Notre avis

Génère des sections d'article académique (résumé, introduction, méthodes, expériences, conclusion, annexe) en LaTeX à partir d'un guide de style extrait.

Points forts

Produit un LaTeX valide et structuré selon des squelettes logiques
Évite les mots interdits et respecte la voix passive
S'appuie sur un guide de style et un lexique de domaine pour une cohérence académique

Limites

Nécessite d'abord d'extraire le style via une commande dédiée
Ne fonctionne que pour les sections prédéfinies
La qualité dépend de la richesse du lexique et du guide de style

Quand l'utiliser

Lorsque vous devez rédiger une ou plusieurs sections d'un papier académique en LaTeX avec une cohérence de style et de terminologie.

Quand l'éviter

Pour une rédaction libre sans contrainte de format académique ou lorsque le style n'a pas encore été extrait.

Exemples

Generate abstract

Generate the abstract section for my paper on deep learning for image segmentation. The full draft is in draft.md.

Write introduction

Write the introduction section using the style guide. Problem: improving model robustness. Method: adversarial training. Background: recent advances in robustness.

Create methods section

Generate the methods section in LaTeX. The algorithm is in algocode.py and the style is already extracted.

name: paper-gen description: Generate academic paper sections (Abstract, Intro, Methods, Experiments, Conclusion, Appendix) in LaTeX format based on extracted style. allowed-tools:

Read
Write
Glob
AskUserQuestion

Paper Content Generation Skill

Goal: Generate specific paper sections in valid LaTeX format using the extracted style and domain knowledge.

Prerequisite Check: Before proceeding, check if psmfiles/STYLE_GUIDE.md exists.

If MISSING: HALT immediately. Reply: "❌ Missing Style Assets. Please run /paper-extract-style first to build the knowledge base."
If FOUND: Proceed with the generation task.

Usage

User will specify the section to generate (e.g., "intro", "method", "abstract", "experiment", "conclusion", "appendix").

Context Loading

Read psmfiles/STYLE_GUIDE.md (Style rules).
Read psmfiles/lexicon_domain.md (Domain sentence bank).
Read paper-gen/LEXICON.md (Logical skeletons).

Generation Tasks

1. Abstract

Input: Full draft or summary.
Structure: Follow 'Abstract/Logical Skeleton' in LEXICON.md.
Constraints:
- Third-person passive voice.
- No forbidden words (delve, underscore, realm, pivotal, intricate, meticulously, comprehensive).
Output: Save valid LaTeX to psmfiles/abstract.tex.

2. Introduction

Input: Problem, Method, Background.
Structure: Follow 'Introduction/Logical Skeleton' (Move 1-3 Model).
Constraints: Same as Abstract.
Output: Save valid LaTeX to psmfiles/section_intro.tex.

3. Methods

Input: Source code or algorithm description.
Role: "Ghostwriter" for the code.
Structure: Follow 'Methods/Logical Skeleton'.
Constraints: Same as Abstract.
Output: Save valid LaTeX to psmfiles/section_methods.tex.

4. Experiments

Input: Results, datasets, setup.
Structure: Follow 'Experiments/Logical Skeleton'.
Constraints: Same as Abstract.
Output: Save valid LaTeX to psmfiles/section_experiments.tex.

5. Conclusion

Input: Draft or outline.
Structure: Follow 'Conclusion/Logical Skeleton'.
Constraints: Same as Abstract.
Output: Save valid LaTeX to psmfiles/section_conclusion.tex.

6. Appendix

Input: Code, proofs, extra figures.
Structure: Follow 'Appendix/Logical Skeleton'.
Formatting: Strictly follow 'Appendix' style in STYLE_GUIDE.md.
Output: Save valid LaTeX to psmfiles/appendix.tex.

Génération de sections d'articles académiques

Recommandé pour