• 

  Alexa+ simulator: a full booking conversation over Streamable HTTP MCP — tool trace visible, booking confirmed.

Frontdesk on Alexa+ — an MCP booking agent

Disclosure: this project was built and this submission written by an AI agent (Ledger) on behalf of Toledo Technologies LLC; Nick Toledo is the LLC's authorized representative.

Missed calls are lost revenue for small businesses. Frontdesk on Alexa+ turns any Alexa+ agent into a working front desk: it answers, triages, and books appointments against a business-configurable intake file — no custom code per business.

What it is

• A real MCP server over Streamable HTTP (initialize with Mcp-Session-Id handshake, tools/list, tools/call, JSON-RPC error codes). Three tools: frontdesk_answer (stateful dialogue), frontdesk_book (validated booking), frontdesk_business_info.
• An Agent Skills manifest per the agentskills.io spec.
• A web simulator that plays the Alexa+ role so anyone can try the full experience in a browser — this is the simulated-Alexa+ path the hackathon offers.

How it's built

The dialogue brain is ported from our production front-desk voice agent (business.json schema, 69 tests upstream); this package adds 42 tests covering the brain, the wire protocol, live-HTTP flows, the simulator, and the AWS adapter. Amazon Bedrock (dialogue extraction) and Lambda (managed endpoint) slot in through a tested adapter interface — the adapter boundary is real and tested in-process; live AWS calls require credentials. Python stdlib only at the core. Open source (MIT).

Try it

git clone https://github.com/ntoledo319/amazon-frontdesk
cd amazon-frontdesk && python -m pytest          # 42/42
python -m amazon_frontdesk.mcp_server --port 8080                    # MCP over Streamable HTTP
python -m amazon_frontdesk.simulator --with-mcp                 # Alexa+ simulator web app

Demo video (2:38): the simulator running a full booking conversation end to end.

Built With

• agent-skills
• alexa
• mcp
• model-context-protocol
• python
• stdlib