data — Claude Code plugin: what it does, how to install
A plugin for Claude Code that enables advanced data analysis and SQL workflows. It turns your agent into a data collaborator capable of exploring datasets, generating insights, and building dashboards.
Apache-2.0 ★ 36,654 updated 2026-10-24
/plugin marketplace add anthropics/knowledge-work-plugins
/plugin install data@knowledge-work-pluginsWhat data does
The plugin integrates data analysis tools into Claude Code to help you query datasets, profile tables, and validate findings. It supports various SQL dialects and analytical stacks. If a direct data warehouse connection is not available, you can upload CSV or Excel files or paste raw query results for the agent to process.
For best performance, configure MCP servers for your data infrastructure, such as Snowflake, Databricks, or BigQuery, in your .mcp.json file. Once connected, the agent explores schemas, runs queries, and iterates based on results. The plugin also builds interactive HTML dashboards and performs quality assurance checks on your methodology before you share results with stakeholders.
What it helps with
- profile a dataset for quality issues
- write optimized SQL queries
- build an interactive HTML dashboard
- validate analysis methodology for bias
- visualize product revenue trends
Example prompts for the agent
- /analyze What is the revenue trend for product A over the last six months?
- /explore-data
users.csv - /build-dashboard Create a sales dashboard using the provided data
How to install data
Plugin marketplace
/plugin marketplace add anthropics/knowledge-work-plugins/plugin install data@knowledge-work-pluginsStep by step, with commands for each platform: Claude Code plugins: marketplaces, installing, and building your own
Questions about data
How to install data in Claude Code?
Run /plugin marketplace add anthropics/knowledge-work-plugins followed by /plugin install data@knowledge-work-plugins in your console.
What does data do?
It enables Claude to act as a data analyst, covering SQL generation, data exploration, Python-based visualization, and interactive dashboard construction.
What is required to use data?
To query your infrastructure directly, you must configure an MCP server for your specific data warehouse or analytics stack.
Source text
data · Apache-2.0
# Data Analyst Plugin
A data analyst plugin primarily designed for [Cowork](https://claude.com/product/cowork), Anthropic's agentic desktop application — though it also works in Claude Code. SQL queries, data exploration, visualization, dashboards, and insight generation. Works with any data warehouse, any SQL dialect, and any analytics stack.
## Installation
```
claude plugin marketplace add anthropics/knowledge-work-plugins
claude plugin install data@knowledge-work-plugins
```
## What It Does
This plugin transforms Claude into a data analyst collaborator. It helps you explore datasets, write optimized SQL, build visualizations, create interactive dashboards, and validate analyses before sharing with stakeholders.
### With a Data Warehouse Connection
Connect your data warehouse MCP server (e.g., Snowflake, Databricks, BigQuery, or any SQL-compatible database) for the best experience. Claude will:
- Query your data warehouse directly
- Explore schemas and table metadata
- Run analyses end-to-end without copy-pasting
- Iterate on queries based on results
### Without a Data Warehouse Connection
Without a data warehouse connection, paste SQL results or upload CSV/Excel files for analysis and visualization. Claude can also write SQL queries for you to run manually, and then analyze the results you provide.
## Commands
| Command | Description |
|---------|-------------|
| `/analyze` | Answer data questions -- from quick lookups to full analyses |
| `/explore-data` | Profile and explore a dataset to understand its shape, quality, and patterns |
| `/write-query` | Write optimized SQL for your dialect with best practices |
| `/create-viz` | Create publication-quality visualizations with Python |
| `/build-dashboard` | Build interactive HTML dashboards with filters and charts |
| `/validate` | QA an analysis before sharing -- methodology, accuracy, and bias checks |
## Skills
| Skill | Description |
|-------|-------------|
| `sql-queries` | SQL best practices across dialects, common patterns, and performance optimization |
| `data-exploration` | Data profiling, quality assessment, and pattern discovery |
| `data-visualization` | Chart selection, Python viz code patterns, and design principles |
| `statistical-analysis` | Descriptive stats, trend analysis, outlier detection, and hypothesis testing |
| `data-validation` | Pre-delivery QA, sanity checks, and documentation standards |
| `interactive-dashboard-builder` | HTML/JS dashboard construction with Chart.js, filters, and styling |
## Example Workflows
### Ad-Hoc Analysis
```
You: /analyze What was our monthly revenue trend for the past 12 months, broken down by product line?
Claude: [Writes SQL query] → [Executes against data warehouse] → [Generates trend chart]
→ [Identifies key patterns: "Product line A grew 23% YoY while B was flat"]
→ [Validates results with sanity checks]
```
### Data Exploration
```
You: /explore-data users table
Claude: [Profiles table: 2.3M rows, 47 columns]
→ [Reports: created_at has 0.2% nulls, email has 99.8% cardinality]
→ [Flags: status column has unexpected value "UNKNOWN" in 340 rows]
→ [Suggests: "High-value dimensions to explore: plan_type, signup_source, country"]
```
### Query Writing
```
You: /write-query I need a cohort retention analysis -- users grouped by signup month,
showing what % are still active 1, 3, 6, and 12 months later. We use Snowflake.
Claude: [Writes optimized Snowflake SQL with CTEs]
→ [Adds comments explaining each step]
→ [Includes performance notes about partition pruning]
```
### Dashboard Building
```
You: /build-dashboard Create a sales dashboard with monthly revenue, top products,
and regional breakdown. Here's the data: [pastes CSV]
Claude: [Generates self-contained HTML file]
→ [Includes interactive Chart.js visualizations]
→ [Adds dropdown filters for region and time period]
→ [Opens in browser for review]
```
### Pre-Share Validation
```
You: /validate [shares analysis document]
Claude: [Reviews methodology] → [Checks for survivorship bias in churn analysis]
→ [Verifies aggregation logic] → [Flags: "Denominator excludes trial users
which could overstate conversion rate by ~5pp"]
→ [Confidence: "Ready to share with noted caveat"]
```
## Connecting Your Data Stack
> If you see unfamiliar placeholders or need to check which tools are connected, see [CONNECTORS.md](CONNECTORS.md).
This plugin works best when connected to your data infrastructure. Add MCP servers for:
- **Data Warehouse**: Snowflake, Databricks, BigQuery, Definite, or any SQL-compatible database
- **Analytics/BI**: Amplitude, Looker, Tableau, or similar
- **Notebooks**: Jupyter, Hex, or similar
- **Spreadsheets**: Google Sheets, Excel
- **Data Orchestration**: Airflow, dbt, Dagster, Prefect
- **Data Ingestion**: Fivetran, Airbyte, Stitch
Configure MCP servers in your `.mcp.json` or Claude Code settings to enable direct data access.
© Anthropic · Apache-2.0 · anthropics/knowledge-work-plugins