Generate publication-quality matplotlib figures and LaTeX tables from experiment data files.
npx skills add https://github.com/zjunlp/Mechanist --skill paper-figure-zjunlp
Please help me install this Agent Skill. Skill: paper-figure Source: https://github.com/zjunlp/Mechanist/tree/main/skills/paper-figure Command: npx skills add https://github.com/zjunlp/Mechanist --skill paper-figure-zjunlp
latex_includes.tex for paper writing; auto-ledger mode produces PDF+PNG per claim with a machine-readable INDEX.json for orchestrator integration. - First-class table support: Generates comparison tables as LaTeX (.tex) and Markdown (.md) with mirrored content for ledger embedding and paper inclusion. - Use Case: After running an ablation study, ask for paper figures and receive reproducible per-figure Python scripts, vector PDFs at 300 DPI, and ready-to-paste LaTeX include snippets. ## Quick Start Ask the assistant to generate publication-quality figures from your experiment results in runs/ or figures/ using the paper-figure skill.High-intent search queries and answers about installing and using this skill.
Provide a figure plan or let the skill scan your data files, and it creates one Python script per figure using a shared publication style config. Outputs are vector PDFs at 300 DPI with serif fonts, no in-figure titles, and consistent color palettes.
Line plots, bar charts, grouped bars, scatter plots, heatmaps, box/violin plots, multi-panel subfigures, and comparison tables are auto-generated. Architecture diagrams, generated image grids, and screenshots must be created manually.
Yes, comparison tables are first-class outputs. Standalone mode writes a .tex file, while auto-ledger mode writes both a Markdown table for inline ledger embedding and a mirrored LaTeX table for paper inclusion.
Standalone mode defaults to vector PDF for LaTeX inclusion. Auto-ledger mode writes both PDF for later paper use and PNG for Markdown embedding, with each figure script selecting formats based on its type.
In auto-ledger mode the entry is recorded as skipped in INDEX.json without raising an error, and script failures are captured as status error entries. The batch continues so one bad figure never aborts the rest.
EmaRimoldiAggregate experiment outputs into structured results with gap reporting.
dz306271098Generate publication-quality figures and LaTeX tables from JSON or CSV experiment results.
EvoScientistApply a 5-aspect checklist, reverse-outlining, and figure-quality checks to academic paper drafts.
arih04xGenerate editable research figures for Word, PowerPoint, and LaTeX from draw.io or TikZ sources.