Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when partitioning content between an Advanced Materials main text and its Supporting Information, ensuring the paper stands alone while extended data, full experimental detail, and secondary characterization live in the SI. Organizes the Supporting Information; does not write the main text.
.claude/skills/brycewang-stanford-advmat-supplementary/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 115% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 1% | 0% |
The main text must be fully understandable and convincing on its own. The Supporting Information supports the paper; it must never be load-bearing for the central advance. A reader who never opens the SI should still grasp the material, the decisive evidence that it is what you claim, the property, and the benchmarked function.
If removing the SI collapses the argument, the partition is wrong — pull the essential piece back into the main text (and trim elsewhere via advmat-length-management).
| Content | Home | |-----------------------------------------------------|------------| | Full synthesis recipes, optimization sweeps | SI | | Extended characterization (more samples, techniques) | SI | | Full device fabrication protocols, control devices | SI | | Complete benchmarking table with all references | SI | | Per-sample statistics, histograms, batch data | SI | | Additional figures, movies, large data tables | SI | | DFT/computational details, input files | SI | | The central advance and its decisive evidence | Main text | | The property mechanism linking structure to function | Main text | | The headline benchmarked metric with n and spread | Main text |
【Main text stands alone】yes / fix (pull back: ...)
【SI sections】S1: ... / S2: ... (cited from body?)
【Decisive evidence only in SI?】none / move back: ...
【SI figure/table labels】S-prefixed? yes / fix
【Reproducible recipe present】yes / fix
【Nomenclature consistent body↔SI】yes / fix
【Next】advmat-writing-style or advmat-length-management> SI scope, file formats, and data-deposition rules are set by Wiley and evolve — verify on the official Advanced Materials author page.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 14,967 | 10,046 | -33% | 1 | 1 | 0% | 2,598 | 2,523 | -3% | 0 | 0 | — |
case-02 | fail→pass | 19,320 | 9,691 | -50% | 1 | 1 | 0% | 3,073 | 2,510 | -18% | 0 | 0 | — |
case-03 | fail→pass | 14,618 | 8,893 | -39% | 1 | 1 | 0% | 2,445 | 2,467 | +1% | 0 | 0 | — |
case-04 | pass→pass | 6,621 | 6,271 | -5% | 1 | 1 | 0% | 1,171 | 1,986 | +70% | 0 | 0 | — |
case-05 | pass→fail | 14,680 | 11,135 | -24% | 1 | 1 | 0% | 2,297 | 2,662 | +16% | 0 | 0 | — |
case-06 | pass→pass | 5,027 | 4,248 | -15% | 1 | 1 | 0% | 921 | 1,651 | +79% | 0 | 0 | — |
case-07 | fail→pass | 23,786 | 5,763 | -76% | 1 | 1 | 0% | 923 | 1,980 | +115% | 0 | 0 | — |
case-08 | pass→pass | 10,762 | 5,412 | -50% | 1 | 1 | 0% | 1,607 | 1,816 | +13% | 0 | 0 | — |
case-09 | pass→pass | 12,258 | 6,384 | -48% | 1 | 1 | 0% | 2,053 | 2,231 | +9% | 0 | 0 | — |
case-10 | pass→pass | 7,304 | 4,884 | -33% | 1 | 1 | 0% | 1,220 | 1,821 | +49% | 0 | 0 | — |
case-11 | fail→pass | 9,946 | 5,281 | -47% | 1 | 1 | 0% | 1,879 | 1,894 | +1% | 0 | 0 | — |
case-12 | pass→pass | 11,268 | 6,397 | -43% | 1 | 1 | 0% | 1,750 | 2,027 | +16% | 0 | 0 | — |
case-13 | pass→pass | 12,386 | 7,204 | -42% | 1 | 1 | 0% | 2,007 | 2,181 | +9% | 0 | 0 | — |
case-14 | fail→pass | 7,036 | 5,111 | -27% | 1 | 1 | 0% | 1,303 | 1,929 | +48% | 0 | 0 | — |
case-15 | pass→pass | 10,256 | 7,238 | -29% | 1 | 1 | 0% | 1,835 | 2,159 | +18% | 0 | 0 | — |
case-16 | fail→pass | 12,664 | 4,452 | -65% | 1 | 1 | 0% | 1,990 | 1,796 | -10% | 0 | 0 | — |
case-17 | pass→pass | 11,381 | 7,523 | -34% | 1 | 1 | 0% | 1,826 | 2,214 | +21% | 0 | 0 | — |
case-18 | pass→pass | 12,822 | 6,698 | -48% | 1 | 1 | 0% | 2,054 | 2,160 | +5% | 0 | 0 | — |
case-19 | pass→pass | 11,911 | 7,083 | -41% | 1 | 1 | 0% | 1,903 | 2,123 | +12% | 0 | 0 | — |
case-20 | fail→pass | 12,254 | 7,287 | -41% | 1 | 1 | 0% | 1,752 | 2,138 | +22% | 0 | 0 | — |
case-21 | fail→pass | 4,791 | 3,061 | -36% | 1 | 1 | 0% | 769 | 1,559 | +103% | 0 | 0 | — |
case-22 | pass→pass | 13,075 | 7,575 | -42% | 1 | 1 | 0% | 2,277 | 2,297 | +1% | 0 | 0 | — |
case-23 | fail→pass | 9,531 | 5,036 | -47% | 1 | 1 | 0% | 1,779 | 1,885 | +6% | 0 | 0 | — |
case-24 | pass→pass | 10,637 | 5,253 | -51% | 1 | 1 | 0% | 1,688 | 1,807 | +7% | 0 | 0 | — |
case-25 | pass→pass | 10,208 | 5,653 | -45% | 1 | 1 | 0% | 1,571 | 1,921 | +22% | 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, and 24 counted toward the lift figure. The other 1 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 +36 percentage points is the difference between those two pass rates over the 24 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.