# 🐝 BeeMind: An Ambient AI Agent for Amazon Bee

### Technical Subtitle
**Proactive Memory, Commitment Tracking & MCP-Powered Tool Orchestration with Amazon Bedrock**

<p align="center">
  <img src="public/logo.png" alt="BeeMind Logo" width="140" />
</p>

[![Hackathon](https://img.shields.io/badge/Amazon%20Developer%20Hackathon-2026-orange.svg)](https://amazonappdev2026.devpost.com/)
[![Track](https://img.shields.io/badge/Track-Bee%20(Wearable%20AI)-amber.svg)](https://amazonappdev2026.devpost.com/)
[![AWS Builder Challenge](https://img.shields.io/badge/Challenge-AWS%20Bedrock%20AgentCore-blue.svg)](https://amazonappdev2026.devpost.com/)
[![License: MIT](https://img.shields.io/badge/License-MIT-emerald.svg)](LICENSE)

> Built for the **[Build, Ship, Shape: Amazon Developer Hackathon 2026](https://amazonappdev2026.devpost.com/)**  
> Targeting the **Bee (Wearable AI)** Track, **AWS Builder Mini Challenge**, and **Open Source Mini Challenge**.

---

## 🌟 Overview

**BeeMind** is an ambient intelligence companion and open **Model Context Protocol (MCP)** server designed for the **Amazon Bee Wearable** ecosystem (Bee Frontier, Bee Eclipse, and Apple Watch running Bee software).

By converting continuous acoustic streams into actionable intelligence without storing raw audio, BeeMind powers:
1. **🧠 Proactive Second Brain & Recall:** Semantic Q&A over daily memories and recommendations using Amazon Bedrock embeddings.
2. **🤝 Real-Time Meeting Commitments:** Instant extraction of action items, assignees, and deadlines, syncing to Jira, Notion, and Slack.
3. **👷 Hands-Free Field & Industrial Workflow:** Voice-checked audit reports, vibration/pressure telemetry, and hazard alerts for deskless workers.
4. **⚡ Ambient Agentic Bridge & Bee MCP Server:** Open Model Context Protocol (MCP) server enabling external AI agents (Claude, Bedrock AgentCore) to interact with wearable context.

---

## 🏗️ Architecture

```
                       ┌────────────────────────────────────────┐
                       │  Amazon Bee Wearable / Apple Watch     │
                       │  (Acoustic Diarization & Telemetry)    │
                       └───────────────────┬────────────────────┘
                                           │
                        Bee CLI / REST Ingestion Endpoint
                                           │
                                           ▼
┌─────────────────────────────────────────────────────────────────────────────┐
│                         BeeMind Ambient Core Engine                         │
│                                                                             │
│   ┌────────────────────────┐                   ┌────────────────────────┐   │
│   │   Amazon Bedrock       │                   │    Model Context       │   │
│   │   Runtime Service      │                   │    Protocol (MCP)      │   │
│   │   (Claude 3.5 Sonnet / │◄─────────────────►│    stdio Server        │   │
│   │    Nova / Titan)       │                   │    (5 Active Tools)    │   │
│   └────────────────────────┘                   └────────────────────────┘   │
└──────────────────────────────────────┬──────────────────────────────────────┘
                                       │
                                       ▼
┌─────────────────────────────────────────────────────────────────────────────┐
│                    BeeMind Interactive Companion UI                         │
│                                                                             │
│   [Live Audio Stream]    [Second Brain]    [Action Items]    [Field Mode]   │
└─────────────────────────────────────────────────────────────────────────────┘
```

---

## 🚀 Quickstart & Setup

### Prerequisites
- Node.js 18+ (tested on Node.js v24)
- npm or pnpm

### 1. Clone & Install
```bash
git clone <your-repo-url>
cd Amazon
npm install
```

### 2. Environment Variables (Optional)
Copy `.env.example` to `.env.local`:
```bash
cp .env.example .env.local
```
> **Note for Judges:** If AWS credentials are not set, BeeMind runs automatically with its **Intelligent Zero-Config Local Engine** so you can evaluate all features immediately!

### 3. Run the Development Server
```bash
npm run dev
```
Open **[http://localhost:3000](http://localhost:3000)** in your browser.

### 4. Run the Model Context Protocol (MCP) Server
To connect BeeMind to Claude Desktop, Cursor, or any MCP-compatible agent client:
```bash
npm run mcp:run
```

MCP Configuration snippet (`claude_desktop_config.json`):
```json
{
  "mcpServers": {
    "beemind-wearable": {
      "command": "npx",
      "args": ["tsx", "server/mcp/index.ts"],
      "cwd": "/path/to/project"
    }
  }
}
```

### 5. Run the Bee Wearable Stream Simulator
Simulate real-time streaming audio events into the companion dashboard:
```bash
npm run simulate:bee
```

---

## 🛠️ MCP Tools Reference

The BeeMind MCP server exposes 5 standardized tools:

| MCP Tool Name | Description | Parameters |
|---|---|---|
| `bee_get_device_status` | Returns wearable battery level, connection, and recording mode | None |
| `bee_get_recent_transcripts` | Returns the latest diarized spoken transcripts with confidence | `limit`, `mode` |
| `bee_search_memory` | Semantic RAG query across user's wearable memory | `query`, `category` |
| `bee_get_action_items` | Retrieves extracted commitments and assigned tasks | `status`, `priority` |
| `bee_dispatch_task` | Dispatches autonomous agent workflows from spoken intent | `actionType`, `summary` |

---

## 📁 Repository Structure

```
├── app/
│   ├── api/
│   │   ├── bedrock/chat/route.ts     # Natural language memory recall
│   │   ├── bedrock/status/route.ts   # Device & AI engine status
│   │   └── bee/stream/route.ts       # Audio chunk ingestion
│   ├── globals.css                   # Glassmorphism & audio waveform styles
│   ├── layout.tsx                    # Root layout with BeeMind branding
│   └── page.tsx                      # Crextio luxury bento companion app
├── lib/
│   ├── bedrock-service.ts            # AWS Bedrock + Zero-Config fallback engine
│   ├── bee-store.ts                  # Seed data & store definitions
│   └── types.ts                      # TypeScript models
├── public/
│   └── logo.png                      # Official BeeMind cybernetic logo
├── server/
│   ├── bee-simulator.ts              # CLI stream generator
│   └── mcp/index.ts                  # Official Model Context Protocol Server
├── AMAZON_FEEDBACK.md                # Feedback for Amazon Bee Developer team
├── DEMO_SCRIPT.md                    # 3-minute video pitch script
├── DEVPOST_SUBMISSION.md             # Devpost text ready to submit
├── LICENSE                           # Open source MIT license
└── package.json
```

---

## 📄 License
Released under the [MIT License](LICENSE).