完成数据分析全流程

Data Analytics Skills for Claude & Codex

31 portable AI-powered skills that turn Claude or Codex into a hands-on analytics partner

No setup required · Works for any company or industry

Version: 1.2.0  ·  Maintainer: Nimrod Fisher  ·  Last Updated: September 2026

🚀 Quick Start

Install

Each skill is a folder containing a SKILL.md (plus optional scripts/, references/, and assets/). Clone the repo and copy the skills you want into your agent's skills directory:

git clone https://github.com/nimrodfisher/data-analytics-skills.git

Claude Code: personal skills (all projects) go in ~/.claude/skills/, project skills in .claude/skills/ inside your project:

cp -r 01-data-quality-validation/programmatic-eda ~/.claude/skills/

Codex: user skills (all repos) go in ~/.agents/skills/, repo skills in .agents/skills/ at the repository root:

cp -r 01-data-quality-validation/programmatic-eda ~/.agents/skills/

New here? QUICKSTART.md walks you through your first skill in 5 minutes, and STRUCTURE.md explains how every skill is packaged.

[!TIP] Describe your task to Claude or Codex naturally — it will select and activate the right skill automatically. No slash commands needed.

Example:

You:    "I need to understand why our activation rate dropped 12% last week"
Agent:  [activates root-cause-investigation, asks for metric data and context]
You:    [provides data and business context]
Agent:  [runs structured investigation with hypothesis testing]

Which skill to start with?

📖 How skills work

Each skill follows the same on-demand context pattern:

1. Request minimum viable context — your agent asks only what's essential to start
2. Execute the workflow — structured, step-by-step analytical process
3. Surface assumptions — anything uncertain is flagged, not silently assumed
4. Deliver a consistent output — templated result you can share or iterate on

[!NOTE] Skills degrade gracefully: if you can't provide everything, your agent states what it's assuming and proceeds.

🛠️ Customization

Skills work out-of-the-box. To make them company-specific, add a references/ folder inside any skill with:

skill-name/
├── SKILL.md
└── references/
    ├── company-schema.md       ← your table/column definitions
    ├── metric-definitions.md   ← standard metric formulas
    └── business-rules.md       ← thresholds, edge cases, etc.

Your agent will pull this context automatically when the skill runs.

🎓 Suggested ramp-up

Week 1 — Get comfortable

• Run programmatic-eda on a familiar dataset
• Practice providing context when your agent asks
• Use analysis-planning at the start of your next project

Week 2–3 — Add your core toolkit

• Set up semantic-model-builder for your key metrics (saves time forever)
• Add query-validation to your SQL workflow
• Pick 2 analysis skills that match your domain

Week 4+ — Go advanced

• Chain 4–5 skills end-to-end on a full project
• Add company-specific references to the skills you use most
• Build team context documents for shared onboarding