Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when formatting a paper, poster, or proposal for a named academic venue: apply that venue's exact layout, citation-marker, page-limit, and anonymization rules, which differ from generic defaults.
.claude/skills/academic-venue-formatting/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 89% | 10 |
| Model | Lift | Δ tokens | Δ turns | Cases | Verified |
|---|---|---|---|---|---|
| gemini-3.6-flashbest | +21% | +78% | 0% | 24 | 56d ago |
| gemini-3.5-flash | pending re-run | — | |||
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✗→✓ | ▲ Improved | — | — |
| case-05 | ✗→✓ | ▲ Improved | — | — |
| case-01 | ✗→✓ | ▲ Improved | — | — |
| case-13 | ✗→✓ | ▲ Improved | — | — |
| case-09 | ✗→✓ | ▲ Improved | — | — |
When a task names an academic venue (NeurIPS, Nature, CVPR, ACL, NSF, ...), emit that venue's EXACT spec for layout, body font, citation markers, anonymization, page limits, figures, and posters. Do not emit a generic article; the generic default is usually the wrong choice for at least one of these dimensions.
text^1. Not bracketed, not author-year.(1). Not superscript, not author-year.[1].(Smith, 2021). Not numbered.(Devlin et al., 2019). Not numbered.Remove author names and affiliations (use "Anonymous Author(s)" / blank author block) and anonymize self-citations so authorship is not revealed.
Author names and affiliations stay on the manuscript.
BEFORE: \documentclass[twocolumn]{article} at 10pt. AFTER: \documentclass{article} (SINGLE-column) at 10pt body, \usepackage{neurips_2024}.
BEFORE: two-column at 10pt. AFTER: two-column at 9pt body (\documentclass[sigconf]{acmart}).
BEFORE: "...as shown by prior work 12]." AFTER: "...as shown by prior work^12." (raised superscript numeral, no brackets).
BEFORE: "...prior work^12." (superscript). AFTER: "...prior work (12)." (parenthetical numeral).
BEFORE: "...prior work 12]." AFTER: "...prior work (Devlin et al., 2019)." (author-year inline).
BEFORE: \author{Jane Doe (MIT) \and John Smith (Stanford)}. AFTER: \author{Anonymous Author(s)} with no affiliations; self-cites written as "prior work (Anonymous, 2023)" rather than "our earlier work 4]".
BEFORE: "Aim for about 12 pages total." AFTER: "8 pages of main text; references and appendix do not count toward the limit."
BEFORE: "15 pages, any readable font." AFTER: "15 pages, 11pt minimum font, 1-inch margins on all sides."
BEFORE: "Export figures at screen resolution (72 dpi), RGB only." AFTER: "Export at 300+ dpi; RGB or CMYK is accepted."
BEFORE: "Make it A0, roughly 1 x 1.5 m, font ~18pt." AFTER: "A0 = 841 x 1189 mm; title 60-85pt, headers 36-48pt, body 24-32pt; add a QR code."
SINGLE-column. Decide column count per venue, never by paper-vs-journal.
version restores author names for the same double-blind venues.
^1vs Science parenthetical (1). Do not collapse them into bracketed [1].
[1] (Vancouver) citations.bracket [1], not anonymized.
^1 for Nature. DON'T use [1] or (Author, year) for Nature.(1) for Science. DON'T use superscript for Science.[1] everywhere (wrong for Nature, Science, Cell Press, ACL/EMNLP).| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-17 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-24 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-23 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
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. 24 cases were attempted. The headline lift of +21 percentage points is the difference between those two pass rates over the 24 comparable cases.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.5-flash | verified | 7/9/2026 | +54% |
Other measured skills in the registry, with their headline benchmark lift.