Skill：karthik data visualization skill
Public data visualization skill for Codex and Claude
适用场景
教研备课
适合角色
介绍
以下内容整理自开源仓库 `skthewimp/karthik-data-visualization-skill` 的 README / SKILL.md 摘要： # Karthik Data Visualization Skills Public data visualization skills for Codex, Claude, and Hermes. This repo contains thirteen related skills, coordinated as a context-sensitive visualization workflow: 1. **`dataviz-fix`** - iterative repair and learning-loop rules: rebuild a pasted chart, require an independent artifact-hashed review, preserve revisions and feedback, then route reusable lessons to the owning skill. 2. **`dataviz-eval`** - artifact and creator-system evaluation rules for separate blind review, scoped send/revise/redesign decisions, failure analysis, and regression benchmarks. 3. **`dataviz-selector`** - chart-selection rules for deciding what kind of visualization fits a dataset plus question, hypothesis, or data story. 4. **`karthik-data-visualization`** - style rules for producing charts in Karthik's preferred visual language: low chartjunk, direct labels, careful typography, meaningful colour, and Tufte-inspired restraint. 5. **`karthik-powerpoint-style`** - slide and deck rules for making PowerPoint-style presentations in Karthik's analytical, claim-first style. 6. **`dataviz-critique`** - chart critique and redesign rules for diagnosing existing visuals using the question-data-visual trifecta plus Karthik's clarity-first standards, then proposing better alternatives. 7. **`karthik-analysis-planner`** - analysis-contract rules for turning fuzzy natural-language questions into operational definitions, denominators, comparisons, metrics, caveats, and falsifiers before evidence-building. 8. **`dataviz-orchestrator`** - end-to-end workflow for turning a dataset, loose question, and audience into an analysed, styled, critiqued visual story. 9. **`dataset-question-generator`** - upstream skill for profiling raw datasets and generating fresh, visualisable questions before planning or charting. 10. **`karthik-data-cleaning`** - data-cleaning rules for Karthik-style exploratory analysis: inspect, clean in context, inspect again, and avoid generic unsupervised fixes. 11. **`chart-annotations`** - annotation rules for deciding what a chart should mark, which competing candidate wins, how the label is worded, and where it sits. 12. **`chart-explainer`** - accompanying-note rules for writing the two lines that travel with a finished chart or table into an email, notebook, or message. 13. **`karthik-r-analysis-style`** - notebook rules for how an exploratory R scratchpad is written: local precedent, analyst-first probes, tidyverse/tidytable idiom, and the working-note register. The split is deliberate. The orchestrator routes the work and preserves handoffs; it does not duplicate every specialist procedure. Planning defines the analytical claim and evidence contract. Cleaning establishes provenance, grain, and data validity. Question gene ---
安装方式
1. 打开仓库：https://github.com/skthewimp/karthik-data-visualization-skill 2. 查找 SKILL.md 或 README 中的安装章节 3. 克隆或下载所需 Skill 子目录到本地 skills 路径（如 `.cursor/skills/` 或 `.claude/skills/`） 4. 重启 / 重载 Agent 客户端 5. 用一条真实任务验证 Skill 是否被正确加载
适用平台
- Claude
- Codex