agentleFS
Calix-LCalix-L/awesome-latex-skills/latex-fmt/SKILL.md

latex-fmt

Skill182 starsChanged 42 days agoMITView on GitHub

Reformat LaTeX papers for a specified venue, year, track, and submission stage. Apply official templates and check layout, bibliography, page limits, and anonymization while preserving scientific content.

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 8 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the root document, build configuration, bibliography backend, and included files.

    line 24
  2. Very specific:

    Read references/templates/venue-guide.md for official entry points and pitfalls.

    line 29
  3. Distinctive:

    If authoritative rules are unavailable, preserve the existing format, identify unresolved requirements, and make any provisional …

    line 34
  4. Very specific:

    Do not silently choose a previous year's kit or infer filenames by incrementing a year.

    line 36
  5. Very specific:

    Follow the kit's example preamble, options, and author macros.

    line 42
  6. Very specific:

    Do not invent neurips_2025.cls or icml2025.cls.

    line 51
  7. Distinctive:

    Use the self-contained layout example when checking local widths and resolved references outside an official kit.

    line 56
  8. Very specific:

    Check the exit status and final log, not just whether a PDF exists.

    line 74
  9. Distinctive:

    Check the compiled main-content boundary against the verified limit.

    line 80
  10. Distinctive:

    Use rules for the specified target rather than historical examples.

    line 110
  11. Very specific:

    Keep official .cls, .sty, and .bst files intact.

    line 111
  12. Distinctive:

    Do not remove author identities for camera-ready submissions by default.

    line 112
  13. Distinctive:

    Never guarantee anonymity from a source keyword search alone.

    line 113
  14. Distinctive:

    Do not submit, upload, or publish the paper as part of formatting.

    line 116

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Role

Apply the requested publication format using the matching official author kit and submission instructions. Distinguish successful compilation from verified submission compliance.

When to Activate

  • The user asks to reformat a LaTeX manuscript or switch templates.
  • The user asks for a pre-submission formatting check.
  • The user invokes /latex-fmt or $latex-fmt.

Workflow

Phase 1: Establish the target and rules

Read the root document, build configuration, bibliography backend, and included files. Establish the target venue, year, track/article type, and stage (anonymous review, preprint, or camera-ready). Ask only for missing information that changes the template or rules.

Read references/templates/venue-guide.md for official entry points and pitfalls. Fetch the matching author kit and instructions, or use the user's supplied kit. Record source URLs, check date, template filenames, page/word limits with exclusions, anonymity rules, and required sections/checklists.

If authoritative rules are unavailable, preserve the existing format, identify unresolved requirements, and make any provisional conversion's assumptions clear. Do not silently choose a previous year's kit or infer filenames by incrementing a year. This repository contains guidance, not official .cls, .sty, or .bst files.

Phase 2: Apply the official template

Follow the kit's example preamble, options, and author macros. .cls files belong in \documentclass; .sty files belong in \usepackage. For example, the historical NeurIPS 2025 review pattern is:

\documentclass{article}
\usepackage{neurips_2025} % use the requested year's actual kit

Do not invent neurips_2025.cls or icml2025.cls. Use the kit's bibliography style rather than assuming the style package supplies a .bst. Remove custom margin or spacing overrides only when they conflict with verified rules.

Consult formatting rules for layout guidance. Use the self-contained layout example when checking local widths and resolved references outside an official kit. It is an example, not a substitute for the requested venue's template. Preserve text, math, table values, citations, labels, and figure contents. Flag missing required sections for the author; do not generate unsupported ethical or empirical claims.

Phase 3: Bibliography and build

Respect the project's engine and bibliography backend. Prefer its existing build command or latexmk when available. For manual builds, run the engine, the actual BibTeX/Biber backend, and additional passes to resolve references. Do not run pdflatex on a fontspec project or select Biber solely because biblatex is present; check its backend option.

Preserve textual (\citet/\textcite) versus parenthetical (\citep/\parencite) meaning when translating bibliography commands. Never blanket replace every citation with \cite or \citep.

Check the exit status and final log, not just whether a PDF exists. Resolve format-related errors with latex-rescue if available. Report missing tools or template files as unverified compilation.

Phase 4: Length and anonymity

Check the compiled main-content boundary against the verified limit; total PDF pages cannot establish compliance when references or appendices are excluded. Count words according to the article type. Suggest cuts before moving sections or reducing figures. Avoid margin, font-size, or negative-spacing tricks.

For anonymous review, use the official review mode and inspect the main PDF, supplement, author blocks, acknowledgments, funding, self-citations, URLs, and PDF metadata. Flag identifying scientific text for contextual revision; keep citation keys intact. Remove identity metadata in source, rebuild, and inspect the result. An in-place metadata scrub does not prove anonymity. Camera-ready and preprint author information should follow their own rules.

Phase 5: Verification and report

Inspect the rendered PDF for clipped floats, readability, fonts, citations, and applicable anonymity requirements. Check checklist location and required sections against official rules. Report:

  • Target venue/year/track/stage and authoritative sources with check date.
  • Template and bibliography changes, with a reviewable diff.
  • Build result, warnings, and actual page/word boundaries measured.
  • Requirements as pass, fail, or unverified, with evidence.
  • Missing author content or decisions and remaining actions.

A successful build is not a submission-ready claim. Use provisional/unverified status if a required rule, build, or visual check could not be confirmed.

Guardrails

  • Preserve scientific content; do not invent citations or missing sections.
  • Use rules for the specified target rather than historical examples.
  • Keep official .cls, .sty, and .bst files intact.
  • Do not remove author identities for camera-ready submissions by default.
  • Never guarantee anonymity from a source keyword search alone.
  • Preserve caption/author interfaces and textual versus parenthetical citations; package or command substitutions need verification in the actual kit.
  • Do not submit, upload, or publish the paper as part of formatting.

Reference Files

  • references/templates/venue-guide.md — official venue entry points and template pitfalls.
  • references/formatting-rules.md — layout, citation, and project organization guidance.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.