OPEN LINK ↗
// 4h agoOPENSOURCE RELEASE
figures4papers turns paper plots into agent skill
Yale CS PhD candidate Chen Liu open-sourced Python scripts behind publication figures from Nature Machine Intelligence, ICML, NeurIPS, and ECCV, alongside a scientific-figure-making skill for AI coding agents. It guides agents toward publication-ready bars, radar plots, trends, and multi-panel figures exported as PNG and PDF.
// ANALYSIS
The real contribution is packaging researcher's visual taste as reusable agent instructions, not merely releasing plotting snippets.
- –Includes real project-specific scripts, demos, design theory, API conventions, and tutorials.
- –Works through path-based prompting or symlink installation in Cursor, Claude Code, and Codex.
- –Covers quantitative comparisons, composition breakdowns, radar plots, line charts, concept plots, and trends.
- –It is a reference toolkit rather than a polished plotting library, so developers still need to adapt scripts and visually inspect outputs.
- –This could meaningfully reduce repetitive figure-polish work for ML researchers while keeping final design judgment human-led.
// TAGS
figures4papersopen-sourcedevtoolcode-generationai-codingresearch
DISCOVERED
4h ago
2026-11-11
PUBLISHED
5h ago
2026-11-10
RELEVANCE
8/ 10
AUTHOR
0x0SojalSec