_Includes

Configuring Helix Apps

Helix Apps are version-controlled configurations for LLM-based applications. Here’s how to configure them:

Basic App Structure

A Helix App consists of:

  1. helix.yaml - Main configuration file
  2. Supporting files (prompts, datasets, etc.)

helix.yaml Example

name: my-awesome-app
version: 1.0.0
description: An example Helix App

assistant:
  model: gpt-4
  system_prompt: "You are a helpful assistant."
  temperature: 0.7

knowledge:
  sources:
    - type: url
      url: https://example.com/docs
      schedule: "0 0 * * *"

integrations:
  - type: api
    name: weather
    url: https://api.weather.com/v1
    auth:
      type: api_key
      key: WEATHER_API_KEY

Deployment

Deploy apps using git:

Configuring Caddy for HTTPS

Caddy can automatically provision and renew TLS certificates. Here’s a basic Caddyfile configuration:

{
  email your-email@example.com
}

helix.yourdomain.com {
  reverse_proxy localhost:8080
}

Installation

Install Caddy on Ubuntu:

sudo apt install -y debian-keyring debian-archive-keyring apt-transport-https
curl -1sLf 'https://dl.cloudsmith.io/public/caddy/stable/gpg.key' | sudo gpg --dearmor -o /usr/share/keyrings/caddy-stable-archive-keyring.gpg
curl -1sLf 'https://dl.cloudsmith.io/public/caddy/stable/debian.deb.txt' | sudo tee /etc/apt/sources.list.d/caddy-stable.list
sudo apt update
sudo apt install caddy

Configuration

  1. Edit the Caddyfile:

    sudo nano /etc/caddy/Caddyfile

  2. Replace the contents with your configuration

Configuring External Providers

Helix can integrate with various external providers for enhanced functionality:

OpenAI Integration

To use OpenAI models, add the following to your configuration:

OPENAI_API_KEY=your_openai_api_key
OPENAI_BASE_URL=https://api.openai.com/v1

TogetherAI Integration

For TogetherAI models:

TOGETHER_API_KEY=your_together_api_key

Hugging Face Integration

For Hugging Face models and datasets:

HF_TOKEN=your_huggingface_token

GitHub Integration

To enable GitHub integration for app deployment:

GITHUB_TOKEN=your_github_token
GITHUB_WEBHOOK_SECRET=your_webhook_secret

Database Configuration

For external database connections:

DATABASE_URL=postgresql://user:password@host:port/database

Ensure you restart the Helix services after making configuration changes.

Configuring Runners

Runners are responsible for executing the actual inference tasks in Helix. Here’s how to configure them:

Local Runner Configuration

For a local runner, ensure the following in your .env file:

RUNNER_ENABLED=true
RUNNER_NAME=local-runner
RUNNER_GPU_ENABLED=true

Start the runner with:

./runner.sh

Remote Runner Configuration

To attach a remote runner to your control plane:

  1. On the control plane, generate a runner token:

    docker exec helix-controlplane helixctl generate-runner-token

  2. On the runner machine, configure the .env file:

Security Configuration

Helix comes with reasonable security defaults, but you should consider the following for production deployments:

Network Security

  • Restrict access to the API port (default 8080) using firewall rules
  • Use HTTPS in production with valid certificates
  • Consider placing Helix behind a reverse proxy like Nginx or Caddy

Authentication

  • Change default passwords immediately after installation
  • Use strong, unique passwords for all accounts
  • Enable two-factor authentication where possible
  • Regularly rotate API keys and tokens

Data Protection

  • Regularly backup your data and test restoration procedures
  • Encrypt sensitive data at rest
  • Monitor access logs for suspicious activity
  • Apply security updates promptly

Container Security

  • Run containers with non-root users where possible
  • Regularly update base images
  • Scan images for vulnerabilities
  • Limit container privileges and capabilities

For more detailed security guidance, refer to the Security Best Practices documentation.

Control Plane Helm Chart Configuration

To install the Helix Control Plane using Helm, use the following configuration:

# values.yaml
controlplane:
  image:
    repository: helixml/controlplane
    tag: "latest"
  
  service:
    type: ClusterIP
    port: 8080
  
  ingress:
    enabled: true
    hosts:
      - helix.yourdomain.com
    tls:
      - hosts:
          - helix.yourdomain.com
        secretName: helix-tls
  
  postgresql:
    enabled: true
    auth:
      postgresPassword: "securePassword"
      password: "securePassword"
  
  redis:
    enabled: true

Install with:

helm repo add helix https://charts.helixml.tech
helm install controlplane helix/controlplane -f values.yaml

Keycloak Helm Chart Configuration

To install Keycloak using Helm, use the following configuration:

# values.yaml
keycloak:
  image:
    repository: quay.io/keycloak/keycloak
    tag: "21.1"
  
  service:
    type: ClusterIP
    port: 8080
  
  ingress:
    enabled: true
    hosts:
      - keycloak.yourdomain.com
    tls:
      - hosts:
          - keycloak.yourdomain.com
        secretName: keycloak-tls
  
  postgresql:
    enabled: true
    auth:
      postgresPassword: "securePassword"
      password: "securePassword"

Install with:

helm repo add bitnami https://charts.bitnami.com/bitnami
helm install keycloak bitnami/keycloak -f values.yaml

Adding Helm Repositories

Before installing Helm charts, you need to add the appropriate repositories:

# Add Bitnami repository (for Keycloak, PostgreSQL, etc.)
helm repo add bitnami https://charts.bitnami.com/bitnami

# Add ingress-nginx repository
helm repo add ingress-nginx https://kubernetes.github.io/ingress-nginx

# Update all repositories
helm repo update

Verify the repositories are added:

helm repo list

System Requirements

  • Operating System: Linux (Ubuntu 20.04+), macOS (12+), or Windows with WSL2
  • CPU: Minimum 4 cores recommended
  • RAM: Minimum 8GB, 16GB recommended
  • Disk Space: Minimum 20GB free space
  • Docker: Version 20.10 or higher
  • Git: Version 2.20 or higher

For GPU Support

  • NVIDIA GPU: CUDA-compatible GPU with at least 8GB VRAM
  • NVIDIA Drivers: Latest drivers installed
  • NVIDIA Container Toolkit: For Docker GPU support

For Local Development

  • Terminal: Bash or compatible shell
  • Text Editor: VS Code, Sublime Text, or similar
  • Browser: Modern browser (Chrome, Firefox, Safari, Edge)