GitHub - azhakhan/dribble: AI-powered, open source IDE for Databases

An AI-powered, open-source SQL IDE for your databases.

Dribble is a web-based SQL IDE with a built-in AI data analyst. Connect to a Postgres database, browse its schema, run queries in a notebook, explore tables with sort/filter/pagination, and ask an AI agent questions about your data — all in one tabbed workspace that remembers where you left off.

Features

• AI data analyst — chat with an agent (Claude Opus 4.8) that inspects your schema, writes and runs read-only SQL, iterates on errors, and renders the final result set as a table.
• SQL notebooks — write and execute queries in a Monaco editor with syntax highlighting. Run with Cmd/Ctrl + Enter. Notebooks and their results are saved.
• Schema browser — navigate schemas and tables from a collapsible sidebar tree.
• Table explorer — browse table data with server-side pagination, column sorting, and a raw WHERE-clause filter.
• Fast results grid — large result sets render in a virtualized data grid.
• Persistent workspace — open tabs, layout/panel sizes, the expanded tree, and cached query/chat results survive reloads (and follow you across browsers, since state is stored server-side).
• Smart connection lifecycle — database drivers are kept warm while in use and idle out when not, with the sidebar reflecting live connection status.
• Flexible auth — runs with no login at all for local use, or behind Google sign-in (with an email/domain allowlist) for multi-user deployments, where each person's connections, notebooks, and chats are private. Stored database credentials are encrypted at rest. See docs/authentication.md.
• Pluggable drivers — Postgres ships today; the driver registry is built to add more engines (MySQL, Snowflake, …).

Tech stack

Next.js 16 · React 19 · TypeScript · Tailwind CSS 4 · Monaco Editor · glide-data-grid · Zustand · Vercel AI SDK (@ai-sdk/anthropic) · Postgres (pg)

Getting started

Prerequisites

• Node.js 20+
• A Postgres database for storing app metadata (connections, notebooks, chat history). Any Postgres works — local, Neon, Supabase, Vercel Postgres, etc.
• An Anthropic API key for the AI agent.

Install

git clone <your-repo-url> dribble
cd dribble
npm install

Configure

Copy the example env file and fill in the values:

cp .env.example .env.local

# Metadata storage (connections, notebooks, chat history).
# Any Postgres works — Vercel Postgres / Neon / Supabase / local.
DATABASE_URL=postgres://user:pass@host:5432/dribble

# Secret used to encrypt stored DB credentials (and sign the auth session).
# Required. Generate with: openssl rand -hex 32
APP_SECRET=

# Powers the AI chat agent (claude-opus-4-8).
ANTHROPIC_API_KEY=

That's all you need to run locally — with no auth configured, the app starts without a login screen and all data belongs to a single built-in user.

The required metadata tables are created automatically on first run.

Optional: Google sign-in (multi-user)

To require login and keep each user's data private, configure Google OAuth — see docs/authentication.md for the full setup. In short, add to .env.local:

AUTH_GOOGLE_ID=
AUTH_GOOGLE_SECRET=
# Restrict who may sign in (leave empty to allow any Google account):
AUTH_ALLOWED_EMAILS=you@example.com
AUTH_ALLOWED_DOMAIN=example.com

Register <origin>/api/auth/callback/google as an authorized redirect URI on the Google OAuth client. Setting these enables the login screen automatically.

Run

Open http://localhost:3000 (you're in directly when no auth is configured; otherwise sign in with Google), add a database connection, and start querying.

To build and run a production server:

A note on AI-generated code

This project was written largely with the help of AI coding tools (Claude Code). All code has been reviewed before being committed, but you should review it yourself before relying on it in production.

License

Released under the MIT License.