Xiaoyi Health Nutrition Skill - Intelligent Diet Planning and Nutrition Analysis

Do you also struggle with this?

Every day, you agonize over what to eat, scroll through food delivery apps for half an hour, and end up ordering fried chicken…

To be honest, I used to be in that exact state. I wanted to eat healthier, but faced with a dazzling array of food, I had no clue how to combine them properly. Have you ever calculated how many calories you consume in a day? Is your protein intake sufficient? Which vitamin—A, B, C, D, or E—are you lacking?

Then I discovered this Xiaoyi Health Nutrition Skill, and it felt like finding a treasure! It's an open-source project hidden on a certain code hosting platform (you know which one). Its name is quite long — xiaoyi-claw-omega-final/skills/xiaoyi-health/nutrition — but don't be intimidated by the name; using it is absolutely amazing!

What does this skill do? Simply put, it helps you create a personalized diet plan. You input your basic information — age, gender, height, weight, activity level — and tell it whether you want to lose weight, gain muscle, or control blood sugar. Then it generates a complete daily plan of what and how much to eat. And it's not one of those generic recipes you find everywhere online; it's calculated based on your personal data, very scientific.

Six-layer architecture? Sounds advanced, but it's actually…

Behind it is a six-layer architecture. I think of it like building a house — the foundation is a food database (which stores nutritional data for over 5,000 foods), and on top of that are the calculation engine, rule engine, AI recommendation module, and so on. But as ordinary users, we don't need to worry about that — just know that it works offline too. Yes, you read that right: all data is stored locally, so no privacy concerns. Who dares to casually upload health data to the cloud these days, right?

Easy installation, one command does it:

npx skills add <your-repository-link> --skill xiaoyi-health/nutrition

Prerequisite: You need to have Node.js and npm installed. If you don't, go to the official website to download the LTS version, install it, open the terminal, paste the command above, hit Enter, and wait for it to finish. Then open localhost:3001 in your browser to see the dashboard.

Real-world experience: I lost 10 pounds using it

I'm 175 cm tall, weigh 75 kg, and want to get down to 68 kg. Following the skill's guidance, I set my goals: total daily calories 1800 kcal, protein 90 g, carbs 200 g, fat 50 g. Then it recommended a weekly meal plan — breakfast: oatmeal with milk and eggs; lunch: chicken breast salad; dinner: steamed fish with brown rice. At first, I thought I'd be hungry, but after following it for a week, I found it actually satisfying, because the skill takes satiety index into account.

What surprised me most was the barcode scanning feature. I bought a snack, scanned the barcode with the skill, and instantly the nutrition facts popped up, automatically logging it into my daily intake. No more manual calorie counting! Dealing with those tricky packages that say "per serving 30g" but actually contain 100g per bag — scan it and get the real data.

There's also a weekly diet report. Every Sunday, it sends you a summary, telling you how many calories you've exceeded, which nutrients are insufficient, and even teasing you with comments like "Carb intake is high; suggest reducing refined grains." The last time it said "Calcium intake insufficient; suggest drinking 300ml of milk daily," it hit me hard — I really hadn't been drinking milk recently.

Hidden features you might not know about

• Multi-user mode: Family members can use different accounts, each with independent data. I set it up for my parents too; they have high blood pressure, and the skill automatically recommends low-sodium foods.
• Voice interaction: Supports voice input. For example, speak into the mic "I had a bowl of Lanzhou pulled noodles for lunch," and it automatically recognizes and estimates the nutrition. But your Mandarin needs to be standard — dialect might cause errors (don't ask how I know).
• Data export: You can export data as an Excel spreadsheet for your nutritionist. Last time I had a check-up, the doctor asked for my diet records, and I simply printed the report. The doctor even praised it as professional.
• Custom foods: If a food isn't in the database, you can manually add it by entering the name and nutritional content per 100g, and next time it will recognize it.

Precautions and pitfalls to avoid

1. Don't rely entirely on its recommendations. Although the algorithm is powerful, individual differences exist. For example, I found that eating the recommended sweet potatoes gave me bloating. Then I added "intolerance food" in the preference settings, and it stopped recommending them.
2. Calorie estimates may have errors. Especially for Chinese dishes — a stir-fry with lots of oil and salt can only be estimated based on standard preparation. I recommend using a food scale at the beginning to weigh your food; accurate logging helps calibrate the system.
3. Don't forget to update. The first version I installed had only 3,000 foods in the database; after updating to the latest version, it added over 2,000 more. Update command: npx skills update xiaoyi-health/nutrition.
4. If you encounter errors, it's often due to a port being occupied. You can add a parameter in the startup command to specify a different port, for example, by modifying the configuration file.

Is the Xiaoyi Health Nutrition Skill worth installing?

I'd say, for anyone who cares about their diet, it's absolutely the GOAT. Especially for those who want to lose weight but hate counting calories, or those with chronic diseases needing dietary control. And it's open-source, so you can modify the code yourself — for example, adding a "regional cuisine" feature. I'm planning to write a Cantonese dim sum module myself.

In short, installing it is like having a nutritionist in your pocket, available 24/7, and it won't scold you. I've been using it for two months, and my body fat percentage dropped from 22% to 19%. Now every day when I open it, it feels like playing a growth game — seeing the nutrition achievement bars turn green one by one gives me a great sense of accomplishment.

Alright, I'm off to scan my lunch takeout. Give it a try too!