ChenLiu-1996

Paper-Grade Figures, Drawn by an Agent: figures4papers I've actually needed to draw decent-looking infographics myself. I wrote a lot of "recipe" prompts for AI to do it, and I even run a related SaaS site. Then, a few days ago while browsing GitHub, I came across a really nice repo. It's full of scripts — and a skill — for drawing publication-grade figures. It's developed and maintained by a Chinese student studying overseas, and it's called figures4papers. GitHub repo: ChenLiu-1996/figures4papers I'd known another good skill before this one, called vivid-figures-skill. Using figures4papers is dead simple. You don't even need to download anything — just send the repo link to your agent and tell it to go make the figure. It's genuinely useful — real research productivity. And honestly it's nicer than the vivid one, at least in terms of aesthetics. Back in the pre-AI era, the illustrations in a paper cost a lot of money if you hired a professional to make them, and a huge amount of effort if you did them yourself (and in the end you'd usually hire someone anyway — because you can copy a figure's layout, but good taste is much harder to copy). Written by Terry, who also runs Airgrix, an airsoft gear store at airgrix.com.

科研人画论文图，不用每次都从 Matplotlib 配色开始折腾了。 figures4papers 就是一款给 Codex、Claude Code、Cursor 用的论文绘图 Skill。 数据交给它，柱状图、折线图、雷达图、趋势图这些论文级图表就能生成出来。 ChenLiu-1996/figures4papers t.co

@xxxgrl5 画完还得再改反正 ChenLiu-1996/figures4papers

谁说只有数字人，我还有数字狗🐶。 柴犬主播给你分享了一个贼好的科研狗大喜的项目。 以及有9K Star ，用户制作图表、多种图形~ 地址：ChenLiu-1996/figures4papers t.co

耶鲁计算机博士生开源了自己的论文画图脚本 figures4papers 柱状、雷达、折线、3D球体、概念图都支持，8个论文项目的完整画图方案都在里面，改改数据就能直接用 GitHub：ChenLiu-1996/figures4papers --- 规范细到不行——去掉上下坐标轴线、柱子加黑边、配色按语义走（蓝色自己的方法、绿色改进、红色对比基线），300 DPI 导出 直接改数据最方便，不想手动改也行，把仓库接进 Cursor 或 Claude，按 SKILL.md 的规范让 AI 帮写脚本，README 里有现成的提示词 --- 论文快deadline了还在纠结怎么出图的可以收藏备用，用顶会作者的审美标准往往不会翻车