Personal productivity · Claude Code

Build a personal meal library with nutrition data and a symptom log

Idea + starter prompt, not run by us yet

For: People managing food-related symptoms who want to spot patterns

Using Claude Code, the poster built a meal library from recipes found in cookbook photos, PDFs and web pages, edited for portions and swaps. The agent fetches nutrition data from the USDA API, stores it, proposes weekly meals and shopping lists, and helps analyse logged digestive and eczema symptoms. Tomatoes were the first hypothesis after two weeks. The poster says it is still early.

Steps

1. Start by adding ten recipes you actually cook, from photos, PDFs or links.
2. Ask Claude Code to fetch nutrition data from a public source and store it with each recipe.
3. Log meals and symptoms the same way each day, so the patterns are comparable.
4. Treat any hypothesis from the logs as something to test with a doctor or dietitian.
5. Enter the shopping list into your store cart yourself. Ordering was not automated in the source.

Starter prompt

Trial first: use [3] sample recipes and [7] made-up days of symptom entries in a folder called [meal-trial]. Run the whole job on that. Do not use my real log yet.

Help me build a personal meal database in this folder. I will add recipes as photos, PDFs and links. For each one, store the ingredients, servings and nutrition data from [public nutrition database], and tell me which source you used. Show me the file layout first. Then help me log symptoms daily in a simple format: [symptoms I track].

This job only reads and writes files in the folder. After two weeks, summarise possible patterns, label each as a guess and not a diagnosis, and suggest what I should discuss with a doctor. Give no medical advice. Stop and wait for my yes before you order any food or supplements. Do not order anything.

Original post: Reddit · found via agentusecases.ai. Written up in our own words. We haven't run it, so test it on made-up data first.