using ALEXA as a voice assistance
mapping showing urgency

Local problems like plastic waste in a college lake or no easy way to find parking often go unsolved because the people who care and the people with the skills never meet. Nexus Impact connects community problems with student teams and mentors, then tracks the measured result. Nexus Impact is a web platform where:

Citizens and NGOs post community challenges, with a location pin and urgency.
Students are matched into teams by skills, weekly hours and interests, with a plain-language reason for each match.
Mentors guide projects, and the community rates them.
A map shows where problems are, how urgent they are, and a route through nearby ones.
A knowledge library collects completed projects with measured results and evidence, plus guides and lessons learned.
The whole site can be translated into English, Hindi and Marathi. What we built during this hackathon (Alexa+ track) Nexus Impact as a website An MCP server (Model Context Protocol, Streamable HTTP) that exposes the platform to Alexa+ through 7 tools: list_challenges, analyze_problem, find_team, get_progress, map_challenges, get_mentor_ratings and search_library. A simulated Alexa+ experience: a web page that acts as the Alexa+ host. It is a real MCP client, and every answer shows the MCP tool call behind it. We did not use a physical Alexa+ device or an Alexa+ account. An Agent Skill (SKILL.md) that explains how to use the tools.
The tools reuse the website's own logic for analysing a problem and scoring student teams, and read the same Supabase data.

How we built it The website is plain HTML, CSS and JavaScript with Supabase for the database and login. The MCP server is Node.js with Express, the official MCP SDK, zod and supabase-js. The simulated Alexa+ page talks to a small server endpoint that calls the MCP server like any other MCP client.

Challenges we ran into We first planned to add Amazon Bedrock, but our new AWS account was blocked from using it. We switched to the Alexa+ MCP route, which fit our project better and needs no paid services. We also had to port browser-side logic into a server and make sure it read data safely.

What we learned MCP turns an existing app into something an assistant can use without rewriting it. One tool per intent keeps the answers clear.

Honest limits The analysis is rule-based , not an AI model. Team suggestions are only suggestions, and skills are self-declared. The Alexa+ part is simulated.

What's next Test on a real Alexa+ device, add an AI model for richer analysis, and let users submit problems by voice.