Tabularis

Tabularis is a desktop SQL workspace built for teams that use both human developers and AI agents. It supports PostgreSQL, MySQL/MariaDB, and SQLite natively, with additional databases like DuckDB, ClickHouse, Redis, and Firestore available through a plugin system. If you're looking for a DBeaver alternative or something closer to TablePlus but with AI-agent integration baked in, this is worth a look.

The standout feature is its built-in Model Context Protocol (MCP) server. AI tools like Claude, Cursor, and Devin can inspect your database schema and run queries directly through Tabularis, using the same connection profiles you've already set up. No separate scripts, no credentials pasted into chat windows. The agent works through the app.

For human workflows, it holds its own:

• Monaco SQL editor with multi-tab support, syntax highlighting, and run-selection controls
• SQL Notebooks combining SQL and Markdown in multi-cell workflows, with inline charts, shared variables, and parameters
• Visual query builder for assembling joins, filters, and aggregations before dropping to raw SQL
• Visual EXPLAIN that turns execution plans into interactive graphs with optional AI analysis
• ER diagrams and inline schema editing for structural changes
• Data grid for browsing and editing result sets, with CSV and JSON export
• SSH and Kubernetes tunneling with system keychain storage for secrets
• Split view for comparing datasets across multiple connections side by side

The plugin system is genuinely open. Plugins are standalone executables that communicate over JSON-RPC 2.0, so you can write a driver in Rust, Go, Python, or anything else that speaks JSON over stdin/stdout. Plugins install without restarting the app and run in isolated processes. A community bounty market tracks which drivers are funded, in progress, or already shipped.

AI assistance (currently experimental) supports OpenAI, Anthropic, OpenRouter, and Ollama for local models. Built on a Rust and Tauri backend with a React 19 frontend, licensed Apache 2.0. Comparable tools like Beekeeper Studio and Chat2DB cover some of this ground, but the MCP-native design is what sets Tabularis apart for agent-assisted database work.