Open-source desktop SQL workspace. Tabularis is an open-source desktop SQL workspace from TabularisDB for PostgreSQL, MySQL, and SQLite, extended to engines such as DuckDB, ClickHouse, and Redis through plugins. It combines SQL notebooks, a visual query builder, and interactive visual EXPLAIN with a built-in MCP server that lets AI agents read schemas and run queries inside the same app.[1][2]
Quick verdict
A focused desktop SQL workspace. Tabularis is a free, Apache-2.0 desktop SQL workspace that covers a few engines well instead of chasing driver counts, and pairs notebooks and visual EXPLAIN with a built-in MCP server for AI agents. Its vendor comparison concedes breadth to DBeaver, and enterprise controls are undocumented. It fits developers and agent-assisted teams on PostgreSQL, MySQL, or SQLite who want a local-first client.[1][2][3]
What Tabularis can do
Three built-in engines, 21 shipped plugins. PostgreSQL, MySQL/MariaDB, and SQLite ship built in, and 21 official plugins cover engines including DuckDB, ClickHouse, Redis, Firestore, SQL Server, MongoDB, and Oracle. The built-in PostgreSQL driver is deprecated in favour of the PostgreSQL plugin, which Tabularis installs automatically.[1]
SQL notebooks with charts and export. Notebooks mix SQL and Markdown cells with cross-cell variables and inline charts, auto-save as .tabularis-notebook files, and export to HTML, CSV, or JSON.[1]
Visual EXPLAIN as plan graphs. Execution plans render as interactive graphs with per-driver EXPLAIN formats for PostgreSQL, MySQL/MariaDB, and SQLite, and Analyse defaults off for data-modifying statements.[1]
Built-in MCP server for AI agents. The MCP server lets Claude, Cursor, and Devin read the schema and execute queries from chat, with one-click client configuration and JSON or TOON output formats.[1]
Plugins in any language over JSON-RPC. Drivers are standalone executables communicating over JSON-RPC 2.0 via stdin/stdout, installed from the plugin registry without a restart and managed under Installed Drivers.[1]
Local AI text-to-SQL via Ollama. The comparison table documents AI text-to-SQL backed by local Ollama models, in contrast to the cloud-based assistants of the compared tools.[1]
Who it fits
Desktop-first SQL daily drivers. Developers who live in PostgreSQL, MySQL, or SQLite get an installer-based client with notebooks and plan graphs, without provisioning any server.[1][2]
Teams wiring agents to databases. Shops that want Claude, Cursor, or Devin to inspect schemas and run approved queries can reuse the same workspace developers already open, rather than standing up a separate agent gateway.[1][2]
Tinkerers extending engine coverage. Contributors comfortable writing a small executable can add a missing engine in any language through the documented plugin protocol and registry.[1]
Where it falls short
Breadth lags the hundred-driver tools. The vendor's own note says that if you need dozens of drivers, use DBeaver: Tabularis focuses on doing a few databases well, with the rest depending on community plugins.[1]
No managed cloud or team story. Distribution is desktop installers only, with local configuration storage and per-machine plugin installs, so teams wanting hosted access or synced workspaces will not find that story documented.[1][2]
Enterprise controls are undocumented. No SSO, audit, or support SLA documentation was found, so regulated or centrally managed rollouts should treat those as unknowns, not as included features.[1][2]
Community signals
Inspected sample: the TabularisDB organisation discussions index, with active General, Announcements, Ideas, Q&A, Show and tell, and Polls categories and threads from March through October 2026, including maintainer announcements and user questions.
1 cited community source. No Reddit originals or G2 reports about Tabularis were inspected, so no testimonial-grade community quotes are included and the testimonials array is empty. Discussion threads were observed at index level only; no individual report is quoted here.
