BaselHack 2025

This repository contains our work for the BaselHack 2025 event. We selected the Endress+Hauser sponsored challenge focused on developing an application to create surveys, gather people's opinions, and analyze the results with AI.

Project Overview

We have developed the backend with Fastify and the frontend with Flutter. The application allows users to create surveys, distribute them, and analyze the collected data using AI techniques. We also have developed an AI agent to assist users in generating survey questions based on specific topics.

Environment Configuration

Each component has its own .env.example file. Copy the appropriate file next to it, rename it to .env, and fill in the required values before running the app:

• code/.env.example → code/.env: shared variables for Docker Compose (e.g., MISTRAL_API_KEY, default admin credentials, port overrides).
• code/backend/.env.example → code/backend/.env: Fastify API configuration (Mongo URI, JWT expirations/secrets, agent URL, optional default admin fields).
• code/agent/.env.example → code/agent/.env: Python agent settings (Mistral API key, expected JWT issuer/audience, path to the backend public key).

Secrets expected at runtime:

• code/secrets/public_key.pem: backend public key used by the agent.
• code/secrets/private_key.pem: backend private key used to sign requests towards the agent.

Sample .env files with development defaults are already provided (code/.env, code/backend/.env, code/agent/.env). Adjust them as needed or rebuild them from the corresponding *.env.example. Compose variables like DB_URI, AGENT_AUTHENTICATION_PRIVATE_KEY_PATH, or AGENT_AUTHENTICATION_PUBLIC_KEY_PATH let you point services at external infrastructure when required. Once the environment files are in place, you can start the full stack from the code/ directory:

cp code/.env.example code/.env
cp code/backend/.env.example code/backend/.env
cp code/agent/.env.example code/agent/.env
# populate the files with your values, then:
docker compose up --build

Continuous Deployment

Pushes to the main branch can be deployed automatically to a Docker host through the GitHub Actions workflow in .github/workflows/deploy-production.yml. The workflow connects to the server over SSH and runs docker compose up -d --build --remove-orphans inside the code/ directory.

One-time server preparation

1. Install Docker, the Docker Compose plugin, and Git on the target server.
2. Clone this repository to the directory you want to run in production (that directory becomes PROD_APP_DIR).
3. Create the production .env files and the secrets/ folder directly on the server; they stay outside of version control.
4. Ensure the SSH user used for deployment can run Docker commands (e.g., add it to the docker group).

Required GitHub secrets

Add the following secrets (repository or organization level) before enabling the workflow:

• PROD_SSH_HOST: Public hostname or IP of the server.
• PROD_SSH_USER: SSH user with access to the repository directory and Docker.
• PROD_SSH_KEY: Private SSH key (OpenSSH format) matching a deploy/public key on the server.
• PROD_SSH_PORT (optional): SSH port if different from 22.
• PROD_APP_DIR: Absolute path to the repository directory on the server (e.g., /opt/baselhack-2025).

After the secrets are configured, every push to main (or a manual workflow_dispatch) will fetch the latest code on the server and rebuild/restart the stack.

Contributors

• Enzo Lorenzini
• Mael Bertocchi
• Matteo Tutti
• Maxime Entz
• Yann Toison-Chabane
• Romain Schmitz
• Matthias Yildirim
• Matteo Bianchi