v3.0

Documentation

Complete guide to DA Toolkit - 6 core + 6 DA commands for Claude Code, from data analysis to reports and dashboards.

Start Here

Getting Started

Install and setup in 2 minutes.

Installation

1

Default Setup (Core Only)

$ npm install -g tramy
$ tramy setup

Creates CLAUDE.md with 6 core commands. Suitable for any project.

Generated:
your-project/
├── CLAUDE.md
└── .claude/commands/
2

DA Setup (Full Toolkit)

Recommended
$ npm install -g tramy
$ tramy setup da

DA toolkit with 6 DA commands only, data directories, and agent templates.

Generated:
your-project/
├── CLAUDE.md
├── data/raw/          # Raw data files
├── data/processed/    # Cleaned data
├── analysis/          # Analysis outputs
├── reports/           # Generated reports
├── notebooks/         # Jupyter notebooks
└── .claude/commands/da/  # 6 DA commands only

Workflow

Core Setup (tramy setup)
/analyze /plan /build /test /doc /commit
DA Setup (tramy setup da)
/da:analyze /da:clean /da:query /da:report

Choose the right setup: Core for general development, DA for data analysis

New in v3.0

Knowledge Management

Commands automatically update CLAUDE.md with learned knowledge.

🧠

Continuous Learning

Each command automatically updates knowledge to CLAUDE.md when completed. This helps Claude "remember" what it learned across sessions.

Knowledge Sections in CLAUDE.md

## Project Knowledge
Discoveries, insights, patterns learned
Updated by: Core: /analyze | DA: /da:analyze
## Data Sources
Data sources, schemas, relationships
Updated by: Core: /analyze, /doc | DA: /da:query, /da:analyze
## Current Plans
Active plans and their status
Updated by: Core: /plan
## Project Files
Important files and their purposes
Updated by: Core: /build | DA: /da:notebook
## Data Quality
Data quality issues and validations
Updated by: Core: /test | DA: /da:clean
## Documentation
Links to generated docs and reports
Updated by: Core: /doc | DA: /da:report, /da:dashboard

How it works

1

Run a command, e.g.: /analyze "sales data"

2

Command executes and delivers results

3

Command auto-updates CLAUDE.md with learned knowledge (data sources, insights, patterns...)

4

Next time you ask, Claude has full context from CLAUDE.md

tramy setup

Core Commands

6 commands for general development. Install with tramy setup

/analyze

Explore and understand data, code, or problems

Deep dive into any topic - explore structure, patterns, and provide insights before taking action. This is the first step in any workflow.

Usage
> /analyze "describe what you want to analyze"
Examples
Input: /analyze "sales_2024.csv"
Output:

Analyze file structure, columns, data types, missing values

Input: /analyze "user authentication flow"
Output:

Analyze current auth flow, security concerns

Output: Analysis report with key findings and recommendations
Updates CLAUDE.md: Data Sources, Project Knowledge
/plan

Create detailed implementation plan

Break complex tasks into specific steps with clear deliverables. Helps you get an overview before starting work.

Usage
> /plan "describe task to implement"
Examples
Input: /plan "quarterly revenue report"
Output:

5-step plan with data sources, metrics, visualizations

Input: /plan "migrate database to PostgreSQL"
Output:

Migration plan with rollback strategy

Output: Step-by-step plan with milestones
Updates CLAUDE.md: Current Plans
/build

Implement solution - write code, queries, notebooks

Execute the plan - write SQL, Python, create notebooks, build dashboards. This is the main implementation step.

Usage
> /build "describe what to build"
Examples
Input: /build "ETL pipeline for user data"
Output:

Python scripts with data validation

Input: /build "REST API for products"
Output:

API endpoints with CRUD operations

Output: Working code, queries, or artifacts
Updates CLAUDE.md: Project Files
/test

Validate results and verify data quality

Check data quality, validate results, test edge cases. Ensure output is correct before delivery.

Usage
> /test "describe what to test"
Examples
Input: /test "check for null values and outliers"
Output:

Data quality report with issues found

Input: /test "validate API responses"
Output:

Test results with pass/fail status

Output: Test results with pass/fail status
Updates CLAUDE.md: Data Quality
/doc

Generate documentation and reports

Create documentation, reports, methodology notes, presentations. Help stakeholders understand your results.

Usage
> /doc "describe what to document"
Examples
Input: /doc "create analysis summary"
Output:

Executive summary with key metrics

Input: /doc "API documentation"
Output:

OpenAPI spec with examples

Output: Markdown docs, reports, or slides
Updates CLAUDE.md: Documentation, Data Sources
/commit

Git commit with proper message

Analyze changes and create descriptive commit message following conventions. Auto-format message.

Usage
> /commit
Examples
Input: /commit
Output:

feat: add user analytics dashboard with KPI metrics

Output: Git commit with formatted message
tramy setup da

DA Commands

6 specialized commands for Data Analysts. Install with tramy setup da

💡

DA Setup = DA Commands Only

tramy setup da only installs 6 DA commands (no core commands). If you need both, run tramy setup first, then tramy setup da.

/da:query

Write SQL queries from natural language

Convert plain English to SQL. Supports complex queries with JOINs, subqueries, window functions, and CTEs.

