | fail→pass | 26,076 | 19,241 | -26% | 1 | 1 | 0% | 4,182 | 4,187 | +0% | 0 | 0 | — |
case-04 | fail→fail | 20,684 | 31,451 | +52% | 1 | 1 | 0% | 3,247 | 5,754 | +77% | 0 | 0 | — |
case-05 | pass→pass | 26,723 | 27,413 | +3% | 1 | 1 | 0% | 3,772 | 5,234 | +39% | 0 | 0 | — |
case-06 | fail→pass | 22,048 | 9,500 | -57% | 1 | 1 | 0% | 3,473 | 2,749 | -21% | 0 | 0 | — |
case-07 | fail→fail | 27,578 | 18,689 | -32% | 1 | 1 | 0% | 4,737 | 4,169 | -12% | 0 | 0 | — |
case-08 | fail→pass | 21,694 | 14,618 | -33% | 1 | 1 | 0% | 3,426 | 3,244 | -5% | 0 | 0 | — |
case-09 | fail→fail | 22,110 | 21,798 | -1% | 1 | 1 | 0% | 3,244 | 4,230 | +30% | 0 | 0 | — |
case-10 | fail→pass | 19,385 | 19,140 | -1% | 1 | 1 | 0% | 3,276 | 4,137 | +26% | 0 | 0 | — |
case-11 | fail→pass | 23,645 | 17,130 | -28% | 1 | 1 | 0% | 3,855 | 3,940 | +2% | 0 | 0 | — |
case-12 | fail→pass | 19,294 | 20,358 | +6% | 1 | 1 | 0% | 3,116 | 4,224 | +36% | 0 | 0 | — |
case-13 | fail→pass | 20,874 | 21,944 | +5% | 1 | 1 | 0% | 3,334 | 4,488 | +35% | 0 | 0 | — |
case-14 | fail→pass | 30,085 | 22,264 | -26% | 1 | 1 | 0% | 4,510 | 4,392 | -3% | 0 | 0 | — |
case-15 | pass→pass | 17,753 | 19,090 | +8% | 1 | 1 | 0% | 2,447 | 3,995 | +63% | 0 | 0 | — |
case-16 | fail→pass | 20,762 | 9,073 | -56% | 1 | 1 | 0% | 3,562 | 2,498 | -30% | 0 | 0 | — |
case-17 | fail→pass | 21,645 | 23,470 | +8% | 1 | 1 | 0% | 3,348 | 4,501 | +34% | 0 | 0 | — |
case-18 | fail→pass | 23,520 | 19,305 | -18% | 1 | 1 | 0% | 3,652 | 3,849 | +5% | 0 | 0 | — |
case-19 | fail→pass | 29,988 | 23,900 | -20% | 1 | 1 | 0% | 4,406 | 4,584 | +4% | 0 | 0 | — |
case-20 | fail→fail | 19,108 | 19,992 | +5% | 1 | 1 | 0% | 2,948 | 4,304 | +46% | 0 | 0 | — |
case-21 | pass→pass | 26,807 | 26,334 | -2% | 1 | 1 | 0% | 4,786 | 5,065 | +6% | 0 | 0 | — |
case-22 | fail→pass | 22,323 | 19,938 | -11% | 1 | 1 | 0% | 3,122 | 4,116 | +32% | 0 | 0 | — |
case-23 | pass→pass | 15,806 | 14,952 | -5% | 1 | 1 | 0% | 2,447 | 3,647 | +49% | 0 | 0 | — |
case-24 | fail→pass | 18,798 | 18,292 | -3% | 1 | 1 | 0% | 2,872 | 3,868 | +35% | 0 | 0 | — |
case-25 | fail→pass | 21,023 | 21,828 | +4% | 1 | 1 | 0% | 3,000 | 4,406 | +47% | 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. 25 cases were attempted. The headline lift of +68 percentage points is the difference between those two pass rates over the 25 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.