# GoNavi [Health: Active]

**Category:** 🗄️ Databases  
**Repository:** https://github.com/Syngnat/GoNavi  
**GitHub Stars:** 2087  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/gonavi

## Description
High-performance multi-data-source database client — ~30MB, AI & MCP ready, zero Electron bloat. | 高性能多数据源数据库客户端：约 30MB，AI 与 MCP 就绪，告别 Electron 膨胀。

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What GoNavi does

GoNavi combines a desktop database workbench with an MCP-facing service for agent workflows. It is built with Go, Wails, and React rather than Electron, and is designed to work across Windows, macOS, and Linux. The application supports multiple classes of data source, including SQL databases, cache systems, vector stores, message queues, search systems, time-series databases, and domestic database products.

The workbench provides connections, a SQL editor, result grids, object trees, saved queries, execution history, and an audit center. Data-grid workflows include batch editing, export, transaction submission, and rollback. SQL evidence in the audit view is desensitized, which is intended to help review activity across connections.

## How it works

GoNavi uses a Go core and a native system WebView through Wails. Its architecture includes connection handling, audit and synchronization components, an MCP HTTP interface, and a web server. The MCP layer is intended to pass structured context between connected data sources and coding agents without moving database passwords off the host.

The built-in AI assistant can use schema context to help draft SQL. Generated statements can be inserted into the editor, previewed, or run from the application. The README positions the GUI as the place for schema work, edits, and EXPLAIN operations, so GoNavi fits workflows where agent assistance is paired with visible database controls rather than completely unattended changes.

Supported data-source examples include MySQL, PostgreSQL, Redis, Kafka, Milvus, OceanBase, and ClickHouse. These examples represent the range of systems the client is intended to bring together; connection support depends on the available drivers and product configuration.

## Tools and capabilities

- Connect to multiple categories of data source from one desktop workbench.
- Expose an MCP HTTP interface for agent-oriented database context.
- Edit and run SQL with syntax highlighting, schema context, and result grids.
- Review SQL execution history and refill previous statements into the editor.
- Record desensitized SQL evidence in an audit center.
- Export data, perform batch edits, and submit or roll back transactions.
- Draft SQL with the schema-aware assistant and preview or insert generated statements.

## Limitations and notes

GoNavi’s small installer footprint should not be treated as a memory-use measurement. The project distinguishes roughly 20–26 MB release installers from runtime RSS and reports a Linux WebKit sample with substantially higher memory use. Results can differ by operating system, WebView implementation, display setup, and workload.

The product is desktop-first, so the MCP interface is part of a larger local workbench rather than a standalone database gateway described separately. Agent users should account for the GUI-controlled workflow around schema changes, edits, and EXPLAIN operations. GoNavi is released under the Apache-2.0 license.

_Full upstream README: https://allmcps.com/mcp/gonavi/readme_