How I Track Nutrition with AI

I stopped using apps and just roll with Claude code models now.

The System

One log file per month, with foods filled per day.

On profile json with my essentials (target kcal, weight, etc.).

All goes to Github Repo so I can interact and fill data from Claude code, remote terminal, etc.

## Jan 18

- 6 eggs, avocado, sauerkraut, mozzarella
- 70g dark chocolate
- Whey shake, berries
- Workout: 400kcal gym

One config with my targets:

Calories: 2835 | Protein: 160g | Fat: 100g

That’s it.

How It Works

I type what I ate. The AI adds it to my log, calculates macros, shows where I stand.

Me: add 6 eggs and avocado

Claude: Added. +350 kcal, +21g protein.

Me: summary

Claude:
Calories   [===========>--]  2689 / 3235   83%
Protein    [============>-]   153 / 160g   96%

Rec: cottage cheese closes protein gap

Why It Works

• No friction — type “add eggs” not search → select → adjust serving
• Context-aware — knows my patterns, warns me when fat is high
• Flexible — “150g mozzarella” or “some salad” both work
• Analysis on demand — patterns, gaps, recommendations

What I Learned

• My fat runs high daily (nuts + chocolate + olive oil)
• Protein needs evening catch-up
• Vegetables are my real gap
• Omega-3 low