How lightdash/lightdash works
Project type and purpose
• Lightdash is a full-stack, open-source BI platform (an alternative to Looker).
• The monorepo contains:
• packages/backend – a Node.js/TypeScript Express API (TSOA-generated REST)
• packages/frontend – a React/Vite single-page app
• packages/cli – a Node.js CLI to scaffold and manage projects
• packages/common – shared TypeScript types and compiler logic
• packages/warehouses – adapters for external data warehouses (e.g. BigQuery, Snowflake)
• packages/e2e – Cypress end-to-end tests
• It connects to your dbt project, fetches metadata, compiles queries, and serves interactive dashboards, plus a scheduler for reports.
File structure
• Top-level “packages” folder holds each component; each has its own src, tests, configs, package.json.
• Backend is organized by routers → controllers → services → models → database (entities/migrations) → clients → analytics → scheduler.
• Frontend is organized by features, components, hooks, providers, plus an “ee” folder for enterprise-only features (AI copilot, embed, SCIM).
• There are Dockerfiles, docker-compose setups, CI/CD pipelines, Helm charts, and Kubernetes configs for deployment.
• The scheduler submodule uses in-process schedulers to run deliveries (email/Slack) and background jobs.
• External integrations: OAuth/OIDC/SSO, S3 cache, email templates, Slack, Google Drive, Prometheus, Sentry.
README for additional insights
• Confirms core use cases: connect dbt, define metrics in YAML, build/explore dashboards, schedule delivery, share URLs.
• Describes CI/CD validation, self-hosting guide, 1-click deploy, and contributions.