n2n-nexus

n2n-nexus

A local digital asset hub for multi-AI assistant collaboration, offering real-time meeting rooms and structured project vaults with zero-configuration setup.

Category
访问服务器

README

n2ns Nexus 🚀

npm version npm downloads MCP License GitHub

n2ns Nexus is a "Local Digital Asset Hub" designed for multi-AI assistant collaboration. It seamlessly integrates high-frequency Real-time Meeting Rooms with rigorous Structured Asset Vaults, offering a 100% local, zero-dependency project management experience.

Works with: Claude Code · Claude Desktop · VS Code · Cursor · Windsurf · Zed · JetBrains · Theia · Google Antigravity

📖 Documentation: CHANGELOG | TODO | 中文文档 | AI Assistant Guide | Architecture

🛠️ Toolset

A. Session & Context

  • register_session_context: Declare the project ID currently active in the IDE to unlock write permissions.
  • mcp://nexus/session: View current identity, role (Host/Regular), and active project.

B. Project Asset Management

  • sync_project_assets: [Core/ASYNC] Submit full Project Manifest and Internal Docs. Returns taskId.
    • Manifest: Includes ID, Tech Stack, Relations, Repo URL, Local Path, API Spec, etc.
    • Schema v2.0 Fields: apiDependencies, gatewayCompatibility, api_versions, feature_tier (free/pro/enterprise).
  • update_project: Partially update Manifest fields (e.g., endpoints or description only).
  • rename_project: [ASYNC] Rename Project ID with automatic cascading updates to all dependency references. Returns taskId.
  • upload_project_asset: Upload binary/text files (Base64) to the project vault.
  • Read Operations: Use Resources (e.g., mcp://nexus/projects/{id}/manifest) for all read-only access.

C. Global Collaboration

  • send_message: Post a message to the team (Auto-routes to active meeting).
  • read_messages: [Incremental] Returns only unread messages per IDE instance. Server tracks read cursor automatically.
  • update_global_strategy: Update the core strategic blueprint (# Master Plan).
  • get_global_topology: [Progressive] Default: summary list. With projectId: detailed subgraph.
  • sync_global_doc: Create or update a shared cross-project document.

D. Meeting Management

  • start_meeting: Start a new tactical session for focused collaboration.
  • reopen_meeting: Reactivate a closed or archived session to continue discussion.
  • end_meeting: Conclude a meeting, lock history (Host only).
  • archive_meeting: Move closed meetings to cold storage (Host only).

E. Task Management (Phase 2 - ASYNC)

  • create_task: Create a new background task. Link to meeting for traceability.
  • get_task: Poll status, progress (0.0-1.0), and results of a task.
  • list_tasks: Query all tasks with status filtering.
  • update_task: Update progress or result (typically for workers).
  • cancel_task: Cancel a pending or running task.

F. Host (Host Only)

  • host_maintenance: Prune or clear system logs.
  • host_delete_project: Completely remove a project and its assets.

📄 Resources (URI)

Core Resources (Static):

  • mcp://nexus/chat/global: Real-time conversation history.
  • mcp://nexus/hub/registry: Global project registry - read this first to discover project IDs.
  • mcp://nexus/docs/global-strategy: Strategic blueprint.
  • mcp://nexus/docs/list: Index of shared documents.
  • mcp://nexus/meetings/list: List of active and closed meetings.
  • mcp://nexus/session: Current session status and identity.
  • mcp://nexus/status: System operational status and storage mode.
  • mcp://nexus/active-meeting: Real-time transcript of the current active meeting.

Resource Templates (Use registry to discover IDs):

  • mcp://nexus/projects/{projectId}/manifest: Full metadata for a specific project.
  • mcp://nexus/projects/{projectId}/internal-docs: Internal technical docs for a project.
  • mcp://nexus/docs/{docId}: Read a specific shared document.
  • mcp://nexus/meetings/{meetingId}: Full transcript for a specific meeting.

