Chat AI Compiler container || Cloudflare tunnel container || MongoDB inside

A multi-container application that combines:

1. Chat AI Compiler – A frontend interface to query multiple AI model endpoints concurrently, view responses in grid/list layouts, and export results to CSV.
2. Cloudflare tunnel – a tunnel to push http://0.0.0.0:3000 to public. (optional -> use dev mode to disable)
3. MongoDB – a non-relational db to store stuff, using moongoose to interface with.
4. DEV mode vs PROD mode - in DEV mode, HTTP works for logins; in PROD mode you can only use tunnel (HTTPS) to login

DEV_Log

• force dev mode ontop of PROD -> use http://localhost:3000/

  docker-compose -f docker-compose.yml -f docker-compose.dev.yml up --build

-USE THHIS? no one knows what this does, but it does do something

docker-compose -f docker-compose.yml -f docker-compose.dev.yml up --build --force-recreate

• production use -> use tunnel URL, do not use localhost, logins not working

  docker-compose up --build

  docker-compose up --build  --no-cache

• MongoDB; testing steps below

  docker exec -it mongo-db mongosh

  use chatai

  show collections

  db.chatsessions.find().pretty()

  db.users.find().pretty()

• Dev HELP (run these to fix things)

• 1.commands to clear up docker issues

  docker-compose down --volumes --remove-orphans

• 2.removes ALL build cache

  docker builder prune --all --force

• 3.removes unused volumes not currently attached to containers

  docker volume prune --force

• 4.removes unused images (not just dangling)

  docker image prune --all --force

Features

• Select from a variety of AI models and get responses forEach looped.
• Toggle between grid and list views for responses.
• Export chat history and responses to CSV.
• Expose the web UI via Cloudflare Tunnel for secure public access.

Prerequisites

• Docker Desktop (including Docker Compose)
• A Groq API key (set in .env as GROQ_API_KEY)
• SESSION_SECRET='a-password-you-set-for-privacy/encryption' (set in .env)
• (Optional) A Cloudflare Tunnel token (set in .env as CLOUDFLARED_TOKEN)

Getting Started

1. Clone the repository

   git clone https://github.com/davidsilb/chat-app.git
   cd chat-app

2. Create your .env file | only these three NEEDED

   MONGO_URI=mongodb://mongo:27017/chatai
   SESSION_SECRET=you_can_put_anything_in_here
   GROQ_API_KEY=your_groq_api_key_here

3. Build and run all services

   docker-compose up --build

   • The web service (port 3000) serves the frontend and backend API.
   • The cloudflared service tunnels your web UI to a public HTTPS URL.

4. Access the application

   • Visit http://localhost:3000 locally.
   • Or use the Cloudflare Tunnel URL printed in the cloudflared logs/dashboard.

Project Structure

.
├── docker-compose.yml             # Docker Compose setup (app+MongoDB+cloudF)
├── docker-compose.dev.yml         # Docker Compose dev wrapper (app+MongoDB+bb)
├── README.md                      # YOU ARE HERE
├── LICENSE                        # MIT
├── .gitignore                     # Ignore .env
├── .env                           # YOU MAKE (not committed)
├── .env.example                   # A template for your personal .env setup
└── web/
    ├── Dockerfile                 # Dockerfile that will build the webapp
    ├── package.json               # Express server dependencies
    ├── server.js                  # Express app main file
    ├── public/
    │   ├── dashboard.html         # Dashboard page (semi-protected)
    │   ├── index.html             # Main chat compiler UI using groqHandler.js
    │   ├── login.html             # Login page
    │   ├── register.html          # Register page
    │   └── logos/                 # should be on a CDN 4real
    │       ├── cute_logo.png      # 🐱
    │       ├── logo1.jpg          # 🤪
    │       ├── logo2.jpg          # 🤪
    │       ├── whitelogo.jpg      # 🤪
    │       └── whitelogo2.jpg     # 🤪
    ├── middleware/
    │   └── isAuthenticated.js     # Check for who is login, knock knock
    ├── mongo/
    │   └── ChatSession.js         # Mongoose schema (store chats)
    ├── protected/
    │   └── searchpage.html        # Search for stuff, if logged in
    ├── models/
    │   └── User.js                # Mongoose schema for users (auth)
    └── routes/
        ├── addTagToResponse.js    # tag msgs func, implemented in index.html
        ├── exportTxt.js           # Export chat history as .txt files
        ├── groqHandler.js         # Handle groq calls
        └── searchFuntion.js       # search DB of logged in, highlights

Frontend Usage

1. Select one or more AI services.
2. Enter your message and click Send.
3. View responses in grid or list layout.
4. Export the conversation and responses to CSV at any time.
5. Leaderboard (beta, unfinished) - index.html contained, session contained.
6. Basic user chat search on searchpage.html - locked to users only.

License

This project is licensed under the MIT License.