// the find
TabularisDB/tabularis
Open-source desktop SQL workspace for PostgreSQL, MySQL/MariaDB, SQLite and 12+ more databases like DuckDB, ClickHouse, Redis and Firestore. Built-in MCP server for Claude, Cursor and Devin, SQL notebooks and visual EXPLAIN.
Tabularis is a Tauri-based desktop SQL client (React + Rust) with built-in MCP server support, SQL notebooks, and a JSON-RPC plugin system for adding database drivers in any language. It targets developers who want a free, open-source alternative to TablePlus or Beekeeper Studio and want their AI coding assistant to query their databases directly. At 3.4k stars and v0.13, it's past proof-of-concept but not yet mature.
The JSON-RPC plugin architecture is genuinely clever — drivers are just executables speaking over stdin/stdout, so you can write one in Go, Python, or Rust without touching the core app. SQL notebooks with cross-cell variable references (`{{cellName.column}}`) fill a real gap that DBeaver and TablePlus both ignore. The MCP server integration is practical: one CLI flag exposes schema introspection and query execution to Claude or Cursor without any separate service to run. Tauri v2 with SQLx on the backend means the binary is small and the Postgres/MySQL drivers are doing real async work, not wrapping a JS library.
The database coverage table is misleading — many listed databases are 'claimed' or 'scoped', meaning someone said they'd build a driver but hasn't. SQL Server, MongoDB, Oracle, and Snowflake are all in that category, which matters if you're evaluating this for a team. SQLite support is on the roadmap as needing improvement, which is awkward for a tool that ships SQLite as a built-in driver. No SQL formatter yet (it's on the roadmap), which is a daily-use gap. The project's own origin story admits it was AI-assisted development, and the codebase shows it in places — the `.agents/` and `.claude/` skill directories in the repo root are training scaffolding that got committed alongside production code.