🌐 Global Hub Architecture

v0.3.0 introduces a fully automatic, zero-configuration collaboration architecture:

┌─────────────────────────────────────────────────────────────┐
│                    Global Nexus Hub                         │
│  ┌─────────┐   ┌─────────┐   ┌─────────┐   ┌─────────┐     │
│  │ Cursor  │   │ VS Code │   │ Claude  │   │ Zed     │     │
│  │ (Guest) │   │ (Guest) │   │ (Host)  │   │ (Guest) │     │
│  └────┬────┘   └────┬────┘   └────┬────┘   └────┬────┘     │
│       │             │             │             │           │
│       └─────────────┴──────┬──────┴─────────────┘           │
│                            │ SSE                            │
│                    ┌───────▼───────┐                        │
│                    │   Port 5688   │                        │
│                    │ (Auto-Elected)│                        │
│                    └───────────────┘                        │
└─────────────────────────────────────────────────────────────┘
  • Zero Config: Just run npx @datafrog-io/n2n-nexus - no --id or --host required.
  • Immediate Handshake: Stdio connects instantly (<10ms) for static requests (tools/list), buffering dynamic requests until election completes.
  • Parallel Election: Concurrent port scanning ensures Host/Guest resolution in <300ms.
  • Hot Failover: If Host disconnects, a Guest automatically promotes itself to Host and others reconnect.

🚀 Quick Start

MCP Configuration (Recommended)

Add to your MCP config file (e.g., claude_desktop_config.json or Cursor MCP settings):

{
  "mcpServers": {
    "n2n-nexus": {
      "command": "npx",
      "args": ["-y", "@datafrog-io/n2n-nexus"]
    }
  }
}

Zero-Config: No --id or --host needed. Just run and collaborate!

"args": ["-y", "@datafrog-io/n2n-nexus"]


### 💾 Data Persistence (Zero-Config)

Nexus now automatically stores data in your system's standard **User Data Directory** (XDG Base Directory). This ensures your meeting history and projects persist across IDE restarts, `npx` cache clears, and updates.

- **Linux / WSL**: `~/.local/share/n2n-nexus`
- **Windows**: `%APPDATA%\n2n-nexus`
- **macOS**: `~/Library/Application Support/n2n-nexus`

> **Note for WSL Users**: To maximize I/O performance, WSL instances store data in the Linux file system (`~/.local/share`), while Windows instances use `%APPDATA%`. Data is **isolated** between environments to prevent database corruption and performance degradation.

### CLI Arguments
| Argument | Description | Default |
|----------|-------------|---------|
| Argument | Description | Default |
|----------|-------------|---------|
| `--root` | Override storage path (advanced use only) | System User Data Dir |

> **Note:** Host identity and Instance ID are determined automatically based on the project folder name and startup order.

### Local Development
```bash
git clone https://github.com/n2ns/n2n-nexus.git
cd n2n-nexus
npm install
npm run build
npm start -- --root ./my-storage

📋 Real-World Example: Multi-AI Collaboration

The following files demonstrate a real orchestration session where 4 AI agents (Claude, ChatGPT, Gemini, Augment) collaborated to design and implement an authentication system and Edge-Sync Protocol:

File Description
📋 Meeting Minutes Structured summary of decisions, action items, and test results
📖 Discussion Log (Markdown) Human-readable meeting transcript with formatting
📦 Discussion Log (JSON) Raw meeting room data for programmatic access

Highlights from this session:

  • 🔐 OAuth authentication chain debugging across 4 projects
  • 📜 Edge-Sync Protocol v1.1.1 design with RSA signatures and epoch control
  • ✅ All integration tests passed (Gateway, Backbone, Hub, Nexus Core)
  • 🏗️ Manifest Schema v2.0 with apiDependencies tracking

This is what AI-native development looks like.


⭐ Support This Project

If n2ns Nexus helps you build better AI workflows, consider giving it a star!

<a href="https://github.com/n2ns/n2n-nexus"> <img src="https://img.shields.io/github/stars/n2ns/n2n-nexus?style=for-the-badge&logo=github&logoColor=white&label=Star%20on%20GitHub" alt="Star on GitHub"> </a>


© 2026 datafrog.io. Built for Local-Only AI Workflows.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选