research-paper-writing

Write ML papers for NeurIPS/ICML/ICLR: design→submit.

Quality

63%

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

Fix and improve this skill with Tessl

tessl review fix ./skills/research/research-paper-writing/SKILL.md

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 and exceptionally well-structured with strong validation checkpoints and feedback loops, but its ~2300-line length, repeated examples, and large inline LaTeX-tooling sections cost it conciseness and some progressive-disclosure credit.

Suggestions

Move the professional LaTeX preamble, TikZ diagram patterns, siunitx/subfigure, and SciencePlots material into a dedicated references/latex-tooling.md file and link to it from the main body to reduce inline bulk.

Deduplicate the booktabs table example and the figure/caption rules, which currently appear in both Phase 4.4 and Phase 5, to tighten conciseness.

Trim general writing heuristics (title/abstract advice, Gopen & Swan principles already in writing-guide.md) in the body, keeping only the pointer to the reference.

Dimension	Reasoning	Score
Conciseness	At ~2300 lines the body is concrete and domain-specific, but it repeats material (the booktabs table example appears in both Phase 4.4 and Phase 5; figure rules are restated) and includes guidance Claude largely already knows (title/abstract heuristics, basic LaTeX), fitting the level-2 anchor of mostly efficient with some unnecessary or tighten-able content.	2 / 3
Actionability	It provides extensive copy-paste-ready, executable artifacts — the cost tracker, doi_to_bibtex, result aggregation, full LaTeX preamble, TikZ patterns, and chktex/validation scripts — matching the level-3 anchor of fully executable code and specific examples.	3 / 3
Workflow Clarity	The phased 0–8 structure with numbered steps, explicit validation checkpoints (citation 5-step process, pre-compilation validation, "only when valid" compilation gating) and feedback loops matches the level-3 anchor of a clear sequence with explicit validation and error-recovery loops.	3 / 3
Progressive Disclosure	Nine reference files are clearly signaled with one-level-deep markdown links and a summary table, but large inline blocks (the professional LaTeX preamble, TikZ diagram patterns, SciencePlots, siunitx/subfigure sections) could plausibly live in their own reference file, fitting the level-2 anchor where some content that should be separate is inline.	2 / 3
	Total	10 / 12 Passed

Description

50%

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 concise and names the right domain and venues, but it is too terse to convey the full pipeline's scope and lacks an explicit "Use when…" trigger. It sits at level 2 across all dimensions.

Suggestions

Add an explicit trigger clause, e.g. "Use when writing or revising an ML/AI research paper for NeurIPS, ICML, ICLR, ACL, AAAI, or COLM — from experiment design through submission and rebuttal."

Expand the action list to reflect the full pipeline: "design and run experiments, analyze results, draft and revise the paper, prepare camera-ready and rebuttal materials".

Differentiate from the superseded ml-paper-writing skill by noting this covers the full experiment→analysis→writing lifecycle, not just writing.

Dimension	Reasoning	Score
Specificity	The phrase "Write ML papers for NeurIPS/ICML/ICLR: design→submit" names the domain and a few actions (write, design, submit) but does not enumerate the pipeline's many concrete capabilities (experiments, analysis, rebuttals, formatting), matching the level-2 anchor that names a domain and some actions but is not comprehensive.	2 / 3
Completeness	It states what the skill does ("Write ML papers") and implies the when via venue names, but there is no explicit "Use when…" trigger clause; per the judging guidelines a missing explicit trigger caps completeness at 2.	2 / 3
Trigger Term Quality	"ML papers" and the venue names "NeurIPS/ICML/ICLR" are natural terms a user might say, but common variations users actually use ("research paper", "experiments", "submission", "rebuttal", "camera-ready") are absent, fitting the level-2 anchor of some relevant keywords missing common variations.	2 / 3
Distinctiveness Conflict Risk	The specific venue list gives the skill a niche, but the description does not distinguish it from the closely related ml-paper-writing skill it claims to supersede, so it could still overlap with similar skills — the level-2 anchor.	2 / 3
	Total	8 / 12 Passed

Validation

62%

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 — 10 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (2378 lines); consider splitting into references/ and linking	Warning
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 2 missing	Warning
referenced_paths_exist	Referenced path issues: 2 missing	Warning

	Total	10 / 16 Passed

Repository: NousResearch/hermes-agent
Path: skills/research/research-paper-writing/SKILL.md
Commit: 74a56b7

Reviewed: 3 days 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.