Usage
> /da:query "describe the data you need"
Examples
Input:

"top 10 customers by revenue last quarter"

Output:

SELECT with ORDER BY, LIMIT, date filtering

Input:

"month-over-month growth rate"

Output:

Window functions with LAG()

Input:

"products that never sold"

Output:

LEFT JOIN with NULL check

Output: Optimized SQL query with explanatory comments
Updates CLAUDE.md: Project Knowledge, Data Sources
/da:analyze

Automatic Exploratory Data Analysis

Auto-detect the type of analysis needed (descriptive, diagnostic, predictive) and perform comprehensive EDA.

Usage
> /da:analyze "describe data or problem to analyze"
Examples
Input:

"analyze sales_data.csv"

Output:

Full EDA with distributions, correlations, outliers

Input:

"compare Q1 vs Q2 performance"

Output:

Comparative analysis with statistical tests

Input:

"find anomalies in transaction data"

Output:

Anomaly detection with IQR/Z-score

Output: Analysis report with visualizations
Updates CLAUDE.md: Project Knowledge, Data Sources
/da:clean

Clean and transform data

Transform raw data to processed data. Handle missing values, duplicates, outliers, and standardize formats.

Usage
> /da:clean "describe data to clean"
Examples
Input:

"remove duplicates and nulls from sales.csv"

Output:

Python script with pandas cleaning

Input:

"standardize date formats in transactions"

Output:

Data transformation script

Input:

"handle outliers in revenue data"

Output:

IQR/Z-score filtering

Output: Cleaned data + transformation script
Updates CLAUDE.md: Project Knowledge, Data Quality
/da:report

Generate professional analysis reports

Create professional reports with executive summary, methodology, findings, and recommendations.

Usage
> /da:report "type of report to create"
Examples
Input:

"weekly sales summary"

Output:

Markdown report with KPIs and trends

Input:

"customer segmentation analysis"

Output:

Full report with methodology and insights

Input:

"A/B test results"

Output:

Statistical report with confidence intervals

Output: Markdown/HTML report with charts
Updates CLAUDE.md: Documentation, Project Knowledge
/da:dashboard

Design BI dashboards

Design dashboard layouts, define metrics, generate code for Tableau, Metabase, or custom solutions.

Usage
> /da:dashboard "describe dashboard to create"
Examples
Input:

"sales performance dashboard"

Output:

Layout + metrics + SQL for each widget

Input:

"real-time monitoring dashboard"

Output:

Time-series charts with alerts

Input:

"customer health scorecard"

Output:

Scorecard with color-coded metrics

Output: Dashboard spec + SQL queries
Updates CLAUDE.md: Documentation, Project Knowledge
/da:notebook

Create Jupyter notebooks for analysis

Generate complete Jupyter notebooks with data loading, cleaning, analysis, and visualization cells.

Usage
> /da:notebook "type of analysis to perform"
Examples
Input:

"cohort analysis notebook"

Output:

Python notebook with retention curves

Input:

"regression analysis"

Output:

Notebook with model training + evaluation

Input:

"data profiling template"

Output:

Reusable profiling notebook

Output: .ipynb file ready to run
Updates CLAUDE.md: Documentation, Project Knowledge
Step-by-Step

Tutorials

Detailed guides for common use cases.

1

Sales Data Analysis

From raw data to executive report (DA Setup)

1
tramy setup da
Setup DA toolkit with 6 DA commands
2
Place file in data/raw/
Copy sales_2024.csv to data/raw/
3
/da:analyze "explore sales_2024.csv"
Explore data structure, columns, patterns
4
/da:clean "remove nulls and duplicates"
Clean data and save to data/processed/
5
/da:query "top 10 products by revenue"
Write SQL to get top products
6
/da:report "Q4 sales performance summary"
Generate summary report
2

Create KPI Dashboard

Design executive dashboard (DA Setup)

1
tramy setup da
Setup DA toolkit
2
/da:dashboard "revenue, users, churn metrics"
Design dashboard structure
3
/da:query "daily active users last 30 days"
SQL for DAU chart
4
/da:query "MRR breakdown by plan"
SQL for revenue breakdown
5
/da:query "churn rate by cohort"
SQL for churn analysis
6
/da:report "dashboard documentation"
Document metric definitions
3

Data Cleaning Pipeline

Raw → Clean → Analysis (DA Setup)

1
tramy setup da
Setup DA toolkit
2
/da:analyze "profile raw data"
Data profiling and quality assessment
3
/da:clean "standardize dates, handle nulls"
Clean and transform data
4
/da:query "validate cleaned data"
SQL to verify data quality
5
/da:notebook "data cleaning documentation"
Create notebook documenting process
4

DA Workflow

Data analysis workflow

1
/da:analyze
Start with EDA - understand data first
2
/da:clean
Clean and transform data
3
/da:query
Write SQL to extract insights
4
/da:report
Create report for stakeholders
5
/da:dashboard
Design dashboard if needed
6
/da:notebook
Document analysis in notebook
CLI

CLI Reference

All available CLI commands.

tramy setup

Default setup with 6 core commands

Creates CLAUDE.md and command templates

tramy setup da

DA toolkit setup

6 DA commands only + data directories

tramy setup -y

Quick setup without prompts

Uses defaults, skip interactive mode

tramy list

List roles and commands

Shows available roles and command counts

tramy context

View project context

Shows CLAUDE.md content

tramy context update

Update context

Re-scan tech stack and regenerate CLAUDE.md

tramy doctor

Health check

Verify installation and diagnose issues