The Grass Receipt -- Turn your offline walks into printable physical receipts.

This is a submission for the Hacktoberfest Weekend Challenge: Touch Grass

The Problem: The Screen-Time Paradox

When you spend your days wrestling with complex code, algorithms, and computational physics, your brain hits a wall. Staring at a display longer yields zero marginal productivity. The best debugger is walking away from the screen.

However, standard habit trackers feel like chores. I wanted to build something lighthearted, fast, and completely offline that rewards offline time with physical art rather than another screen notification.

What I Built: The Grass Receipt

The Grass Receipt is a local app that turns your offline walks and screen-free breaks into a physical, vintage store receipt.

You input your micro-stats (minutes outside, burnout level, and screen-free time). TabPFN instantly evaluates your metrics to calculate a "Mental Reset Score," and a local Gemma 2 model formats a witty, monospaced "receipt" of your outdoor experience. You hit print, grab your glossy paper, and tape it straight to your wall.

How I Built It

• TabPFN (Tabular Foundation Model): Evaluates the user's offline habit data to generate the mental reset classification probability in milliseconds without tedious manual training loops.
• Gemma 2 (gemma2:2b via Ollama): Acts as the local text generation engine, turning raw scores into aesthetic, structured store receipts.
• Sentry Agent Tracing: Tracks transaction performance across the TabPFN calculation and LLM inference pipeline.
• DevRelay Integration: Session telemetry tracking the development and building process.

Why Open Source & Local AI Matter

An app designed to disconnect you from screens shouldn't rely on cloud subscription tokens or active Wi-Fi connections when you're sitting on a bench outside. By utilizing a zero-shot tabular model like TabPFN and a fast local weight like Gemma 2, everything runs locally on machine hardware in milliseconds.

The Handover

Take a 30-minute walk, drop your stats into the app, print the receipt, and turn off your computer.

Prize Categories Entered

• Best Use of TabPFN: Predicting mental reset scores from tabular offline habit data.
• Best Use of Gemma: Generating localized offline receipt text via Ollama.
• Best Use of Sentry Agent Tracing: Execution telemetry tracking.
• Best Use of DevRelay: Session documentation.