Skip to main content
Install the Lightdash skills to enable AI-powered development with Claude Code or Cursor. This teaches Claude everything it needs to know about building in Lightdash, from writing metrics to creating dashboards.

ā€‹
Quick install

lightdash install-skills
This command installs skills for Claude by default. For other AI coding agents, see the detailed installation guide.

ā€‹
What you can build

With Lightdash skills installed, you can work conversationally to build complete BI solutions. Claude can help you:
  • Build from scratch. Create complete projects with no existing dbt files. Just describe the report you want, and Claude will profile your warehouse, suggest metrics with descriptions and categories, and build dashboards with layouts, colors, tabs, and filters.
  • Warehouse profiling. Query your data warehouse to discover available tables and suggest relevant metrics for your use case.
  • Bulk operations. Make updates to 50+ charts in production, rename tables or metrics, and bulk update all dependent content.
  • Migrations. Move analysis from other tools to Lightdash by translating existing logic into Lightdash metrics.
  • Best practices. Review dashboards and charts for improvements, suggest better metric definitions, and refactor code.
  • Rapid development. Work conversationally to iterate on metrics, dimensions, and dashboards without context switching.
Perfect for teams without an existing semantic layer. Just connect your warehouse, describe the report you want, and Claude works backwards to build the models.

ā€‹
How it works

Claude needs verification to work well, and the Lightdash CLI provides everything needed for success. You describe what you want, and Claude:
  1. Profiles the warehouse to understand available data
  2. Suggests and creates metrics with proper metadata
  3. Builds dashboards with thoughtful layouts and configurations
  4. Verifies each step before executing

ā€‹
Example workflows

Build a dashboard from scratch: 
I want to build a customer retention dashboard. Profile my BigQuery warehouse
to find relevant tables, suggest key metrics like churn rate and retention
cohorts, and create a dashboard with tabs for monthly and quarterly views.
Create metrics with metadata: 
Create revenue metrics for our orders table with descriptions and categories
for the metrics catalog. Include total revenue, average order value, and
revenue per customer.
Bulk refactoring: 
Rename the "user" table to "customer" across all metrics, dimensions, and
dashboards. Update all references and maintain the relationships.

ā€‹
Next steps

For detailed installation options (global install, custom paths, different AI agents), see the Agent skills guide. Once installed, you can use Claude Code or Cursor to build your Lightdash project conversationally without opening the UI.
When using Claude, prefix your prompts with /developing-in-lightdash to ensure Claude uses the Lightdash skills effectively.