Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →围绕毕业论文主线重构章节：把原论文式叙事改成学位论文式叙事，重写章节目标、内容比重、前后承接与论证方式。 **Trigger**: 章节重构, chapter reconstructor, 毕业论文主线重构, paper 变 thesis, 章节改写. **Use when**: 你已经知道每份材料在毕业论文中的角色，但当前章节仍然像原论文翻译、拼接或卖点展示。 **Skip if**: 当前只是做局部句子润色，或章节角色尚未确定。 **Network**: none. **Guardrail**: 不是翻译论文；不是简单删减；必须围绕毕业论文主线重写章节目标与承接关系。
.claude/skills/willoscar-thesis-chapter-reconstructor/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 133% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -5% | 0% |
这是 graduate-paper pipeline 的核心重构 skill。
codex_md/chapter_role_map.mdcodex_md/question_list.mdcodex_md/chapter_rewrite_rules.mdAlways read:
references/overview.mdreferences/rewrite_rules.mdreferences/examples_good.mdreferences/examples_bad.mdMachine-readable contract:
assets/chapter_rewrite_contract.jsonchapter_rewrite_rules.md任何一章如果还像：
都说明这一步还没做完。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 15,115 | 32,926 | +118% | 1 | 1 | 0% | 2,230 | 5,192 | +133% | 0 | 0 | — |
case-02 | fail→pass | 25,891 | 26,138 | +1% | 1 | 1 | 0% | 3,583 | 4,051 | +13% | 0 | 0 | — |
case-03 | fail→pass | 24,746 | 34,177 | +38% | 1 | 1 | 0% | 3,753 | 5,453 | +45% | 0 | 0 | — |
case-04 | fail→fail | 2,885 | 6,903 | +139% | 1 | 1 | 0% | 413 | 1,433 | +247% | 0 | 0 | — |
case-05 | fail→fail | 2,856 | 3,916 | +37% | 1 | 1 | 0% | 454 | 807 | +78% | 0 | 0 | — |
case-06 | fail→fail | 2,189 | 4,729 | +116% | 1 | 1 | 0% | 287 | 851 | +197% | 0 | 0 | — |
case-11 | fail→fail | 9,368 | 2,413 | -74% | 1 | 1 | 0% | 1,489 | 622 | -58% | 0 | 0 | — |
case-07 | fail→pass | 9,884 | 11,633 | +18% | 1 | 1 | 0% | 1,446 | 1,957 | +35% | 0 | 0 | — |
case-08 | fail→pass | 12,373 | 8,919 | -28% | 1 | 1 | 0% | 1,615 | 1,528 | -5% | 0 | 0 | — |
case-09 | pass→pass | 15,308 | 16,095 | +5% | 1 | 1 | 0% | 2,186 | 2,735 | +25% | 0 | 0 | — |
case-10 | pass→pass | 15,858 | 20,585 | +30% | 1 | 1 | 0% | 2,279 | 3,162 | +39% | 0 | 0 | — |
case-12 | fail→pass | 7,644 | 1,857 | -76% | 1 | 1 | 0% | 1,100 | 546 | -50% | 0 | 0 | — |
case-13 | fail→pass | 17,776 | 1,775 | -90% | 1 | 1 | 0% | 1,046 | 518 | -50% | 0 | 0 | — |
case-14 | fail→pass | 11,141 | 2,006 | -82% | 1 | 1 | 0% | 1,572 | 516 | -67% | 0 | 0 | — |
case-15 | fail→pass | 6,941 | 2,180 | -69% | 1 | 1 | 0% | 961 | 554 | -42% | 0 | 0 | — |
case-16 | fail→pass | 11,364 | 1,685 | -85% | 1 | 1 | 0% | 1,873 | 493 | -74% | 0 | 0 | — |
case-17 | fail→pass | 9,031 | 1,835 | -80% | 1 | 1 | 0% | 1,490 | 558 | -63% | 0 | 0 | — |
case-18 | fail→pass | 7,411 | 2,018 | -73% | 1 | 1 | 0% | 1,173 | 573 | -51% | 0 | 0 | — |
case-19 | pass→pass | 12,281 | 11,519 | -6% | 1 | 1 | 0% | 1,674 | 1,795 | +7% | 0 | 0 | — |
case-20 | pass→pass | 13,355 | 6,920 | -48% | 1 | 1 | 0% | 1,808 | 1,287 | -29% | 0 | 0 | — |
case-21 | pass→pass | 12,825 | 12,049 | -6% | 1 | 1 | 0% | 1,891 | 2,062 | +9% | 0 | 0 | — |
case-22 | pass→pass | 11,635 | 8,679 | -25% | 1 | 1 | 0% | 1,635 | 1,533 | -6% | 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 +55 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.