evident-charts: open-source skill teaches AI coding agents to make clear, honest charts
randal_olson · x · 2026-09-26
Data visualization researcher Randy Olson released evident-charts, an open-source agent skill that distills visualization research and expert consensus into rules for AI coding agents (Claude Code, Codex, Cursor, Gemini and more).
How it works
- Auto-loads whenever the agent makes or reviews a chart; you can also ask it to critique and fix an existing chart (attach the PNG, the script, or both)
- Checks the data first: totals mixed with parts, duplicate rows, placeholder codes, preliminary months
- Writes the takeaway as the title; checks in code for color-blind conflicts and overlapping labels; a fresh reviewer inspects the rendered image
- Every rule is tagged by support level (experimental evidence, practitioner consensus, or taste) and cited
The README shows a typical bad draft (raw counts on a log scale, EU total ranked among its own members, meaningless rainbow colors) before and after critique and rebuild.
Related event: Open-source agent skill teaches AI to make honest charts(3 posts)→
More from coding & agent
- Dev builds SDF modeling engine from scratch in C++/Vulkan with AI-generated scenes — bilawalsidhu · 2026-09-27
- MCP isn't the USB-C of AI: LLMs improvise, so put a deterministic runtime in charge — ThomasBuildLab · 2026-09-27
- Devs Build a Roblox Game With Opus 5.5, Cancel ChatGPT Subscription — cringe_keys · 2026-09-27
- Indie Dev Adds MCP to His Website Monitoring Tool So AI Agents Can Check Sites — bincura · 2026-09-27
- Replacing Zapier with conversational agents: engineering 100% deterministic workflows — JanBreslau · 2026-09-27
- Claude Opus 5.5 One-Shots a Launch Video, Hailed as Solving Video Animation — JosephJacks_ · 2026-09-27