nodespace-mcp
Enables AI coding assistants to query a local knowledge base for persistent, searchable project context, reducing re-explanation and token usage.
README
NodeSpace
Faster context. Fewer tokens.
AI coding assistants forget everything between sessions. NodeSpace gives them persistent, searchable access to your project knowledge — so you stop re-explaining your codebase every time you start a conversation.
nodespace.ai · Download · Discord
⚠️ Alpha Preview — NodeSpace is in early development. Features may change and data formats are not yet stable.

Documentation
For full documentation, see nodespace-docs.
Why NodeSpace
Developers using AI assistants waste time copying files, re-explaining architecture, and watching context degrade mid-session. NodeSpace fixes this by sitting between your knowledge and your AI tools:
- 80% fewer roundtrips — AI agents query your knowledge base via MCP instead of scanning files with grep/ripgrep
- Runs entirely on your machine — no cloud accounts, no API calls, no data leaving localhost
- Works offline — on planes, behind VPNs, anywhere
You write things down once. Every AI tool you use can find them instantly.
Installation
Download the Desktop App
| Platform | Format |
|---|---|
| macOS (Apple Silicon) | .dmg |
| Windows | .exe or .msi |
Build from Source
Prerequisites:
- Bun 1.0+ —
curl -fsSL https://bun.sh/install | bash - Rust 1.80+ —
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
git clone https://github.com/NodeSpaceAI/nodespace-core
cd nodespace-core
bun install
bun run tauri:dev
MCP Setup
NodeSpace includes a built-in MCP server that starts automatically when you open the app. Your AI tools connect to it locally — there's nothing to deploy or host.
Claude Code / Cursor / Codex / Other MCP Clients
Add to your MCP settings (e.g., ~/.claude.json for Claude Code):
{
"mcpServers": {
"nodespace": {
"type": "http",
"url": "http://localhost:3100/mcp"
}
}
}
Running Tests
bun run test # Fast unit tests (Happy-DOM)
bun run test:browser # Browser integration tests (Playwright)
bun run test:all # All tests (unit + browser + Rust)
bun run rust:test # Rust backend tests only
Community
- 💬 Join our Discord — ask questions, share feedback, follow development
- 🌟 Star this repo if NodeSpace is useful to you
- 🐛 Report a bug
License
NodeSpace is licensed under the Functional Source License 1.1 (Apache 2.0).
- ✅ Use NodeSpace freely for any purpose
- ✅ Modify the code to fit your needs
See LICENSE for the full text.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。