Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →对中文毕业论文进行编译、warning 分级、模板模式检查、数据与引用复查，并把问题回写成可继续迭代的 review checklist。 **Trigger**: 毕业论文编译检查, thesis compile review, warning 分级, 终稿复查, main.pdf 检查. **Use when**: 论文已经回写到 TeX 交付层，需要确认是否真正达到“可提交”的质量，而不是只做到能编译。 **Skip if**: 还处于中间层重构阶段，`chapters/*.tex` 尚未形成稳定交付稿。 **Network**: none. **Guardrail**: 不在这里重构章节主线；如果发现结构问题，明确回退到上游修复。
.claude/skills/willoscar-thesis-compile-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | -46% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -69% | 0% |
main.texreferences/*.bibclaude_md/review_checklist.mdoutput/THESIS_BUILD_REPORT.mdclaude_md/review_checklist.mdAlways read:
references/overview.mdreferences/warning_triage.mdMachine-readable contract:
assets/build_review_contract.jsonmain.tex“能编译”只是最低标准。 真正的目标是形成：
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | fail→pass | 22,002 | 1,872 | -91% | 1 | 1 | 0% | 949 | 510 | -46% | 0 | 0 | — |
case-15 | fail→pass | 7,779 | 3,148 | -60% | 1 | 1 | 0% | 1,046 | 729 | -30% | 0 | 0 | — |
case-20 | pass→pass | 15,767 | 15,181 | -4% | 1 | 1 | 0% | 2,528 | 2,769 | +10% | 0 | 0 | — |
case-01 | fail→fail | 1,884 | 4,977 | +164% | 1 | 1 | 0% | 234 | 471 | +101% | 0 | 0 | — |
case-02 | fail→fail | 5,178 | 5,450 | +5% | 1 | 1 | 0% | 189 | 500 | +165% | 0 | 0 | — |
case-03 | fail→fail | 18,657 | 5,094 | -73% | 1 | 1 | 0% | 2,864 | 495 | -83% | 0 | 0 | — |
case-04 | fail→pass | 9,595 | 1,650 | -83% | 1 | 1 | 0% | 1,325 | 441 | -67% | 0 | 0 | — |
case-05 | fail→pass | 7,567 | 2,829 | -63% | 1 | 1 | 0% | 1,085 | 652 | -40% | 0 | 0 | — |
case-07 | pass→pass | 7,629 | 6,478 | -15% | 1 | 1 | 0% | 1,213 | 1,259 | +4% | 0 | 0 | — |
case-08 | fail→pass | 9,994 | 1,680 | -83% | 1 | 1 | 0% | 1,493 | 470 | -69% | 0 | 0 | — |
case-09 | fail→pass | 19,999 | 1,876 | -91% | 1 | 1 | 0% | 997 | 473 | -53% | 0 | 0 | — |
case-10 | pass→pass | 12,267 | 2,997 | -76% | 1 | 1 | 0% | 1,812 | 705 | -61% | 0 | 0 | — |
case-11 | fail→pass | 14,080 | 10,409 | -26% | 1 | 1 | 0% | 2,040 | 1,797 | -12% | 0 | 0 | — |
case-12 | fail→pass | 14,014 | 4,531 | -68% | 1 | 1 | 0% | 1,944 | 977 | -50% | 0 | 0 | — |
case-13 | pass→pass | 11,473 | 6,542 | -43% | 1 | 1 | 0% | 1,599 | 1,113 | -30% | 0 | 0 | — |
case-14 | pass→pass | 12,451 | 3,543 | -72% | 1 | 1 | 0% | 1,885 | 752 | -60% | 0 | 0 | — |
case-16 | fail→pass | 15,659 | 8,779 | -44% | 1 | 1 | 0% | 2,486 | 1,573 | -37% | 0 | 0 | — |
case-17 | fail→fail | 7,441 | 1,550 | -79% | 1 | 1 | 0% | 1,020 | 421 | -59% | 0 | 0 | — |
case-18 | pass→pass | 10,449 | 4,729 | -55% | 1 | 1 | 0% | 1,488 | 940 | -37% | 0 | 0 | — |
case-19 | fail→pass | 14,981 | 4,393 | -71% | 1 | 1 | 0% | 1,979 | 874 | -56% | 0 | 0 | — |
case-21 | pass→pass | 8,897 | 7,562 | -15% | 1 | 1 | 0% | 1,617 | 1,451 | -10% | 0 | 0 | — |
case-22 | pass→pass | 13,342 | 13,085 | -2% | 1 | 1 | 0% | 2,370 | 2,566 | +8% | 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, and 17 counted toward the lift figure. The other 5 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +45 percentage points is the difference between those two pass rates over the 17 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.