Nutrition Advisor: AI-Powered Dietary Analysis & Personalized Meal Planning

Preface: A "Nutritionist" Powered by Large Language Models to Help You Watch Your Diet

To be honestI'm the kind of person who is constantly anxious about what to eatI want to lose weightbut I'm afraid of going hungryI want to gain musclebut I have no idea what to eatI've tried some calorie counting apps beforebut after two days I got annoyed and gave upThenwhile browsing GitHubI stumbled upon the health-llm-driven project by the developer itsosowhich includes a skill called nutrition-advisorIt immediately caught my interestThis thing is built on a large language modelso it feels much smarter than those rigid databases.

Simply putthis skill acts as a "large-model nutrition consultant." You just tell it your situation—for example"I'm 28 years oldmaleweigh 75kgwork in an officejog occasionallyand want to lose fat"—and it will give you a complete dietary plan based on the information you provideMoreoverit doesn't just offer generic advice; it can engage in conversation with you and answer all sorts of specific questionslike "I had a spicy hot pot for lunch todaydid I exceed my calorie limit?" or "Recommend some high-proteinlow-fat dinner recipes." It's like having a personal nutritionist available online anytime.

Installation and Experience: One Command to Set It Upbut Pay Attention to Details

The installation process is very straightforwardSince this is a Skill projectyou need to run it on the platformIf you haven't installedyetyou'll need to install it first (you can search for the official website of )Then open a terminal and type the following command:

npx skills add https://github.com/itsoso/health-llm-driven --skill nutrition-advisor

After running itit will automatically download the repository and register the skillIt took me about twenty to thirty secondsbut it might take longer if your network is slowOnce the download is completeopenand you'll see the Nutrition Advisor card in the skill listClick "Run," and a chat interface will pop up—this is where you'll talk to the AI nutritionist in the future.

A small reminder: If you're using Windowsit's recommended to run the terminal as an administrator; otherwiseyou might run into permission issuesAlsomake sure your Node.js version is greater than 16or npx might throw an errorBy the waythis skill doesn't require an internet connection to work—all models and data are loaded locallyso your privacy is well protected.

Live Test: Let the AI Create a Fat-Loss Breakfast Plan for Me

I started with a simple testIn the chat boxI typed: "Generate a breakfast suitable for fat losswith calories控制在 within 400 kcaland provide protein and dietary fiber."

It replied quicklylisting three options:

• Two slices of whole wheat bread + two boiled eggs + one cup of unsweetened soy milk + a small handful of blueberries
• 30g oatmeal + 200ml milk + 10g nuts + half an apple
• Egg pancake (with whole wheat flour) + lettuce salad (low oil) + unsweetened yogurt

Each option indicated the calorie count and nutritional contentI chose the first optionand it further asked: "Do you have any history of egg allergy or high cholesterol?"That was quite thoughtfulI honestly answered noand it confirmed the planThen I asked: "What about lunch?" It said: "Based on the 300 kcal you've already consumed at breakfastI recommend keeping lunch to around 500 kcalSuggested meal: 150g chicken breast + 100g brown rice + 200g broccoli." This chain of reasoning felt way better than my own random guesses.

I also tested its depth of knowledgeFor exampleI asked: "Can diabetic patients eat watermelon?" It not only gave cautious advice (watch the portionpair with meals) but also explained the glycemic index and glycemic load of watermelonciting nutritional guidelinesThis wasn't a simple Q&A; behind it was the LLM's understanding of medical knowledge.

Feature Highlights: What Makes It So "Smart"?

• Personalized modeling: Not generic advicebut suggestions generated based on your body dataactivity leveland goalsThe more detailed your descriptionthe more accurate it becomes.
• Conversational interaction: You can continue asking follow-up questions without losing contextFor instanceif you ask about breakfast and then lunchit will incorporate the previous information.
• Real-time calculation: Enter a food name and portion sizeand it can instantly estimate calories and nutrientsVery practical for takeout lovers—like "About how many calories are in a bowl of Lanzhou beef noodles?" It gives a reliable estimate and even reminds you that "the broth is high in fatso it's better not to drink too much."
• Food substitution suggestions: "I don't like chicken breastWhat can I substitute?" It will recommend protein sources like fishshrimptofuetcand adjust the entire recipe accordingly.
• Multi-language support: I asked in Chinese and got a Chinese answer; switching to Englishit seamlessly switched as wellThis is very friendly for bilingual users.

Notes and Suggestions

Although this skill is powerfulit is not a real doctor or registered dietitianSo keep in mind:

• For diseases requiring strict medical dietary management (e.gkidney diseasediabetes requiring insulin therapy)it is recommended to consult a professionalAI advice should only be used as a reference.
• The food database may be incompleteespecially for nutritional data on some local specialty snacksIt will estimate based on reasoningbut it's not 100% accurate.
• If it makes a mistakeyou can point it outsuch as "You said chicken breast 300g last timenow you say 250g?" It has a correction mechanismIn generalits common knowledge is sufficient.
• Regularly updating the skill can yield better performanceUse npx skills update nutrition-advisor to pull the latest code.

Summary of My Experience

This nutrition advisor skill has really helped me a lotNow I no longer stress about what to eat for breakfast—I just ask it directlyand it's fast and scientificAnd its