What is this page for?

Receipt Payload Parser

Extract items via Computer Vision and sync structures directly to Digital Pantry Nodes.

Receipt Ingestion

Database Target Node

Manual Asset Injection

Append nodes that circumvented CV parsing.

Synchronization Relay

0 Rows read
0 Manual Add
0 Total Nodes
$0.00 Tax Split
$0.00 Tip Split
🤖 What this page is for

Photograph a grocery receipt and the page reads off each line item with a confidence score, you check the rows, then save them into one pantry — Manhattan, Beach Haven, Mountain Lakes or the closet. You come here to stock a pantry from a paper receipt without typing every item, and to correct what the reader made of it before anything is written: each row's name, category, quantity and price can be edited, corrected rows are marked, and the commit sends what the panel holds rather than the untouched model output.

Technical Infrastructure

Vision model (configured)

The image goes to the vision model configured for this app (VISION_MODEL, default gemini-3.1-flash-lite) and the reading is returned with its provenance — the model that read it and how many OCR candidates were compared.

Relational Sync

Entities are bound to their respective node (`templates/food_storage.html` or property matrices) via the PostgreSQL async engine managed implicitly inside `data/db.py`.