Alexa+ Bedrock Orchestrator MCP

An open-source Streamable HTTP MCP server empowering Alexa+ with multi-step reasoning, real-time IoT context, and agentic task orchestration powered by AWS Bedrock Claude 3.5 Sonnet.

Comment

Inspiration

Voice interfaces and smart home AI are transitioning from basic command-and-response engines to fully autonomous, context-aware agents. With the launch of Alexa+ and the Model Context Protocol (MCP) specification (2025-11-25)[cite: 4, 15], developers can now build external server capabilities that reason dynamically over complex tools[cite: 4, 15]. We built Alexa+ Bedrock Orchestrator MCP to bridge Alexa+ with AWS Bedrock, providing real-time multi-step intelligence across smart devices and ambient surfaces[cite: 4, 16].

What it does

Our project is a production-ready Model Context Protocol (MCP) Server running over Streamable HTTP[cite: 4, 15, 17]. It provides:

  • Alexa+ MCP Tool Discovery: Native implementation of /mcp/rpc handling standard JSON-RPC 2.0 requests, tool listing (tools/list), and execution dispatching (tools/call)[cite: 4, 15, 17].
  • AWS Bedrock Reasoning Engine: Dynamically routes user requests to Claude 3.5 Sonnet on AWS Bedrock via boto3 for multi-step contextual planning[cite: 4, 16].
  • Ambient Context Aggregation: Allows Alexa+ to query real-time IoT device state (Ring doorbells, Fire TV state, and Wearable AI metrics) through unified MCP tools[cite: 4, 15, 16].

How we built it

  • Backend Framework: Built with Python and FastAPI to support both standard JSON-RPC HTTP POST endpoints and asynchronous Streamable HTTP Server-Sent Events (SSE)[cite: 4, 15, 17].
  • Agentic Workflows: Integrated AWS Bedrock Runtime (boto3 client) using the anthropic.claude-3-5-sonnet model[cite: 4, 16].
  • Protocol Standard: Strictly compliant with the MCP spec version 2025-11-25[cite: 4, 15, 17].

Challenges we ran into

  • Streamable HTTP Compatibility: Ensuring low-latency SSE connections and full JSON-RPC 2.0 handshake compatibility required precise path and payload normalization[cite: 4, 15, 17].
  • Credential Fallback Handling: Designing robust error boundaries so that the server gracefully provides simulated responses when live AWS credentials aren't active in local testing[cite: 4, 16].

Accomplishments that we're proud of

  • Fully functional Streamable HTTP MCP server that instantly connects Alexa+ agent capabilities with AWS Bedrock models[cite: 4, 15, 16, 17].
  • 100% open-source under the MIT license with complete setup documentation[cite: 4, 5, 16].

What's next for Alexa+ Bedrock Orchestrator MCP

  • Adding cross-device orchestration triggers between Ring event streams and Fire TV overlay notifications[cite: 4].
  • Introducing specialized MCP agent tools for long-term user memory state across sessions[cite: 4].

Built With

  • alexa-plus
  • aws-bedrock
  • boto3
  • fastapi
  • json-rpc
  • mcp
  • python
  • streamable-http
  • uvicorn
Share this project:

Updates

Submission history