Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when CrossFrame Suite routes explicit Chinese teaching of CrossFrame concepts, misreading boundaries, plain-language examples, signals, or exercises.
.claude/skills/sickn33-crossframe-teach/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 40% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 113% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 112% | 0% |
crossframe-suite routes explicit CrossFrame work into concept teaching, misreading correction, plain-language examples, observable signals, or exercises.This AAS-ready copy preserves the original CrossFrame skill body below. Chinese remains the canonical semantic layer; English metadata is only for discovery, installation, and repository review.
crossframe-suite routing; do not apply it as a generic default reasoning layer.> 本 skill 不独立触发。 所有 CrossFrame 任务统一从 crossframe-suite 入口调度。用户无需直接调用本 skill；suite 根据路由规则在需要时自动加载。
如果概念教学要连接文章写作、案例沉淀、读书研究或输出评审，先读取 ../crossframe-suite/SKILL.md 做总调度；本 skill 只负责教学解释、误读边界和练习。
中文是权威语义。CrossFrame Teach 只是教学入口，不重写、不替代、不压缩 canonical CrossFrame。
每次触发后先读取相邻 canonical 材料：
../crossframe/SKILL.md../crossframe/references/read-routing-map.md../crossframe/references/continuity-bundles.md，并按需使用 ../crossframe/worksheets/source-continuity-check.md；未完成联读时只能降档。../crossframe/templates/read-state-capsule.md 规定的 v5-read-state-capsule，并在高责任、公共、AI/过程性产物、生命周期、无法退出主体或文章输出场景执行 ../crossframe/worksheets/source-anchor-integrity-check.md。如果胶囊缺失，回到 ../crossframe/SKILL.md 补齐；本 skill 不重新发明源路由。不要把 canonical 全文复制进回答。只按本次概念需要读取 canonical 的协议、术语保真材料、概念卡或模板；教学表达使用本 skill 的轻量协议和模板。
protocols/teach-protocol.md，确定本次是概念课、误读纠偏、现实信号训练，还是练习题生成。references/teaching-fidelity.md，防止术语堆砌、解释过短失真、道德化和漏练习。templates/concept-lesson.md；只生成练习时使用 templates/micro-exercises.md。examples/ 中对应概念；需要自测时读取 evals/smoke-tests.md。默认按这个顺序输出，不要把术语放在第一段当结论：
如果用户要求极简，也至少保留一个极短自测问题，除非用户明确说不要练习。
一次合格的教学回答必须能回答：
protocols/teach-protocol.md：教学解释流程。references/teaching-fidelity.md：教学保真与反误用规则。templates/concept-lesson.md：完整概念课模板。templates/micro-exercises.md：练习题模板。examples/chengjie-huiliu.md：承接/回流教学样例。examples/open-assertion.md：开放断言教学样例。examples/love-open-action.md：爱/开放行动教学样例。examples/failure-patterns.md：失败样例。evals/smoke-tests.md：smoke tests。| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | pass→fail | 17,444 | 16,174 | -7% | 1 | 1 | 0% | 2,701 | 4,028 | +49% | 0 | 0 | — |
case-01 | fail→pass | 20,014 | 18,311 | -9% | 1 | 1 | 0% | 3,282 | 4,367 | +33% | 0 | 0 | — |
case-02 | pass→pass | 14,690 | 17,260 | +17% | 1 | 1 | 0% | 2,787 | 4,073 | +46% | 0 | 0 | — |
case-04 | fail→pass | 15,092 | 15,468 | +2% | 1 | 1 | 0% | 2,538 | 3,544 | +40% | 0 | 0 | — |
case-05 | fail→fail | 12,603 | 15,511 | +23% | 1 | 1 | 0% | 2,083 | 3,797 | +82% | 0 | 0 | — |
case-06 | fail→pass | 14,163 | 14,875 | +5% | 1 | 1 | 0% | 2,515 | 3,832 | +52% | 0 | 0 | — |
case-07 | fail→pass | 7,284 | 8,529 | +17% | 1 | 1 | 0% | 1,286 | 2,740 | +113% | 0 | 0 | — |
case-13 | pass→pass | 13,311 | 13,186 | -1% | 1 | 1 | 0% | 2,151 | 3,404 | +58% | 0 | 0 | — |
case-08 | pass→pass | 14,194 | 21,421 | +51% | 1 | 1 | 0% | 2,390 | 4,412 | +85% | 0 | 0 | — |
case-09 | pass→pass | 14,481 | 16,220 | +12% | 1 | 1 | 0% | 2,326 | 4,150 | +78% | 0 | 0 | — |
case-10 | pass→pass | 19,352 | 17,203 | -11% | 1 | 1 | 0% | 3,070 | 4,223 | +38% | 0 | 0 | — |
case-11 | pass→pass | 10,570 | 15,938 | +51% | 1 | 1 | 0% | 1,789 | 3,923 | +119% | 0 | 0 | — |
case-12 | fail→pass | 12,698 | 19,922 | +57% | 1 | 1 | 0% | 1,986 | 4,204 | +112% | 0 | 0 | — |
case-14 | fail→fail | 10,765 | 10,914 | +1% | 1 | 1 | 0% | 1,767 | 3,054 | +73% | 0 | 0 | — |
case-15 | fail→pass | 22,704 | 24,981 | +10% | 1 | 1 | 0% | 3,789 | 5,744 | +52% | 0 | 0 | — |
case-16 | pass→pass | 16,683 | 17,281 | +4% | 1 | 1 | 0% | 2,830 | 3,895 | +38% | 0 | 0 | — |
case-17 | pass→pass | 12,085 | 13,379 | +11% | 1 | 1 | 0% | 2,024 | 3,636 | +80% | 0 | 0 | — |
case-18 | pass→pass | 17,201 | 18,362 | +7% | 1 | 1 | 0% | 2,712 | 4,233 | +56% | 0 | 0 | — |
case-19 | pass→pass | 15,037 | 19,181 | +28% | 1 | 1 | 0% | 2,315 | 4,009 | +73% | 0 | 0 | — |
case-20 | fail→pass | 13,827 | 15,270 | +10% | 1 | 1 | 0% | 2,175 | 3,593 | +65% | 0 | 0 | — |
case-21 | pass→fail | 15,532 | 22,705 | +46% | 1 | 1 | 0% | 2,566 | 4,853 | +89% | 0 | 0 | — |
case-22 | pass→fail | 27,209 | 27,293 | +0% | 1 | 1 | 0% | 3,690 | 5,177 | +40% | 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 +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 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.