VibeDB is a fully local, open-source, and privacy-focused text-to-SQL tool designed to help developers extract valuable insights from their PostgreSQL databases.
This project is my personal take on combining LLM tool calling, structured responses, and modern AI integrations to create a smooth and intuitive workflow.
- Self-hosted via Docker
- Multi-user profiles
- Organize your work in workspaces
- Automatic SQL generation from natural-language prompts (only
SELECT) - Automatic database table context generation with optional manual management
- Chat and SQL history
- Data visualization with auto-generated ShadcnUI charts
- Fully local, open source and powered by
Ollama
- Next.js 15
- TypeScript
- Tailwind CSS
- ShadcnUI
- Docker Desktop installed and running
- A configured
.envfile with the necessary environment variables (see.env.examplefor guidance)
-
Clone the repository
git clone <repo-url> cd vibedb
-
Configure ENV
- see
.example.envfor example configuration
- see
-
Build docker containers
docker compose up --build -d
- For testing use some spare database, I suggest Pagila
Licenced under MIT licence
Contributions are welcome and appreciated!
Whether it’s fixing a bug, adding a new feature, or improving the documentation, your help makes VibeDB better.