Drizby

www.drizby.com

Drizby is an open-source BI app built around drizzle-cube. It lets you connect a database, define or import Drizzle schemas, build cube definitions, and explore the results through dashboards, notebooks, and a visual analysis builder.

Status: active work in progress. The README below reflects what is in the repo today.

What Drizby does today

• Runs as a self-hosted web app with an onboarding flow and demo seed data
• Manages database connections and compiles a semantic layer per connection
• Lets admins author Drizzle schema files and cube definitions in the browser
• Can run drizzle-kit pull from the app to bootstrap schema files from an existing database
• Includes AI-assisted cube planning/generation when an AI provider is configured
• Provides dashboards, notebook-style analysis, and a no-code analysis builder
• Supports role-based app access (admin, member, user/pending approval)

Important scope notes

• Single-tenant today: the codebase stores organisationId on records, but the app currently runs with that value hardcoded to 1. Treat the current product as a single workspace deployment, not full multi-tenant SaaS.
• Schema pull has limits: Drizby shells out to drizzle-kit pull, then cleans up the generated file for use in the editor. For PostgreSQL-style dialects it currently targets the public schema, and complex generated extras such as indexes/constraints may need manual cleanup after import.
• AI features are optional: notebooks still exist without AI configured. AI-assisted cube generation uses the provider configured in Settings, while notebook chat can also use a per-user API key entered in the notebook UI.

Quick start with Docker

Create a persistent volume, generate two random secrets, then run the published image with the required production env vars:

docker volume create drizby-data
openssl rand -hex 32   # use one value for OAUTH_JWT_SECRET
openssl rand -hex 32   # use another value for ENCRYPTION_SECRET

docker run --rm \
  -p 3461:3461 \
  -e APP_URL=http://localhost:3461 \
  -e OAUTH_JWT_SECRET=replace-with-generated-hex-value \
  -e ENCRYPTION_SECRET=replace-with-generated-hex-value \
  -v drizby-data:/app/data \
  ghcr.io/cliftonc/drizby:main

Then open http://localhost:3461.

On first run you will create the initial admin account in the setup wizard, then Drizby seeds a demo SQLite connection with sample data/content so you can explore the product immediately.

Docker notes

• External app port: 3461
• Internal Node port inside the container: 3462 (used behind Caddy; you do not need to publish it)
• Persistent app data path: /app/data
• The named volume stores the internal app database and local file-based data used by the container.
• If you publish a different host port, set APP_URL to the public URL users will actually open.
  • Example: -p 8080:3461 => APP_URL=http://localhost:8080
• To reset a local Docker install completely:

docker rm -f $(docker ps -aq --filter ancestor=ghcr.io/cliftonc/drizby:main) 2>/dev/null || true
docker volume rm drizby-data

Optional Docker env vars

These are not required for a basic local run, but are supported by the app:

• ADMIN_EMAIL, RESEND_API_KEY, RESEND_FROM_EMAIL to auto-create the admin account and send a setup/reset email
• APP_NAME to customize instance branding

Build the Docker image yourself

git clone https://github.com/cliftonc/drizby.git
cd drizby
docker build -t drizby .
openssl rand -hex 32   # use one value for OAUTH_JWT_SECRET
openssl rand -hex 32   # use another value for ENCRYPTION_SECRET
docker run --rm \
  -p 3461:3461 \
  -e APP_URL=http://localhost:3461 \
  -e OAUTH_JWT_SECRET=replace-with-generated-hex-value \
  -e ENCRYPTION_SECRET=replace-with-generated-hex-value \
  -v drizby-data:/app/data \
  drizby

Source development

Prerequisites

• Node.js 24+ recommended (matches the Dockerfile/runtime used in this repo)
• npm

Setup

git clone https://github.com/cliftonc/drizby.git
cd drizby
CI=true npm ci --legacy-peer-deps --no-audit --no-fund
npm run setup

npm run setup generates migrations, applies them, and seeds the local demo content.

Run the dev servers

npm run dev

That starts:

• Vite client: http://localhost:3460
• Backend/API: http://localhost:3461
• Vite proxies /api, /cubejs-api, /oauth, /.well-known, /mcp, and /health to the backend

In normal source development, open http://localhost:3460.

Build and run a local production build

npm run build
npm start

The standalone server listens on PORT (default 3461). In Docker, Caddy listens on PORT and proxies to Node on NODE_PORT (default 3462).

Current capabilities

Semantic layer and modeling

• Browser-based schema editor with Monaco
• Browser-based cube definition editor with compilation/validation
• Per-connection semantic layer compilation
• Schema import via drizzle-kit pull
• AI-assisted cube planning, generation, and join application

Analysis surfaces

• Dashboards with draggable/resizable grid widgets
• Notebook-style analysis documents with mixed content blocks
• No-code analysis builder for measures, dimensions, and filters
• Chart/table rendering for saved and exploratory analysis

Connections

The connection registry in the app currently includes presets for:

• PostgreSQL variants: postgres.js, pg, Neon, Supabase, PGlite, AWS Data API, Aurora DSQL
• MySQL variants: mysql2, PlanetScale, TiDB
• SingleStore
• SQLite variants: better-sqlite3, LibSQL/Turso
• Databend
• Snowflake
• DuckDB

Each saved connection gets its own schema files, cube definitions, and semantic-layer instance inside Drizby.

Auth and administration

• Email/password auth
• Role-based permissions (admin, member, pending user)
• Admin approval flow
• Settings UI for AI providers and auth providers
• OAuth/SSO-related routes and settings for Google, GitHub, GitLab, Microsoft, Slack, magic links, and SAML/SCIM configuration

Typical workflow

1. Start Drizby and create the admin account
2. Add a database connection
3. Pull or write Drizzle schema files
4. Create cube definitions manually or with AI assistance
5. Explore data in the analysis builder or notebooks
6. Save/publish results to dashboards

Verification commands

The repo includes these main checks:

npm test
npm run typecheck
npm run lint

Tech stack

Layer	Technology
Semantic layer	drizzle-cube
Backend	Hono, TypeScript, Drizzle ORM
Frontend	React 18, TanStack Query, Recharts, Tailwind CSS
Editor	Monaco
Internal app DB	SQLite (better-sqlite3)
Build tooling	Vite, esbuild/tsx

License

MIT

---

This repository is maintained with the assistance of Last Light, an automated GitHub maintenance bot that handles issue triage, pull request reviews, and routine repository upkeep.