Mint Parchment - Prototyped AI Quick Review Assistant

Mint Parchment is an AI-powered study assistant prototype designed to help students refresh academic concepts via concise, auto-generated jot notes.

Comment

Inspiration

As a student, I often needed quick refreshers before exams—simple, structured notes that summarize a topic without forcing me to dig through long lectures or messy notes. With Chrome’s built-in AI Prompt API, I wanted to explore if I can create a lightweight, local, privacy-friendly study assistant could be created.

This sparked Mint Parchment, a prototype AI tool designed to give instant academic jot-notes and support efficient studying and reviewing.

What it does

Mint Parchment is a browser-based AI study assistant that generates short, structured “jot notes” for any academic topic a student enters. Key features include:

  • ✅ AI-powered concept summaries (via Chrome Prompt API)
  • ✅ Auto-organized headings, bullet lists, and micro-definitions
  • ✅ Personal notes panel for users to expand on or edit the AI’s output
  • ✅ Sidebar history that lets users revisit recent topics
  • ✅ Prototype mind map canvas for visual study aids

The tool runs entirely client-side, ensuring speed, simplicity, and privacy.

How we built it

This project was built solo. I implemented all major components, including the Chrome Prompt API integration, topic history management, note-taking interface, and prototype mind-map canvas, from front-end logic to UI design.

Mint Parchment is built entirely with:

  • HTML/CSS/JavaScript
  • Chrome’s Prompt API for local LLM-powered text generation
  • localStorage for persisting session data (topics, notes, summaries)
  • HTML for early mind-map rendering experiments

I also wrote a custom mini-parser that formats model responses into HTML:

  • Converts markdown-like bullets into lists
  • Adds headings
  • Styles code, bold, italics
  • Produces clean, readable jot-notes

The site loads the Prompt API on demand, creates an LLM session, and generates new summaries in real time.

Challenges we ran into

No backend support: I intended to integrate cloud servers and more advanced AI flows, but couldn't get stable cross-domain deployments working in time.

Mind map functionality: I wanted a full node-graph editor where concepts auto-arrange or link together. I built the canvas page and drag logic foundation, but the generative map features are still incomplete.

Prompt API quirks: Because the Prompt API is new, debugging session creation failures, availability states, and formatting inconsistencies took significant time.

LocalStorage structure: Managing topic history, notes, and generated content while keeping the UI clean required repeated restructuring.

Accomplishments that we're proud of

  • ✅ Implemented a working integration of the Chrome Prompt API
  • ✅ Created a polished UI with navigation, session history, and toggled note editing
  • ✅ Designed a clean custom rendering engine for jot-note formatting
  • ✅ Built a standalone mind-map canvas page as a foundation for future visual tools

What we learned

  • How to work with the Chrome Prompt API, including availability checks, session creation, and prompt flows
  • Frontend-only architecture can still support meaningful AI applications
  • The importance of UI simplicity when dealing with multi-step user interactions
  • How challenging mind-map interaction logic (dragging, node creation, linking) can be without dedicated libraries
  • How to balance ambition with the time constraints of a hackathon

What's next for Mint Parchment - Prototyped AI Quick Review Assistant

The next steps focus on turning the prototype into a more complete study platform:

  1. Full mind-map generation Auto-create topic nodes based on LLM summaries Drag-and-drop node editing Relationship detection between subtopics

  2. Cloud backend User accounts & cloud storage Sync notes across devices

  3. Smarter topic recap Session-based review history Personalized concept reinforcement Flashcard generation from jot notes

  4. Polished UI improvements Multi-topic comparison mode

Built With

Share this project:

Updates