Azimutt is an open source database schema visualizer built for developers and DBAs who need to understand, document, and analyze databases with dozens or hundreds of tables without wading through a generic ERD tool that renders everything at once.
Most ERD tools show the full schema simultaneously. On a 50-table database, the result is an unreadable tangle of lines. dbdiagram.io and DrawSQL work well for small schemas designed from scratch, but they are not built for exploring existing production databases you did not design and cannot fully map from memory.
Azimutt takes a progressive approach: you load your schema, then reveal tables and relationships incrementally, focusing on what matters for the current investigation. The path-finding feature automatically traces how two tables connect through foreign key chains, which saves hours on unfamiliar schemas. Table and column notes let you document findings directly in the diagram layout.
Key Features
- Interactive ERD with progressive reveal: show only the tables relevant to your current investigation
- Automatic path-finding between any two tables through foreign key chains
- AML (Azimutt Markup Language) for designing new schemas at typing speed without a visual canvas
- Database linting to surface inconsistencies like missing foreign keys and nullable columns that should not be null
- Per-table and per-column documentation notes attached directly to diagram layouts
- Support for Postgres, MySQL, SQLite, MongoDB, and other relational and document databases
Self-Hosting
Deploy with Docker. The self-hosted version includes all exploration, path-finding, and documentation features. The repository provides a Docker Compose file for local development and a production configuration.
License
MIT. Azimutt is free to self-host, fork, and use commercially with no restrictions.
Best For
Azimutt is best for developers and DBAs onboarding to an unfamiliar codebase, teams creating living database documentation, and engineers debugging query performance issues on schemas too large to hold in memory at once.
