Container Requirements

NeoRun automatically generates Dockerfiles using AI analysis, but understanding the container environment helps you build more reliable deployments.

How It Works

Clone: NeoRun clones your repo into a build container
Analyze: Gemini AI scans your project structure, package.json, requirements.txt, Dockerfile, etc.
Plan: AI generates build steps (install deps, build, configure entry point)
Build: Docker image is built layer by layer with caching
Verify: Health checks confirm the service is responding on the expected port

Port Exposure

NeoRun looks for the primary port your application listens on. Common defaults:

Framework	Default Port
Next.js	3000
React (Vite)	5173
FastAPI / Uvicorn	8000
Flask	5000
Express.js	3000
Django	8000
Streamlit	8501
Gradio	7860
Jupyter	8888

Tip: Set the PORT environment variable or bind to 0.0.0.0 — NeoRun injects the PORT env var automatically.

Resource Limits

Resource	CPU Container	GPU Container
Memory	4 GB	8 GB
CPU	2 cores	4 cores
Shared Memory	Default	2 GB
Disk	Build-time only	Build-time only

Working Directory

All containers use /app as the working directory. Your repo is cloned into /app.

Environment Variables

NeoRun injects these environment variables automatically:


PORT=<detected_port>
GIT_TERMINAL_PROMPT=0
DEBIAN_FRONTEND=noninteractive

User Secrets

You can add custom environment variables (API keys, database URLs) via:

Dashboard: Settings → Secrets
CLI: Deploy-NeoRunProject -Secrets @{MY_KEY="value"}

Secrets are stored encrypted in Supabase Vault and injected at container runtime.

Custom Dockerfile

If your repo contains a Dockerfile, NeoRun will detect and use it directly — skipping AI analysis. This gives you full control over the build process.

Requirements for custom Dockerfiles:

Expose at least one port
Use CMD or ENTRYPOINT for the main process
Support the PORT environment variable if possible

Supported Languages & Frameworks

NeoRun’s AI agents can analyze and build most common stacks:

JavaScript/TypeScript: Node.js, Next.js, React, Vue, Angular, Svelte, Express, Fastify
Python: FastAPI, Flask, Django, Streamlit, Gradio, Jupyter, PyTorch, TensorFlow
Go: Standard library, Gin, Echo, Fiber
Rust: Cargo projects, Actix, Axum, Rocket
Java/Kotlin: Spring Boot, Gradle, Maven
Ruby: Rails, Sinatra
PHP: Laravel, Symfony

Network Isolation

Each pod runs in its own Docker network. Containers from different users/pods cannot communicate with each other. Only explicitly mapped ports are accessible through the Traefik reverse proxy.