# tabularis — Desktop DB Client (AI + MCP Built-In)

> tabularis is an open-source desktop DB client for PostgreSQL, MySQL/MariaDB, and SQLite, with notebooks plus optional AI and plugins for extra drivers.

## Install

Copy the content below into your project:

## Quick Use

```bash
brew tap TabularisDB/tabularis
brew install --cask tabularis

# Launch from Applications after install
```

## Intro

A modern, hackable DB client that handles multiple SQL engines and lets you add drivers via plugins—useful when you want one UI for many databases.

**Best for:** Developers who juggle PostgreSQL + MySQL + SQLite and want notebooks, explain, and plugin-based drivers in one desktop app

**Works with:** macOS/Windows/Linux; PostgreSQL, MySQL/MariaDB, SQLite; optional AI providers (e.g., Ollama) per README

**Setup time:** 5–12 minutes

### Key facts (verified)

- README release downloads reference v0.10.3 installers for Windows/macOS/Linux.
- Supports multi-schema PostgreSQL and multi-database selection for MySQL/MariaDB in one connection (per README features).
- Configuration lives under OS-specific app data dirs (e.g., `~/Library/Application Support/tabularis/` on macOS) per README.
- GitHub: 1,474 stars · 98 forks; pushed 2026-05-12 (GitHub API verified).

## Main

A productive way to use tabularis for day-to-day debugging:

1. Create one connection per environment (local / staging / prod read-replica) and name them clearly.
2. Use notebooks for “investigation threads”: keep the query, the result snippet, and the conclusion together.
3. When you need another engine (DuckDB, MongoDB, etc.), use the plugin mechanism the README describes—treat drivers as modular.
4. Keep “fast-safe defaults”: start with `EXPLAIN`, add `LIMIT`, and avoid heavy scans until you’ve narrowed the target tables and indexes.

If you enable AI features, keep prompts scoped to schema + query intent, not secrets or full data dumps.

### FAQ

**Q: Which databases does it support?**
A: The README lists PostgreSQL, MySQL/MariaDB, and SQLite as first-class supported drivers.

**Q: Can I add other databases?**
A: Yes. The README describes a plugin system to add drivers such as DuckDB or MongoDB.

**Q: Is AI required?**
A: No. AI features are optional per README; you can use it as a normal DB client.

## Source & Thanks

> Source: https://github.com/TabularisDB/tabularis
> License: Apache-2.0
> GitHub stars: 1,474 · forks: 98

---

<!-- ZH -->

## 快速使用

```bash
brew tap TabularisDB/tabularis
brew install --cask tabularis

# 安装后从「应用程序」启动
```

## 简介

一个更“可折腾”的现代 DB 客户端：多引擎支持 + 插件扩展驱动，适合用一个 UI 管多种数据库。

**最适合：** 同时用 PostgreSQL/MySQL/SQLite 的开发者：希望有 notebook、Explain 与插件驱动的一体化桌面工具

**适配：** macOS/Windows/Linux；PostgreSQL、MySQL/MariaDB、SQLite；README 提到可选 AI 提供方（如 Ollama）

**配置时间：** 5–12 分钟

### 关键事实（已验证）

- README 的 Release 下载示例展示了 v0.10.3 的 Windows/macOS/Linux 安装包。
- README 功能列表写明：PostgreSQL 支持多 schema；MySQL/MariaDB 支持一连接选择多个数据库。
- README 给出配置目录：macOS 在 `~/Library/Application Support/tabularis/`（其他系统亦有对应路径）。
- GitHub：1,474 stars · 98 forks；最近更新 2026-05-12（GitHub API 验证）。

## 正文

用 tabularis 做日常排障的一种高效方式：

1. 按环境拆连接（local/staging/prod 读副本），命名清晰。
2. 用 notebook 记录“调查链路”：同一处保存 SQL、结果片段、结论。
3. 需要新引擎（DuckDB、MongoDB 等）时，按 README 的插件机制扩展驱动，把“驱动”当模块管理。
4. 保持“快且安全”的默认动作：先 `EXPLAIN`，再加 `LIMIT`，避免一上来全表扫描。

开启 AI 功能时，尽量只给 schema + 意图，不要把秘钥或完整数据直接喂进去。

### FAQ

**支持哪些数据库？**
答：README 明确写了 PostgreSQL、MySQL/MariaDB、SQLite。

**能扩展别的数据库吗？**
答：可以。README 描述了插件系统，可通过插件增加驱动（例如 DuckDB、MongoDB 等）。

**一定要开 AI 吗？**
答：不需要。README 表明 AI 功能是可选项，可当普通 DB 客户端用。

## 来源与感谢

> Source: https://github.com/TabularisDB/tabularis
> License: Apache-2.0
> GitHub stars: 1,474 · forks: 98


---
Source: https://tokrepo.com/en/workflows/tabularis-desktop-db-client-ai-mcp-built-in
Author: Script Depot