Skip to content

Repository files navigation

askprisma — AI Data Analysis Skill

npm version License: MIT Works with Claude Code Works with Gemini CLI Works with Codex CLI

Business data analysis skill for AI coding agents. Turns raw data into actionable business intelligence — from CSV exploration to professional PDF reports. Works with Claude Code, Gemini CLI, Codex CLI, OpenCode, and any agent that reads ~/.claude/skills/.


What It Does

/askprisma runs a structured 5-phase analysis workflow:

  1. Data Discovery — profiles all data files or SQL tables (row counts, types, nulls, cardinality, statistics)
  2. Analysis Planning — presents a numbered plan tailored to your data and question; waits for approval
  3. Iterative Execution — writes and runs Python code task by task, fixing errors automatically
  4. Business Synthesis — translates findings into plain language with specific numbers and "so what" implications
  5. PDF Report Generation — optionally generates a professional PDF in executive summary, comprehensive, or slide format

Installation

Option 1 — npx (installs for all detected agents at once)

npx askprisma-skill

Detects which AI tools are present and copies the skill into each one's standard directory:

Agent Install Path
Claude Code / cowork ~/.claude/skills/askprisma/
Gemini CLI ~/.gemini/extensions/askprisma/
Codex CLI ~/.codex/skills/askprisma/
OpenCode ~/.config/opencode/skills/askprisma/

Option 2 — npx skills add (Vercel skills ecosystem)

npx skills add whiteboardmonk/askprisma-skill

Or install for a specific agent only:

npx skills add whiteboardmonk/askprisma-skill --agent claude
npx skills add whiteboardmonk/askprisma-skill --agent cursor

Preview what will be installed without installing:

npx skills add whiteboardmonk/askprisma-skill --list

Option 3 — Manual (Claude Code only)

git clone https://github.com/whiteboardmonk/askprisma-skill
ln -s "$PWD/askprisma-skill/skills/askprisma" ~/.claude/skills/askprisma

Usage

Basic analysis

/askprisma What are the main trends in sales_data.csv?

With a focused question

/askprisma Which product categories are driving churn this quarter?

Request a PDF report upfront

/askprisma Segment our customers and generate an executive summary PDF

After analysis completes, the agent will ask if you want a PDF. Choose from:

  • Executive Summary — 1-2 pages for leadership
  • Comprehensive Report — full analysis with methodology
  • Slide Presentation — visual PDF in 16:9 format

Output

All artifacts are saved to ./askprisma-outputs/ in your working directory:

  • *.png — charts and visualizations
  • *.csv — intermediate data tables
  • report_[style]_[YYMMDDHHMM].pdf — PDF reports

Python Dependencies

pip install -r requirements.txt

Or individually:

pip install pandas numpy matplotlib seaborn scipy scikit-learn statsmodels openpyxl reportlab Pillow

File Structure

askprisma-skill/
├── bin/
│   └── install.js                   # npx cross-platform installer
├── .claude-plugin/
│   └── plugin.json                  # Claude plugin manifest
├── skills/
│   └── askprisma/
│       ├── SKILL.md                 # Skill instructions (frontmatter + content)
│       ├── references/
│       │   ├── coding-patterns.md   # Python coding conventions
│       │   ├── business-translation.md  # Technical-to-business translation guide
│       │   └── report-styles.md     # PDF report JSON schemas
│       └── scripts/
│           └── generate_report.py   # PDF generator (standalone CLI)
├── gemini-extension.json            # Gemini CLI extension manifest
├── package.json                     # npm package (agents.skills field for npx skills add)
├── LICENSE
├── README.md
├── CHANGELOG.md
└── requirements.txt

Testing the PDF Generator Standalone

python skills/askprisma/scripts/generate_report.py \
  --input /tmp/test_report.json \
  --output /tmp/test_output.pdf \
  --charts-dir /tmp

License

MIT — see LICENSE.

About

Business data analysis skill for Claude Code. Explore data, plan analysis, write Python, synthesize business insights, generate PDF reports.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages