AI-Powered Health & Fitness Companion

A Flask web application that helps users track their BMI, generate personalized meal and workout plans, and interact with an AI health coach. The application stores user data in a local SQLite database and uses OpenRouter (via ai_caller.py) to produce AI-generated plans and chat responses.

Key features

• User authentication (signup/login/logout) with hashed passwords
• BMI calculator with history and categorized tips
• Persistent user preferences (diet, allergies, goals, activity level, target weight, etc.)
• AI-generated weekly meal plans (stored per-week)
• AI-generated weekly workout plans (stored per-week)
• Chat interface with AI assistant that uses recent chat history and BMI context
• Progress view that shows BMI trends for the user and accepted friends
• Friend request system (send/accept/reject) to share progress
• Toggle-tracking for completed meal/workout items

Project structure

ai-bmi-calc/
├── app.py                # Flask application and routes
├── ai_caller.py          # Lightweight OpenRouter client wrapper used for AI calls
├── health.db             # SQLite database (created at runtime)
├── requirements.txt      # Python dependencies
├── static/               # CSS, favicon, client assets
└── templates/            # Jinja2 HTML templates (login, signup, dashboard, etc.)

Dependencies

Dependencies are listed in requirements.txt. Key packages:

• Flask
• requests
• cs50 (lightweight SQLite wrapper used here)
• matplotlib (used for chart rendering)

Install them with:

pip install -r requirements.txt

AI backend

This project uses OpenRouter (https://openrouter.ai/) via ai_caller.py to call hosted models (the default code sets the model to a Llama 3 variant). You must provide an OpenRouter API key via an environment variable named OPENROUTER_API_KEY (see notes below). The app prepares system prompts from diet_coach_prompt.txt and workout_coach_prompt.txt and sends user context for plan generation.

Note: older README referenced Ollama — the current code uses OpenRouter. Some legacy Ollama-sidecar code remains but the AI calls are performed by ai_caller.py.

Database

The app uses a local SQLite database file named health.db. On first run the app will create required tables automatically. You can remove health.db to reset data (or use a migration script if you add one).

Run locally

1. Create and activate a virtual environment (recommended):

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

2. Set up required variables:

In app.py:

app.config["SECRET_KEY"] = "qwertyuiopasdfghjklzxcvbnm"  # Use your own secret key here to encrypt the data

In ai_caller.py:

API_KEY = "openrouter_api_key"  # Enter your openrouter API key here

3. Start the app (development):

python app.py

By default the app runs on 127.0.0.1:5501 (see app.run(...) in app.py).

Usage overview

• Signup for a new account and login.
• Set your preferences on the Preferences page (dietary preferences, goals, age, activity level, etc.).
• Use the BMI Calculator to record weight/height entries; BMI entries are saved and used by the AI to tailor meal/workout plans.
• Generate weekly meal and workout plans from the Meal Plan and Workout Plan pages.
• Chat with the AI assistant from the Chat page; messages are saved to history.
• Track progress and optionally connect with friends to compare charts.

Security & production notes

• Do not commit secret keys or API keys. Use environment variables or a secrets manager.
• The default SECRET_KEY in app.py is for development only. Replace it before deploying.
• Consider enabling HTTPS, session protection, rate limiting and proper model usage quotas when exposing the app.

Development notes & troubleshooting

• If AI calls return errors, verify OPENROUTER_API_KEY is set correctly and the chosen model is available on OpenRouter.
• The app expects diet_coach_prompt.txt and workout_coach_prompt.txt to exist in the project root — they provide system-level prompts for plan generation.
• If templates or tables are missing, check runtime errors in the console — the app tries to create missing tables on startup.

Tests

There are no automated tests included. For quick smoke testing: create an account, set preferences, add a BMI entry, and try generating plans and chatting.

License

This project is licensed under the MIT License.