Discord and contribution entry points. The site links a Discord server and a contribute page pointing at open issues across every project, including a call for driver-plugin authors.[1]
Eleven README languages. The repository README is published in eleven languages, from Italian and Spanish to Chinese, Japanese, and Tagalog, signalling a broad volunteer translator base.[1]
5.2k stars shown on the site. The official site header displayed a 5.2k star count for TabularisDB/tabularis when observed in October 2026; treat it as a snapshot, not a live value.[1]
Platforms and deployment
Windows, macOS, and Linux installers. Distribution covers WinGet and a Windows setup exe, Homebrew cask plus Apple Silicon and Intel DMGs, and Snap, Flatpak, AUR, AppImage, deb, and rpm on Linux.[1][2]
Local configuration with portable option. Connection profiles, saved queries, and settings live in the platform config directory, relocatable through Settings or the TABULARIS_DATA_DIR variable, for example onto a synced folder; installed plugins stay local.[1]
SSH and Kubernetes tunnel workflows. Solution pages document reaching databases through SSH or Kubernetes tunnels with system keychain storage, aimed at secure and on-call access patterns.[1]
Questions worth asking
How much does Tabularis cost?
Which databases does Tabularis support?
PostgreSQL, MySQL/MariaDB, and SQLite ship built in. Everything else, including DuckDB, ClickHouse, Redis, Firestore, SQL Server, MongoDB, and Oracle, arrives through installable plugins listed in the plugin registry, with further engines tracked on the bounty board.[1]
How do AI agents connect to Tabularis?
Tabularis ships a built-in MCP server started with tabularis --mcp, with one-click install configuration for Claude Desktop, Cursor, and Windsurf documented under Settings, plus JSON or TOON result formats for compact tabular output to language models.[1]
How are new database engines added?
Drivers are standalone executables speaking JSON-RPC 2.0 over stdin/stdout, so a plugin can be written in any language following the plugin guide. Official plugins install from Settings with no restart, and unserved engines can be claimed or sponsored on the bounty board.[1]
Sources and editorial review
Source research. Reviewed on October 13, 2026. Written by Sanjib Sah. AI assisted; reviewed by a human editor.
We have not run or tested this product.
Inspected sample: the TabularisDB organisation discussions index, with active General, Announcements, Ideas, Q&A, Show and tell, and Polls categories and threads from March through October 2026, including maintainer announcements and user questions. No Reddit originals or G2 reports about Tabularis were inspected, so no testimonial-grade community quotes are included and the testimonials array is empty. Discussion threads were observed at index level only; no individual report is quoted here.
- Tabularis README - TabularisDB/tabularis
official · Observed October 13, 2026
GitHub README for TabularisDB/tabularis: desktop SQL workspace, 3 built-in drivers plus 21 plugins, MCP server, notebooks, visual EXPLAIN, JSON-RPC plugin system, Ollama text-to-SQL, Why-tabularis comparison vs DBeaver CE, TablePlus, and Beekeeper Studio dated June 2026, installers, local config, React plus Tauri stack, sponsors, Apache-2.0 license.
- Tabularis - Open-Source Desktop Client for Modern Databases
official · Observed October 13, 2026
Official tabularis.dev homepage: multi-database workspace, notebooks, visual EXPLAIN, query builder, MCP server, plugins, bounty board, engine and workflow solutions, SSH and keychain secure-client positioning, DBeaver, DataGrip, TablePlus, Navicat, pgAdmin, and Beekeeper Studio comparison pages, 5.2k star display.
- Tabularis CHANGELOG
official · Observed October 13, 2026
CHANGELOG.md head: v0.27.0 released 2026-10-08 and v0.26.0 on 2026-10-01, showing active independent maintenance across connections, datagrid, editor, EXPLAIN, and settings.
- Download - Tabularis
official · Observed October 13, 2026
Official download page offers desktop installers only with GitHub release artifacts; no cloud signup or hosted edition is offered.
- TabularisDB Discussions on GitHub
community · Observed October 13, 2026
TabularisDB organisation discussions index: active General, Announcements, Ideas, Q&A, Show and tell, and Polls categories with threads from March through October 2026, including maintainer announcements and user questions. Shows an ongoing user and contributor conversation around the product.
Compared with these products
No explicit proprietary baseline is recorded.