Run with Docker Compose 🚀🧑‍💻

Setup CAIPE to run in a docker environment on a latop or a virtual machine like EC2 instance.

Clone the repository

git clone https://github.com/cnoe-io/ai-platform-engineering.git
cd ai-platform-engineering

Configure environment variables
```
cp .env.example .env
```
Update .env with your configuration. 📚 See the Setup LLM Providers for more details.

Configure A2A Authentication (Optional)

The A2A protocol supports two authentication methods. Choose one based on your security requirements:

Option A: OAuth2 Authentication (Recommended for Production)

Add the following to your .env file:

A2A_AUTH_OAUTH2=true
JWKS_URI=https://your-identity-provider.com/.well-known/jwks.json
AUDIENCE=your-audience
ISSUER=https://your-identity-provider.com
OAUTH2_CLIENT_ID=your-client-id

Getting OAuth2 Tokens:

Use the provided utility to obtain OAuth2 JWT tokens:

# Add these additional environment variables for token generation
OAUTH2_CLIENT_SECRET=your-client-secret
TOKEN_ENDPOINT=https://your-identity-provider.com/oauth/token

# Run the utility to get a token
python ai_platform_engineering/utils/oauth/get_oauth_jwt_token.py

Local Development with Keycloak:

For local development, you can run a Keycloak OAuth server:

# Start local Keycloak server
cd deploy/keycloak
docker compose up

Then configure your environment:

A2A_AUTH_OAUTH2=true
JWKS_URI=http://localhost:7080/realms/caipe/protocol/openid-connect/certs
AUDIENCE=caipe
ISSUER=http://localhost:7080/realms/caipe
OAUTH2_CLIENT_ID=caipe-cli
OAUTH2_CLIENT_SECRET=your-client-secret-from-keycloak
TOKEN_ENDPOINT=http://localhost:7080/realms/caipe/protocol/openid-connect/token

Keycloak Setup Steps:

Access Keycloak admin console at http://localhost:7080
Login with admin/admin
Switch to the caipe realm
Create a new client called caipe-cli
Copy the client secret and use it in your environment variables

Generate JWT Token with Keycloak:

After setting up Keycloak, generate a JWT token using your client credentials:

export OAUTH2_CLIENT_ID=caipe-cli
export OAUTH2_CLIENT_SECRET=<YOUR CLIENT SECRET>  # randomly generated from Keycloak
export TOKEN_ENDPOINT=http://localhost:7080/realms/caipe/protocol/openid-connect/token

python ai_platform_engineering/utils/oauth/get_oauth_jwt_token.py

Option B: Shared Key Authentication (For Development/Testing)

Add the following to your .env file:

A2A_AUTH_SHARED_KEY=your-secret-key

Note: If neither authentication method is enabled, the A2A agent will run without authentication. This is not recommended for production environments.

🏁 Getting Started

Launch with Docker Compose (AI Platform Engineer)
```
docker compose up
```
Launch with Docker Compose (AI Incident Engineer)
```
docker compose -f docker-compose.incident_engineer.yaml up
```

Connect to the A2A agent (host network)

docker run -it --network=host ghcr.io/cnoe-io/agent-chat-cli:stable

Or, clone and run the chat client:

uvx https://github.com/cnoe-io/agent-chat-cli.git <a2a|mcp>

[Optional] Connect to A2A Agent via backstage agent-forge plug-in

# Once the container is started, open agent-forge in browser (in test mode)
https://localhost:3000

🛠️ For local development setup, check out the Local Development Guide.

🏁 Getting Started​

🏁 Getting Started