ValueMunch
A mobile friendly cooking prototype that starts with ingredients already in the pantry and leads to meal ideas, cooking steps, and simulated rewards.
1st Place · Build with TRAE · 31 January 2026
From Pantry to Meal
We built ValueMunch around the question of what to cook with ingredients already available. Users photograph their pantry ingredients or type them in, review what was recognized, and choose a recipe. Cooking steps and timers carry the chosen meal through to completion.
Completing a meal reveals impact information and a simulated reward. These are prototype features, not evidence of measured waste reduction or real reward payouts. Sample or cached recipes keep the cooking flow available when AI processing is unavailable.
The former Firebase demo is retired; source and walkthrough material remain available. A separate pantry app design explores a smaller version starting with three ingredients and an optional cuisine twist.
We built ValueMunch in four hours at ByteDance using TRAE for planning, Google AI Studio for design experiments, and Firebase for deployment.
Carrying Pantry Ingredients Through a Cooking Session
Connecting Discovery to Cooking Progress
The expanded React application separates camera capture, recipe discovery, guided cooking, and impact into coordinated views. Ingredient review comes before recipe selection; the selected recipe then supplies the steps and timers used during cooking. The demo architecture assigns view coordination to the main application and Firebase operations to service functions, connecting cooking history and impact data to the wider experience.
Pantry Recipe App
The separate pantry PWA starts with three ingredients and an optional cuisine twist. Its product requirements specify a structured recipe response containing ingredients, instructions, preparation and cooking times, and servings. That gives the interface defined fields to display rather than a single block of generated prose.
This smaller design also specifies an IndexedDB cache of the last five generated recipes. When the network is unavailable, it presents those saved recipes with an offline indicator instead of requiring a new generation request.