Inspiration
Volunteer coordinators often ask people for the same availability and equipment details in several places. PaperPath explores how an assistant could make that paperwork less repetitive while keeping the applicant in control.
What PaperPath does
PaperPath is a local practice simulator for a fictional community volunteer program. A person can describe an example request; the prototype uses an MCP client and local MCP server to prepare a draft with source evidence. Missing, conflicting, or unsupported details stay unresolved for review. A person's availability correction is shared across the application and availability/equipment form. After explicit practice confirmation, the app exports two PDFs tied to the same revision.
How we built it
The web app uses TypeScript, Node.js, Express, SQLite, and Vite. A local Model Context Protocol client and server communicate over Streamable HTTP; Groq supports candidate extraction; pdfme renders the two forms.
Challenges and learning
The main challenge is separating a suggestion from an approved fact. Showing source text, preserving unknowns, requiring explicit correction and confirmation, and binding both PDFs to one revision are core trust decisions. This simulator reinforced that an assistant should make review easier without pretending to submit on a person's behalf.
Scope and next steps
This is a local prototype, not an Alexa+ integration or a service that submits applications to organizations. Next, we plan to validate the complete journey with more fictional requests and finish release checks. The linked video shows an earlier prototype journey and is illustrative; current behavior may differ. It does not demonstrate Alexa+ integration or real submissions.
Log in or sign up for Devpost to join the conversation.