research-paper-writing

Guides end-to-end ML research paper production from experiment design through conference submission.

Updated Aug 19, 2026
One-click install
npx skills add https://github.com/jakubbartnik/honey-barrel-finale --skill research-paper-writing-jakubbartnik
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: research-paper-writing
Source: https://github.com/jakubbartnik/honey-barrel-finale/tree/main/honey-barrel-finale/workspace/hermes/skills/research/research-paper-writing
Command: npx skills add https://github.com/jakubbartnik/honey-barrel-finale --skill research-paper-writing-jakubbartnik

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires semanticscholar, arxiv, habanero, requests, scipy, numpy, matplotlib, SciencePlots, and includes references (resource) components.

What problem does it solve? Writing a publication-ready ML/AI paper involves coordinating experiments, verified citations, statistical analysis, LaTeX formatting, and venue-specific checklists, and mistakes in any step cause desk rejections or hallucinated references. ## Core Features & Use Cases - Full Research Lifecycle: Covers project setup, literature review, experiment design, execution monitoring, statistical analysis, drafting, self-review, and submission for NeurIPS, ICML, ICLR, ACL, AAAI, and COLM. - Citation Verification Workflow: Fetches BibTeX programmatically via Semantic Scholar, CrossRef, and arXiv APIs instead of generating citations from memory, marking unverifiable ones as placeholders. - Venue Templates and Checklists: Ships official LaTeX templates and mandatory checklist requirements (NeurIPS 16-item checklist, ACL Limitations section, ICLR LLM disclosure) to avoid desk rejection. - Use Case: A researcher with experimental results asks the agent to draft an ICML submission; the skill produces a LaTeX draft grounded in an experiment log, with verified citations, error bars, and the broader impact statement. ## Quick Start Ask the agent to help write a research paper from your experiment results, specifying the target venue such as ICML or NeurIPS.

Frequently Asked Questions about research-paper-writing

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a machine learning paper for NeurIPS or ICML?▼

Follow a phased pipeline: define a one-sentence contribution, map each claim to an experiment, run baselines with statistical tests, then draft using the venue's official LaTeX template. Complete the mandatory checklist (NeurIPS has 16 items) before submission to avoid desk rejection.

How do I prevent hallucinated citations in AI-assisted paper writing?▼

Never generate BibTeX from memory; fetch it programmatically via DOI content negotiation from CrossRef or the Semantic Scholar API. Verify each paper exists in at least two sources and confirm the cited claim appears in the abstract, marking failures as [CITATION NEEDED].

What LaTeX templates are included for conference submissions?▼

The templates directory includes official style files for ICML 2026, ICLR 2026, NeurIPS 2025, ACL, AAAI 2026, and COLM 2025, each with its .sty and .bst files plus example documents. Compile with pdflatex and bibtex, or upload to Overleaf.

Which conferences require a limitations section or paper checklist?▼

NeurIPS requires a 16-item checklist covering claims, reproducibility, and ethics; ACL mandates a Limitations section outside the page limit; ICML requires a Broader Impact Statement; ICLR 2026 requires LLM usage disclosure when LLMs contributed significantly to research or writing.

How should I design human evaluation for an NLP paper?▼

Use pairwise comparisons over Likert scales for reliability, recruit at least 3 annotators per item, and report Krippendorff's alpha for inter-annotator agreement. Document annotator qualifications, compensation, and attention checks, since ACL venues explicitly require these details.

When should I use iterative refinement versus single-pass drafting?▼

Use autoreason-style iterative refinement for mid-tier models on constrained tasks, where the generation-evaluation gap is largest. For frontier models on unconstrained tasks, single pass or critique-and-revise performs better because synthesis drift prevents convergence.