Context-Aware Personal Assistant MCP Server for Alexa+

A Model Context Protocol (MCP) server built for the Amazon Developer Hackathon 2026 (Alexa+ Track) that provides persistent memory, calendar integration, task management, and proactive assistance for Alexa+.

🏆 Hackathon Entry

• Track: Alexa+ (MCP Server)

• Division: New Experience

• Competition: Build, Ship, Shape: Amazon Developer Hackathon

• Deadline: October 23, 2026

Related MCP server: AGI MCP Server

✨ Features

Core Capabilities

• Persistent Memory: Store and retrieve memories with tags, importance scoring, and semantic search

• Task Management: Create, update, and track tasks with priorities, due dates, and tags

• Calendar Integration: Schedule events, get upcoming events, and manage recurrence

• Proactive Assistance: Context-aware suggestions based on current time, events, and tasks

• User Preferences: Persistent preference storage for personalization

MCP Protocol Compliance

• MCP Spec: 2025-11-25 (Streamable HTTP)

• Transport: Streamable HTTP + SSE for real-time updates

• Capabilities: Tools, Resources, Prompts

• Session Management: Multi-session support with proper initialization

Alexa+ Integration Ready

• Self-hosted MCP server deployable anywhere

• Streamable HTTP transport as required by Alexa+

• OpenAPI-compatible endpoints

• Health checks and monitoring

🚀 Quick Start

Prerequisites

• Python 3.11+

• Docker (optional, for containerized deployment)

Local Development

# Clone and setup
git clone <your-repo>
cd amazon-alexa-mcp-hackathon

# Create virtual environment
python3 -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Run STDIO server (for local testing)
python server.py

# Run HTTP server (for Alexa+ integration)
python http_server.py

Docker Deployment

# Build and run
docker-compose up --build -d

# Check health
curl http://localhost:8000/health

📡 MCP Endpoints

HTTP Transport (Alexa+ Compatible)

• POST /mcp - Main MCP endpoint (Streamable HTTP)

• GET /mcp - SSE endpoint for real-time updates

• GET /health - Health check

• DELETE /mcp/session/{id} - Session management

Available Tools

Available Resources

🎯 Use Cases for Alexa+

1. Morning Briefing: "Alexa, what's my day look like?" → Gets context summary with events, tasks, suggestions

2. Memory Recall: "Alexa, remember I need to call mom about the reunion" → Stores memory with tags

3. Task Management: "Alexa, add a task to review the quarterly report by Friday" → Creates task with due date

4. Proactive Alerts: "You have a meeting in 15 minutes" → Proactive suggestions

5. Context Continuity: Memories persist across sessions, enabling long-term personal assistant behavior

🏗️ Architecture

┌─────────────────┐     ┌──────────────────┐     ┌─────────────────┐
│   Alexa+        │────▶│  MCP HTTP Server │────▶│  Context-Aware  │
│   (Client)      │     │  (Streamable HTTP)│     │  Assistant Core │
└─────────────────┘     └──────────────────┘     └─────────────────┘
                                                        │
                                                        ▼
                                               ┌─────────────────┐
                                               │  Persistent     │
                                               │  Storage (JSON) │
                                               └─────────────────┘

📁 Project Structure

amazon-alexa-mcp-hackathon/
├── server.py              # STDIO MCP server (local testing)
├── http_server.py         # HTTP MCP server (Alexa+ integration)
├── requirements.txt       # Python dependencies
├── Dockerfile             # Multi-stage Docker build
├── docker-compose.yml     # Container orchestration
├── data/                  # Persistent data (gitignored)
│   ├── memories.json
│   ├── tasks.json
│   ├── events.json
│   └── preferences.json
└── README.md

🔧 Configuration

Environment variables:

• PORT - Server port (default: 8000)

• PYTHONUNBUFFERED - Unbuffered output for logging

🧪 Testing

Test STDIO Server

echo '{"jsonrpc":"2.0","id":"1","method":"initialize","params":{}}' | python server.py

Test HTTP Server

# Initialize
curl -X POST http://localhost:8000/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":"1","method":"initialize","params":{}}'

# List tools
curl -X POST http://localhost:8000/mcp \
  -H "Content-Type: application/json" \
  -H "Mcp-Session-Id: test-session" \
  -d '{"jsonrpc":"2.0","id":"2","method":"tools/list","params":{}}'

# Call tool
curl -X POST http://localhost:8000/mcp \
  -H "Content-Type: application/json" \
  -H "Mcp-Session-Id: test-session" \
  -d '{"jsonrpc":"2.0","id":"3","method":"tools/call","params":{"name":"get_context","arguments":{}}}'

🏅 Hackathon Highlights

Why This Wins

1. Complete MCP 2025-11-25 Implementation - Full Streamable HTTP transport compliance

2. Real-World Utility - Solves actual Alexa+ limitation: lack of persistent context

3. Production Ready - Docker, health checks, multi-session, CORS

4. Extensible Design - Easy to add new tools, resources, integrations

5. Privacy-First - Local storage, no external dependencies

Judging Criteria Alignment

📝 Submission Checklist

• Working MCP server (STDIO + HTTP)

• MCP 2025-11-25 Streamable HTTP transport

• Self-hosted, deployable anywhere

• GitHub repo with source code

• Demo video (3 min max)

• Product feedback on Amazon Developer tools

• Alexa+ track requirements met

📄 License

MIT License - See LICENSE file for details.

🤝 Acknowledgments

• Amazon Developer Team for the hackathon opportunity

• MCP Specification authors

• NVIDIA Nemotron team (inspiration for context-aware AI)

Built for the Amazon Developer Hackathon 2026 🚀