Tabularis is a desktop SQL workspace for Windows, macOS, and Linux that ships PostgreSQL, MySQL/MariaDB, and SQLite drivers built in and reaches 21 further databases (DuckDB, ClickHouse, Redis, Firestore, MongoDB, Oracle, SQL Server, DynamoDB, Elasticsearch, IBM Db2, Cloudflare D1, and others) through plugins. Plugins are standalone executables that talk to the app over JSON-RPC 2.0 on stdin and stdout, so a driver can be written in any language, installed from the registry without a restart, and uninstalled with one click. Databases not yet covered sit on a public bounty board.
The workspace centers on a Monaco-based SQL editor with tabs, isolated connections per tab, split view, multi-statement execution with per-query result tabs, and query splitting that handles stored procedures and dollar-quoted blocks. Around it sit a schema explorer with an interactive ER diagram, a drag-and-drop visual query builder, a visual EXPLAIN that renders execution plans as navigable node graphs for PostgreSQL, MySQL/MariaDB, and SQLite, and an editable data grid with batch commits, JSON and JSONB cell editors, and CSV or JSON export. SQL notebooks combine SQL and Markdown cells in one document with inline results, bar, line, and pie charts, cross-cell variables that expand to a CTE at run time, and export to HTML, CSV, or JSON.
For agent builders the relevant piece is the built-in MCP (Model Context Protocol) server, which exposes five tools (list connections, list databases, list tables, describe a table, and run a query) so an AI agent can read schema and execute SQL against saved connections from its own chat interface. Settings offer one-click client configuration for Claude Desktop, Cursor, and Windsurf, and each tool call can return JSON or the more compact TOON format. Optional text-to-SQL and query explanation run against OpenAI, Anthropic, MiniMax, OpenRouter, Ollama for local models, or any OpenAI-compatible API, with model lists fetched from the provider and cached locally.
Connection profiles, saved queries, settings, and themes live in a local configuration folder that can be relocated to a synced drive, with passwords held in the system keychain and SSH tunneling for remote hosts. The frontend is React 19 with TypeScript and Tailwind CSS v4, and the backend is Rust on Tauri v2 with SQLx.
Features
- Database coverage: PostgreSQL, MySQL/MariaDB, and SQLite built in, plus 21 plugin drivers including DuckDB, ClickHouse, Redis, Firestore, MongoDB, Oracle, SQL Server, DynamoDB, Elasticsearch, IBM Db2, Cloudflare D1, LibSQL/Turso, CSV folders, and Google Sheets
- Plugin system: standalone driver executables speaking JSON-RPC 2.0 over stdin and stdout in any language, installable from the registry without a restart, with a bounty board for unclaimed databases
- MCP server: a built-in Model Context Protocol server with list_connections, list_databases, list_tables, describe_table, and run_query tools, one-click client setup for Claude Desktop, Cursor, and Windsurf, and JSON or TOON output per call
- AI assist (optional): text-to-SQL and query explanation through OpenAI, Anthropic, MiniMax, OpenRouter, Ollama local models, or any OpenAI-compatible API, with provider-fetched model lists and per-provider custom models
- SQL editor: Monaco editing with completion, tabbed and split views with isolated connections, run-all or run-selected execution with per-query result tabs, saved queries, and .sql file editing
- SQL notebooks: SQL and Markdown cells in one document with inline results and charts, cross-cell result references expanded to CTEs, global parameters, run-all with stop-on-error, and HTML, CSV, or JSON export
- Visual EXPLAIN: execution plans as interactive node graphs with table, raw, and AI-assisted views for PostgreSQL, MySQL/MariaDB, and SQLite
- Visual query builder: drag-and-drop tables and joins with WHERE and HAVING filters, aggregates, sorting, and limits, generating SQL as you build
- Database explorer: tree view of tables, columns, keys, indexes, views, and routines with inline editing, an interactive ER diagram, and SQL dump and import
- Data grid: inline and batch editing, row creation and multi-select, CSV or JSON export, syntax-highlighted JSON and JSONB cells with a dedicated editor, and initial MySQL GEOMETRY support
- Connections: saved and cloneable profiles with keychain password storage, SSH tunneling, multi-database selection per connection, and per-connection icons and colors
- Customization: remappable keyboard shortcuts, declarative theme packages with VS Code imports, and a UI translated into eleven languages
