Deploy

Beta Test

Deploy your ConnectOnion agents to production. Docker, GCP, AWS - your choice.

Why deploy? Run agents 24/7, scale horizontally, integrate with existing infrastructure. Your agents become production services.

Two Deployment Options

co deploy (Easiest)

Quick deployment to ConnectOnion Cloud. Managed hosting, no infrastructure to manage.

One command deployment
Automatic HTTPS
Re-deploy updates same URL

Self-Host

Full control with your own infrastructure. Use Docker, GCP, AWS, or any VPS.

Full control
Custom domains
Compliance requirements

60-Second Quick Start

Deploy your agent with the CLI - one command:

Terminal

# Create an agent project
co create my-agent

# Navigate to the project
cd my-agent

# Deploy to ConnectOnion Cloud
co deploy

Python REPL

Interactive

Deploying to ConnectOnion Cloud...

Project: my-agent

Secrets: 3 keys

Uploading...

Building...

Deployed!

Agent URL: https://my-agent-abc123.agents.openonion.ai

co deploy Requirements

Git repository with committed code

.co/config.toml (created by co create or co init)

Authenticated (co auth)

Configuration

.co/config.toml

# .co/config.toml
[project]
name = "my-agent"
secrets = ".env"

[deploy]
entrypoint = "agent.py"

Secrets

Secrets from .env are securely passed to your agent:

.env

# .env
OPENAI_API_KEY=sk-xxx
DATABASE_URL=postgres://...

Note: URL format is {project_name}-{your_address[:10]}.agents.openonion.ai. Re-deploying updates the same URL.

Self-Host with host()

Deploy to your own infrastructure using host():

agent.py

# agent.py
from connectonion import Agent, host

agent = Agent("my-agent", tools=[my_tool])

# Export ASGI app for uvicorn/gunicorn
app = host.app(agent)

if __name__ == "__main__":
    host(agent)

Run with uvicorn/gunicorn

code

# Direct
python agent.py

# Uvicorn
uvicorn agent:app --workers 4

# Gunicorn
gunicorn agent:app -w 4 -k uvicorn.workers.UvicornWorker

For full API reference, see host() documentation.

Docker Deployment

The simplest way to deploy - works anywhere Docker runs:

1. Create Your Agent

agent.py

# agent.py
from connectonion import Agent, host

def search(query: str) -> str:
    """Search for information."""
    return f"Results for: {query}"

agent = Agent(
    name="my-agent",
    tools=[search],
    system_prompt="You are a helpful assistant."
)

# Host the agent
host(agent)

2. Dockerfile

Dockerfile

FROM python:3.11-slim

WORKDIR /app

# Install dependencies
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

# Copy agent code
COPY agent.py .

# Run the agent
CMD ["python", "agent.py"]

3. Build & Run

Terminal

# Build the image
docker build -t my-agent .

# Run with API key
docker run -d \
  -e OPENAI_API_KEY=$OPENAI_API_KEY \
  --name my-agent \
  my-agent

Python REPL

Interactive

Building image...

Successfully built a1b2c3d4e5f6

Successfully tagged my-agent:latest

Container started: my-agent

Agent serving at: 0x3d4017c3e843895a...

Deploy to Google Cloud Run

Serverless deployment with automatic scaling:

1. Build & Push to Container Registry

Terminal

# Authenticate with GCP
gcloud auth login

# Set project
gcloud config set project YOUR_PROJECT_ID

# Build and push
gcloud builds submit --tag gcr.io/YOUR_PROJECT_ID/my-agent

2. Deploy to Cloud Run

Terminal

gcloud run deploy my-agent \
  --image gcr.io/YOUR_PROJECT_ID/my-agent \
  --platform managed \
  --region us-central1 \
  --set-env-vars "OPENAI_API_KEY=sk-..." \
  --allow-unauthenticated

Python REPL

Interactive

Deploying container to Cloud Run service [my-agent]...

Done.

Service URL: https://my-agent-abc123-uc.a.run.app

Agent address: 0x3d4017c3e843895a92b70aa74d1b7ebc...

Security Note

Use Secret Manager for API keys in production instead of environment variables.

Deploy to AWS

Multiple options depending on your needs:

EC2 / Lightsail

Simple VPS deployment. Best for always-on agents.

Full control over environment
Predictable pricing
Easy SSH access

ECS / Fargate

Container orchestration. Best for scaling.

Auto-scaling built in
Load balancing
Rolling deployments

EC2 Quick Deploy

Terminal

# SSH to your EC2 instance
ssh -i your-key.pem ubuntu@your-instance-ip

# Install dependencies
sudo apt update && sudo apt install -y python3-pip docker.io

# Clone and run
git clone https://github.com/your/agent-repo.git
cd agent-repo
docker build -t my-agent .
docker run -d -e OPENAI_API_KEY=$OPENAI_API_KEY my-agent

Environment Variables

Configure your agent for different environments:

Variable	Description	Required
OPENAI_API_KEY	OpenAI API key for GPT models	Yes*
ANTHROPIC_API_KEY	Anthropic API key for Claude	Optional
GOOGLE_API_KEY	Google API key for Gemini	Optional
RELAY_URL	Custom relay server URL	Optional
LOG_LEVEL	Logging verbosity (DEBUG, INFO, WARN)	Optional

* Or use ConnectOnion managed keys with co auth - no API keys needed!

Best Practices

Security

Use secret managers for API keys
Never commit .co/ or .env
Use non-root container users
Enable HTTPS for all endpoints

Reliability

Add health checks to containers
Set up automatic restarts
Configure logging and monitoring
Use persistent volumes for keys

Ready to Deploy?

Your agents are production-ready. Ship them!

host() Reference Connect to Agents

CLI Reference

Agent