Syngnat/GoNavi

GoNavi – A native‑speed, AI‑aware database client

What it is

• A cross‑platform desktop application (and optional web server) for connecting to a huge variety of data sources – relational DBs, caches, vector stores, message queues, time‑series, search engines, and more.
• Built with Wails v2 (Go + native WebView) and React 18, so the binary is ~30 MB and starts quickly, avoiding the heavyweight Electron runtimes used by many other DB GUIs.
• Provides first‑class MCP (Model‑Control‑Protocol) support: AI agents can be handed live schema information and other context without ever exposing passwords, enabling code‑generation, query‑optimisation, and other LLM‑driven assistance.

Key capabilities

Category	Highlights
AI‑assisted SQL	Connects to OpenAI, Gemini, Claude, or any OpenAI‑compatible endpoint. Slash‑command interface lets the model generate, explain, or optimise SQL using the current table schema.
Unified data cockpit	One UI to manage connections, run queries, edit data in‑place, batch‑edit, commit/rollback transactions, and export results (CSV, XLSX, JSON, Markdown).
Broad source support	Built‑in drivers for MySQL, PostgreSQL, Oracle, Redis, Milvus, ClickHouse, Kafka, RabbitMQ, Elasticsearch, etc.; optional drivers for many more (SQL Server, SQLite, MongoDB, Trino, …).
Performance‑focused UI	Virtualised data grid for massive result sets, Monaco editor with schema‑aware completions, and fast start‑up thanks to native WebView.
Observability & safety	Execution logs, audit centre, secret‑redacted storage, optional write‑protection, and MCP safety defaults (schema‑only unless allowMutating=true).
Deployability	Desktop binaries for macOS, Windows, Linux; experimental HTTP‑served mode; Docker/Podman/K8s/Helm images for the web UI and MCP server; auto‑update checks.

How you would use it

1. Install the binary (or run the Docker image) and add a connection – the UI guides you through DSN creation, SSH tunnelling, or proxy configuration.
2. Open a query tab; the embedded Monaco editor offers syntax highlighting and, if you enable an AI provider, you can type /gen SELECT … to have the model write a query based on the selected tables.
3. Run the query, edit results directly in the grid, or export them.
4. If you have an LLM‑powered coding agent (e.g., Claude Code, DeepSeek Harness), start the MCP server; the agent can request the current schema and receive it without ever seeing the connection password, allowing it to generate safe migration scripts or data‑access code.
5. For remote teams, spin up the experimental web‑server mode behind a reverse proxy and let users access the same UI via a browser.

Getting started

# Prereqs
brew install go node   # or apt/yum equivalents
go install github.com/wailsapp/wails/v2/cmd/wails@v2.11.0

# Development (hot‑reload)
git clone https://github.com/Syngnat/GoNavi.git && cd GoNavi
wails dev

# Production build
wails build          # binaries end up in build/bin

Or simply download a pre‑built binary from the [Releases page] (macOS, Windows, Linux).

CLI companion A headless gonavi executable is shipped alongside the desktop app, letting you list connections, run queries, and export data from scripts or CI pipelines without any UI.

License Apache 2.0 – you can freely use, modify, and redistribute the software.

Who might benefit

• Database administrators who want a lightweight, fast client that can also hand schema data to AI assistants.
• Developers building LLM‑driven code‑generation or data‑analysis tools and need a secure way to expose database metadata.
• Teams that need a single pane‑of‑glass for heterogeneous data stores (SQL, NoSQL, vector, MQ) without the bloat of Electron‑based tools.

---

Links

• 🌐 Website
• 📦 GitHub Releases
• 📖 Full README (English)
• 📖 README in Chinese