ml-paper-writing

Write publication-ready ML/AI papers for NeurIPS, ICML, ICLR, ACL, AAAI, COLM. Use when drafting papers from research repos, structuring arguments, verifying citations, or preparing camera-ready submissions. Includes LaTeX templates, reviewer guidelines, and citation verification workflows.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

The canonical home for this skill is ml-paper-writing in OpenLAIR/dr-claw

Quality

Content

77%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is highly actionable with well-sequenced, validated workflows and a sensible one-level reference structure, but it is verbose for its context window: it duplicates writing-guide material inline and cites time-sensitive conference versions and a non-existent templates/ directory. Strengths are actionability and workflow clarity; conciseness and progressive-disclosure are the weak dimensions.

Suggestions

Move the inline writing-philosophy material (the 7-principles table, Perez micro-tips, and Lipton word-choice guidance) entirely into references/writing-guide.md, keeping only a one-line pointer in SKILL.md — this content is already duplicated there and inflates the body.

Relocate the time-sensitive conference requirements table ('NeurIPS 2025', 'ICML 2026', 'ICLR 2026' page limits and key requirements) into a reference file or a clearly dated 'current requirements' section so stale version numbers don't clutter the main body.

Fix the dangling templates/ references: the body repeatedly cites templates/ and templates/README.md, but no templates/ directory exists in the bundle — either ship the template directories or remove those references and the LaTeX-templates sections that depend on them.

Dimension	Reasoning	Score
Conciseness	The ~920-line body is mostly specialized actionable guidance rather than basic-concept padding, but it is inflated by inline duplication of reference material (Gopen & Swan 7 principles, Perez micro-tips, Lipton word-choice appear both here and in writing-guide.md) and by time-sensitive version tables ('NeurIPS 2025', 'ICML 2026', 'ICLR 2026') placed inline outside any deprecated section, which the guideline says to penalize — so it does not reach the lean score-3 anchor.	2 / 3
Actionability	Provides copy-paste-ready, executable guidance throughout — bash (`find . -name "*.py"`, `latexmk -pdf main.tex`, `cp -r templates/...`), Python (`semanticscholar` search, `doi_to_bibtex` via CrossRef), and LaTeX (booktabs tables, macros, figure environments) — matching the fully-executable score-3 anchor rather than the pseudocode/incomplete level at 2.	3 / 3
Workflow Clarity	Five numbered workflows (0–4) carry explicit step checklists and validation/feedback checkpoints — e.g., citation workflow's 'Verify paper exists in 2+ sources' with 'If ANY step fails → mark as placeholder, inform scientist', and template workflow's 'Verify template compiles as-is (before any changes)' — satisfying the explicit-validation score-3 anchor.	3 / 3
Progressive Disclosure	The five references/ files are real, one-level-deep, and clearly signaled with descriptive link text, but the body duplicates content that lives in writing-guide.md (content that should be separate is inline) and repeatedly points to a templates/ directory and templates/README.md that do not exist in the bundle, so navigation is partly broken and it falls short of the clean score-3 anchor.	2 / 3
	Total	10 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, trigger-rich, and clearly states both capability and use conditions with an explicit 'Use when' clause, using third-person imperative voice throughout. It is among the strongest reference examples and shows no vagueness, over-claiming, or voice violations.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'drafting papers from research repos, structuring arguments, verifying citations, or preparing camera-ready submissions' — tied to named venues, matching the 'lists multiple specific concrete actions' anchor rather than the partial action set at score 2.	3 / 3
Completeness	Explicitly answers both what ('Write publication-ready ML/AI papers... Includes LaTeX templates, reviewer guidelines, and citation verification workflows') and when via a present 'Use when...' clause, so it is not capped at 2 by the missing-trigger guideline.	3 / 3
Trigger Term Quality	Covers natural terms a researcher would actually say — 'NeurIPS, ICML, ICLR, ACL, AAAI, COLM', 'drafting papers', 'research repos', 'verifying citations', 'camera-ready submissions' — giving good coverage rather than the partial keyword set at score 2.	3 / 3
Distinctiveness Conflict Risk	Occupies a clear niche (publication-ready papers for specific top ML/NLP venues with citation-verification workflows) with distinct triggers unlikely to fire for generic writing skills, rather than the overlapping 'Works with document files' level at 2.	3 / 3
	Total	12 / 12 Passed

Validation

81%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (939 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 2 missing	Warning

	Total	13 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/writing/ml-paper-writing/SKILL.md
Commit: 52845c3

Reviewed: about 9 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.