Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when setting up a paper/proposal for a named academic venue: apply that venue's exact LaTeX class, anonymization rule, page limit, and citation style.
.claude/skills/venue-submission-format/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 27 |
| Model | Lift | Δ tokens | Δ turns | Cases | Verified |
|---|---|---|---|---|---|
| gemini-3.5-flashbest | +33% | — | 0% | 24 | 86d ago |
| gemini-3.6-flash | +14% | +90% | 0% | 22 | 54d ago |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | — | — |
| case-02 | ✗→✓ | ▲ Improved | — | — |
| case-12 | ✗→✓ | ▲ Improved | — | — |
| case-01 | ✓→✓ | = Same ✓ | — | — |
| case-09 | ✗→✗ | = Same ✗ | — | — |
When a task names a specific academic venue (a conference, journal, grant agency, or poster format), emit that venue's EXACT machine-checkable setup — LaTeX class/style token, anonymization state, page limit, citation/bibliography style — never a generic article default. The mapping from venue to format is arbitrary and venue-controlled; guessing a plausible-looking default is wrong.
| Venue | documentclass | style package | |---|---|---| | NeurIPS 2024 | \documentclass{article} | \usepackage{neurips_2024} | | ICML 2024 | \documentclass{article} | \usepackage{icml2024} | | ICLR 2024 | \documentclass{article} | \usepackage{iclr2024_conference} | | CVPR | \documentclass[10pt,twocolumn,letterpaper]{article} | \usepackage[review]{cvpr} | | ACL / EMNLP / NAACL | \documentclass[11pt]{article} | \usepackage[review]{acl} | | IEEE conference | \documentclass[conference]{IEEEtran} | (class only) | | ACM (camera-ready) | \documentclass[sigconf]{acmart} | + \setcopyright{...} | | Springer LNCS | \documentclass{llncs} | (class only) | | Elsevier journal | \documentclass{elsarticle} | (class only) | | PLOS | \documentclass[10pt,letterpaper]{article} | (PLOS template) |
The ML conferences (NeurIPS/ICML/ICLR) share the plain article base class and differ ONLY in the style package — the package name is the load-bearing token.
The review version must contain NO author names, NO affiliations, NO email addresses, NO acknowledgments, and NO funding statement.
NeurIPS [final], ICML [accepted], ICLR \iclrfinalcopy, CVPR / ACL drop [review].
never "in our previous work x]" — first-person self-reference de-anonymizes.
names and affiliations appear in the submission.
9 main-text pages.
| Venue | Citation style | bibliographystyle | |---|---|---| | NeurIPS / ICML / ICLR | numeric via natbib (\citep{}, \citet{}) | — | | ACL family | natbib (\citep, \citet) | — | | IEEE | numeric in square brackets | \bibliographystyle{IEEEtran} | | ACM | ACM Reference Format | \bibliographystyle{ACM-Reference-Format} | | Springer LNCS | numeric | \bibliographystyle{splncs04} | | Elsevier | numeric | \bibliographystyle{elsarticle-num} | | PLOS | numbered Vancouver (numeric, NOT author-year) | — |
approved fonts include Times New Roman, Computer Modern, Arial, Palatino Linotype.
(Arial, Helvetica, Georgia, Palatino Linotype); margins ≥ 0.5 inch.
beamerposter or tikzposter; scale fonts large fordistance reading.
\documentclass[11pt]{article}\usepackage{neurips} — invents aplausible package name and a font option.
\documentclass{article} then \usepackage{neurips_2024} — exactpackage name, no font option.
\author{Jane Doe\\MIT\\jane@mit.edu} plus an acknowledgments line — the model'sdefault is to fill in author metadata.
\author{Anonymous Author(s)} (or omitted), no affiliation, noemail, no acknowledgments, no funding statement.
[1], with \bibliographystyle{IEEEtran}.poster package; A0 is about 33×46 inches" — wrong package, imperial guess.beamerposter (or tikzposter); A0 = 841 × 1189 mm.not change their class file between versions — they flip an option. Camera-ready NeurIPS is \usepackage[final]{neurips_2024}; the review version omits [final]. Never delete the package to "de-anonymize."
\usepackage[review]{cvpr}IS the anonymized state; dropping [review] (or using [final]) is the camera-ready state.
by counting them.
submission yet excluded from the 9-page main-text limit.
(page count, margins, font, font size), not a \documentclass. Do not answer a grant question with a LaTeX class.
anonymization applies only to the double-blind set in R2.
\usepackage{neurips}; always load \usepackage{neurips_2024} (exact year tag).anonymize them and de-anonymize only in the camera-ready via the venue's toggle.
scheme the venue mandates.
neurips, icml, iclr_conf) instead of the exact tokens(neurips_2024, icml2024, iclr2024_conference).
[11pt] / [12pt] font option to the ML-conference \documentclass{article} — thestyle package controls the font; the base class takes no font option.
(Author, Year) citations everywhere, when IEEE/ACM/LNCS/Elsevier/PLOSrequire a numeric scheme.
[final] tothe review system).
poster package that doesn't exist.article)?beamerposter/tikzposter?| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.5-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 +14 percentage points is the difference between those two pass rates over the 22 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 | 6/27/2026 | +33% |
Other measured skills in the registry, with their headline benchmark lift.