Launched June 9, 2026
Tabularis

Tabularis

Open-source desktop SQL client — notebooks, visual EXPLAIN, MCP server

FreeMCP3,475 impressions#14 of its week

Tabularis is an open-source desktop database client built in Rust and Tauri: the Windows installer is about 12 MB, not 200. MySQL, PostgreSQL and SQLite are built in. Everything else comes from a registry of 18 plugin drivers — MongoDB, Redis, ClickHouse, DuckDB, Elasticsearch, DynamoDB, Google Sheets and more — each one an ordinary process speaking JSON-RPC over stdin/stdout, so a driver can be written in any language. On top of the data grid and SQL editor: notebooks for multi-query analysis, visual EXPLAIN that turns query plans into diagrams with per-node findings, SSH tunnels, Kubernetes port-forwarding, stored routines and ER diagrams. It also ships an MCP server, so Claude, Cursor and other agents can query your databases — with a read-only mode and per-operation approval gates. Credentials live in the OS keychain, connection exports are encrypted, and there is no account: everything stays on your machine. Apache-2.0, for Linux, macOS and Windows.

Tabularis is an open-source desktop SQL client with notebooks, visual EXPLAIN, and a built-in MCP server for AI agents.

for
Developers and data analysts who need a local, secure, extensible database client.
pricing
free
license
Apache-2.0
TabularisDB/tabularis 5.1k 332TypeScriptupdated 7 days ago
works withRustTauriReact 19OpenAIAnthropicOpenRouterOllamaMySQLPostgreSQLSQLiteSSHKubernetes

Key features

6 features of Tabularis
  • Multi-Database Support — Connect to PostgreSQL, MySQL/MariaDB, SQLite and 18+ plugin drivers via JSON-RPC.
  • AI Assistance — Draft, explain and iterate SQL using OpenAI, Anthropic, OpenRouter or local Ollama models.
  • Built-in MCP Server — Expose schemas and run queries to Claude, Cursor and other agents with read-only mode and approval gates.
  • Visual EXPLAIN — Turn execution plans into interactive diagrams with node-level findings and optional AI analysis.
  • SQL Notebooks — Combine SQL and Markdown in multi-cell workflows with inline charts and reusable variables.
  • SSH & Kubernetes Tunneling — Securely reach remote databases; secrets are stored in the OS keychain.

Use cases

  • Analyze multi-query workflows in notebooks with inline charts
  • Visually build complex joins and filters before generating SQL
  • Allow AI agents to safely query production databases via MCP
  • Debug performance issues using interactive Visual EXPLAIN diagrams
  • Manage connections to diverse databases through a plugin registry

Tabularis vs alternatives

TabularisUnified MCP ServerMonghoulgridctlBrowserGateway
Best forMCP-native desktop SQL clientStandalone MCP serverMongoDB-only GUIYAML-driven MCP orchestrationCloud browser routing
PricingFreeSubscriptionOne time feeFreeFree
DevHunt upvotes229112022
LaunchedJun 2026Jan 2026May 2026Aug 2026Sep 2026
  • Tabularis vs Unified MCP Server: Unified MCP Server provides a central MCP endpoint but no desktop UI or multi-database client features.
  • Tabularis vs Monghoul: Monghoul is a MongoDB-specific GUI, whereas Tabularis supports many databases and plugins.
  • Tabularis vs gridctl: gridctl bundles multiple MCP servers and skills, lacking the local SQL editor and notebook environment of Tabularis.
  • Tabularis vs BrowserGateway: BrowserGateway routes cloud browsers; Tabularis focuses on local database interaction and AI-native tooling.

Tabularis FAQ

Is Tabularis free to use?+

Yes, Tabularis is free and open-source under the Apache-2.0 license.

What databases are supported out of the box?+

PostgreSQL, MySQL/MariaDB and SQLite are built in; additional databases are available via plugins.

How does the MCP server work with AI agents?+

The built-in Model Context Protocol server lets compatible agents inspect schemas and run queries through Tabularis, with optional read-only mode and per-operation approval.

Can I write my own database driver?+

Yes, drivers are standalone executables that communicate over JSON-RPC, and can be written in any language such as Rust, Go, Python or Node.js.

How are credentials stored?+

Credentials are kept in the operating system’s keychain and connection exports are encrypted; no external account is required.

Summarized by DevHunt from tabularis.dev · Sep 28, 2026. Details may change; check the official site.