Inspiration
Drivers often need answers while moving: where can I park, what is nearby, are there road alerts, and should I release my previous parking spot? Traditional map apps are strong for navigation, but they do not focus enough on collaborative parking availability and voice-first parking decisions.
ParkShare was built around a real driver problem: reaching a destination is not enough if the driver still has to search for parking, type while driving, or forget to release a free spot for other users.
What it does
ParkShare is a real parking and driving app extended with a real MCP Streamable HTTP server for the Alexa+ track. It lets a driver:
- Search for any destination or nearby place.
- Find useful places from the current location.
- Get parking-aware guidance around a destination.
- Confirm whether the previous parking spot was left free before starting a new route.
- Report free or occupied parking spaces.
- Receive route context and community road-alert information.
- Use an Alexa+ MCP copilot to ask for parking, nearby places, route alerts, and spot-release actions.
How we built it
The ParkShare mobile product uses Flutter, Google Maps, location services, destination search, speech-to-text, text-to-speech, Android integrations, and backend APIs for parking sessions, routes, places, road alerts, observability, and assistant behavior.
For the hackathon, we built a real Node.js MCP Streamable HTTP server implementing protocol version 2025-11-25. It exposes five tools:
find_nearby_placefind_parking_near_destinationget_road_alertsrelease_parking_spotconfirm_previous_spot_free
The public companion web experience uses the same ParkShare tool functions and provides live location-aware searches, real map routes, parking results, community alerts, action logs, and browser speech output.
What we built during the hackathon
ParkShare existed before the hackathon as a mobile parking and driving app. During the hackathon window, we significantly updated it with the Alexa+ MCP mobility copilot server, improved voice-first flows, safer route-start parking confirmation, stronger nearby-place search, and community road-alert workflows.
The submitted repository contains the MCP server, production worker, setup instructions, validation scripts, smoke tests, demo flows, and the companion web UI shown in the video.
Challenges
The main challenge was turning a real mobile driving workflow into a clear conversational tool interface. ParkShare combines maps, parking state, route context, live user location, voice commands, and community alerts, while Alexa+ needs concise structured tools.
We also had to keep assistant answers safe for driving: short, actionable, and backed by structured results rather than long chat responses.
Accomplishments
We created a focused Alexa+ MCP server around a real mobility product. The experience demonstrates finding nearby services, finding destination parking, checking and reporting road alerts, confirming a previous spot, and releasing a parking spot for the community.
The submission includes an open-source repository, license, README, validation and smoke tests, a public live experience, a public English demo video, product feedback, and friction logs.
What we learned
The strongest mobility assistant is not just a chatbot. It needs location context, route context, parking state, safety rules, and short answers that work while the user is driving.
The Alexa+ track pushed ParkShare toward a more agentic model: a driver should be able to ask one natural question and receive one useful action without digging through screens.
What's next
Next we plan to connect the MCP server more deeply to ParkShare production APIs, improve Android Auto behavior, strengthen voice-only workflows, add confidence and verification rules for road alerts and parking reports, and expand parking guidance around destination arrival.
Log in or sign up for Devpost to join the conversation.