Agent skills for data science (Python, Jupyter, pandas) projects
95 open-license skills in our database are built for data science (Python, Jupyter, pandas), and 638 skills match a typical data science (Python, Jupyter, pandas) project. Below: what coding agents get wrong on data science (Python, Jupyter, pandas), 3 picks to start with, and where each agent loads them.
Updated 2026-10-18. Tags used for this page: python, jupyter, data, ml.
What coding agents get wrong in data science (Python, Jupyter, pandas)
Agents write notebooks that break on a rerun, chain pandas calls that copy data, and leak test data into training. Plots come out without labels or units.
Skills that help
Skills for pandas, scikit-learn and plotting practice, plus notebook hygiene and reproducible experiments.
Cursor rules: the database holds 13 Cursor rule files (.mdc) tagged data science (Python, Jupyter, pandas) (data-analysis-initial-exploration, data-visualization-rules, error-handling-and-data-validation-rules, general-data-analysis-and-jupyter-notebook-rules, jupyter-notebook-best-practices, pandas, and more). Cursor reads them from .cursor/rules/. Other agents load SKILL.md skills, so most picks here are skills.
3 picks to start with
1. data-visualization
Create effective data visualizations with Python (matplotlib, seaborn, plotly).
npx skills add anthropics/knowledge-work-plugins --skill data-visualization
2. jupyter-notebook
Use when the user asks to create, scaffold, or edit Jupyter notebooks (.ipynb) for experiments, explorations, or tutorials; prefer the bundled templates and...
npx skills add openai/skills --skill jupyter-notebook
3. matplotlib-best-practices
Best practices for Matplotlib data visualization, plotting, and creating publication-quality figures in Python
npx skills add Mindrally/skills --skill matplotlib-best-practices
Get the list for your exact project
These 3 fit data science (Python, Jupyter, pandas) in general. The scan box opens with a sample data science (Python, Jupyter, pandas) description you can edit. A scan reads your repo or description, shows the match count and 2 picks free, and the full curated list (up to 30 picks with install commands, reasons and overlap notes) costs $9 once.
Scan a data science (Python, Jupyter, pandas) project free Use the MCP tool
Where each agent loads data science (Python, Jupyter, pandas) skills
- Claude Code: project folder
.claude/skills/, install flag--agent claude-code - Cursor: project folder
.cursor/skills/ or .agents/skills/, install flag--agent cursor - Codex: project folder
.agents/skills/, install flag--agent codex - GitHub Copilot: project folder
.github/skills/ or .agents/skills/, install flag--agent github-copilot
npx skills add owner/repo --skill skill-name --agent claude-code
How we count
We match project tags against 86,803 skills from 5,854 public GitHub repos (refreshed 2026-10-18). A skill built for another framework or language is left out, and so is a skill below our relevance floor. The picks on this page come from the ranked list for these tags. We favor broad practice skills over single-task ones. Licenses are shown as each repo declares them. Read a skill before you install it.
Questions
What agent skills should I use for data science (Python, Jupyter, pandas)?
Start with a broad practice skill built for data science (Python, Jupyter, pandas), then add skills for testing and review. In our database 95 skills are tagged for data science (Python, Jupyter, pandas) and 638 skills match a typical data science (Python, Jupyter, pandas) project. The picks on this page are a place to start. A scan of your repo ranks the full list for your tasks.
Do these skills work in Jupyter?
They work in any agent that edits your notebooks or scripts, such as Claude Code, Cursor or Codex. The skill tells the agent how to work, not where.
Are there Cursor rules for data science (Python, Jupyter, pandas)?
Yes. The database holds 13 Cursor rule files (.mdc) tagged data science (Python, Jupyter, pandas). Cursor loads them from .cursor/rules/. Claude Code, Codex and Copilot do not read .mdc rules, so pick SKILL.md skills for those agents.
Where do data science (Python, Jupyter, pandas) skills go for Claude Code, Cursor, Codex and Copilot?
Claude Code reads .claude/skills/, Cursor reads .cursor/skills/ or .agents/skills/, Codex reads .agents/skills/, and GitHub Copilot reads .github/skills/ or .agents/skills/. The install command takes --agent with the agent name and copies the skill folder there.
Why 3 picks here?
The right list depends on your exact stack and tasks. Run a free scan for your project to see its match count and 2 picks. The full curated list is $9 once.
Related stacks
Python skillsPyTorch skillsHugging Face skillsFastAPI skills
All stacks
Next.js skillsReact skillsPython skillsFastAPI skillsDjango skillsReact Native and Expo skillsSupabase skillsRust skillsGo skillsdata science (Python, Jupyter, pandas) skillsVue and Nuxt skillsAngular skillsSvelte and SvelteKit skillsTailwind CSS skillsNode.js and Express skillsLaravel skillsRuby on Rails skillsSpring Boot skillsJava skills.NET and C# skillsC++ skillsiOS and SwiftUI skillsSwift skillsAndroid and Jetpack Compose skillsFlutter and Dart skillsTerraform skillsKubernetes skillsMongoDB skillsPyTorch skillsHugging Face skillsApache Airflow skillsdbt skillsShopify skillsWordPress skillsSalesforce skillsn8n skillsSolidity and smart contracts skills
By agent
Skills for Claude CodeSkills for CursorSkills for CodexSkills for GitHub Copilot