Track pantry stock from receipts and buy groceries
A custom AI system reads receipts, estimates kitchen stock, checks grocery store availability and prices, buys groceries, and suggests recipes.
Done with a custom-built or unnamed agent
- What they did
- User ingests receipts and spending data into a custom service, then exposes a grocery budget envelope plus recipe history and financial data to a stock estimator tool. A second tool connects to local grocery store APIs for nutrition, availability, and prices. Both tools are exposed over a terse MCP surface to an AI harness (ChatGPT plus opencodev2) that produces stock updates, budgeting, and recipe recommendations.
- How it went
- The system genuinely saves the user time and money through ongoing stock estimates and recommendations. No purchase execution or quantitative results are stated.
- Worth knowing
- User offered to open-source the stack if others want it.
Try it yourself with Claude Code
In the folder [path to receipts] I have grocery receipts. Read them, build a running estimate of what's in my kitchen, and save it as [pantry file name]. Then list what I'm likely low on, suggest [number] recipes using what I have, and prepare a shopping list. You are done when the pantry file and shopping list exist; don't buy anything without showing me the list first.
Source: X · Oct 5, 2026
Five of these in your inbox every morning
The best things people got an AI agent to do, each with the prompt to try it.
More like this
- Qwen 3.6 27B
Order groceries online with a local LLM skill
The author built a grocery skill for a local Qwen 3.6 27B model that knows their fridge contents, past orders and real-time stock, and it ordered products for them online.
Find coffee makers under $100 and add options to the cart
Perplexity Comet searched Amazon for coffee makers with a grinder, built a comparison table and added two options to the cart, stopping before purchase.
Find and add high-meat cat food to a basket
A reviewer had Comet's assistant find the cat food with the most meat and add it to the basket, plus some product comparison and form filling, though imperfectly.
Email Houston dealerships to shop for a car
Cowork built a dealership tracking sheet, sent outreach emails, filled web forms and managed replies; a separate 100+ receipt expense job hit limits and moved to Claude Code.
Hand consumer errands to an always-on agent
A journalist tested Dot on ISP signup, an Ikea account and food ordering; it finished only some with human help, blocked by browser checks and payment steps.
Build an e-commerce agent with OpenClaw
Poster spent 100+ hours and 200M+ tokens building an e-commerce workflow agent with OpenClaw; post focuses mostly on the build effort.