All products

Track grocery prices from receipts with AI

Read item prices from your receipt photos, correct unclear product names and build a local grocery price history.

Track grocery prices from receipts with AI when the till prints a short code that makes sense in the shop and means nothing a week later. In a Meal OS workspace, meal-receipt reads the photos, groups pages from the same purchase and asks you to check its ingredient matches. A corrected abbreviation can help with the next receipt from the same shop.

Look at the unit before confirming a number. Two tubs for one price and loose produce sold by weight belong in different comparisons. The guide uses Claude vision to read the supplied image, including parts left visible around the items. Crop or cover personal details first if they are unnecessary. The accepted ledger records shopping data and the processed image remains in the project.

What it does

  • Group several receipt photos from the same purchase in page order.
  • Check per-item, multi-buy and weight-based prices separately.
  • Review uncertain matches before adding them to the history.
  • Keep confirmed abbreviations for later receipts.
  • Find processed images in a separate folder without losing the originals.

Install and use

  1. Use a Meal OS project with its ingredient reference and a Claude environment that can read images.
  2. Supply a redacted receipt or a set of pages and identify which belong together.
  3. Check the matched items and units, correct any doubtful lines and confirm the update.

FAQ

Source

Made by Scott Curry (scurry). Original skill: meal-receipt, published on GitHub under the MIT licence.

View on GitHub