crewAI/docs/how-to/open-source-deployment.mdx

# CrewAI Open-Source Deployment

CrewAI Open-Source Deployment provides a simple way to containerize and deploy CrewAI workflows without requiring a CrewAI+ account.

## Installation

```bash
pip install crewai
```

## Quick Start

### 1. Create a deployment configuration file

Create a file named `deployment.yaml`:

```yaml
# CrewAI Deployment Configuration
name: my-crewai-app
port: 8000
host: 127.0.0.1  # Default to localhost for security

# Crews configuration
crews:
  - name: research_crew
    module_path: ./crews/research_crew.py
    class_name: ResearchCrew
  
  - name: analysis_crew
    module_path: ./crews/analysis_crew.py
    class_name: AnalysisCrew

# Flows configuration
flows:
  - name: data_processing_flow
    module_path: ./flows/data_processing_flow.py
    class_name: DataProcessingFlow
  
  - name: reporting_flow
    module_path: ./flows/reporting_flow.py
    class_name: ReportingFlow

# Additional configuration
environment:
  - OPENAI_API_KEY=${OPENAI_API_KEY}
  - CREWAI_LOG_LEVEL=INFO
```

### 2. Create a deployment

```bash
crewai os-deploy create deployment.yaml
```

This command will:
- Create a deployment directory at `./deployments/{name}`
- Copy your crew and flow modules to the deployment directory
- Generate Docker configuration files

### 3. Build the Docker image

```bash
crewai os-deploy build my-crewai-app
```

### 4. Start the deployment

```bash
crewai os-deploy start my-crewai-app
```

### 5. Use the API

The API will be available at http://localhost:8000 with the following endpoints:

- `GET /`: Get status and list of available crews and flows
- `POST /run/crew/{crew_name}`: Run a crew with specified inputs
- `POST /run/flow/{flow_name}`: Run a flow with specified inputs

Example request:

```bash
curl -X POST http://localhost:8000/run/crew/research_crew \
  -H "Content-Type: application/json" \
  -d '{"inputs": {"topic": "AI research"}}'
```

### 6. View logs

```bash
crewai os-deploy logs my-crewai-app
```

### 7. Stop the deployment

```bash
crewai os-deploy stop my-crewai-app
```

## API Reference

### GET /

Returns the status of the deployment and lists available crews and flows.

**Response:**
```json
{
  "status": "running",
  "crews": ["research_crew", "analysis_crew"],
  "flows": ["data_processing_flow", "reporting_flow"]
}
```

### POST /run/crew/{crew_name}

Runs a crew with the specified inputs.

**Request Body:**
```json
{
  "inputs": {
    "key1": "value1",
    "key2": "value2"
  }
}
```

**Response:**
```json
{
  "result": {
    "raw": "Crew execution result"
  }
}
```

### POST /run/flow/{flow_name}

Runs a flow with the specified inputs.

**Request Body:**
```json
{
  "inputs": {
    "key1": "value1",
    "key2": "value2"
  }
}
```

**Response:**
```json
{
  "result": {
    "value": "Flow execution result"
  }
}
```

## CLI Reference

| Command | Description |
|---------|-------------|
| `crewai os-deploy create <config_path>` | Create a new deployment from a configuration file |
| `crewai os-deploy build <deployment_name>` | Build Docker image for deployment |
| `crewai os-deploy start <deployment_name>` | Start a deployment |
| `crewai os-deploy stop <deployment_name>` | Stop a deployment |
| `crewai os-deploy logs <deployment_name>` | Show logs for a deployment |

## Comparison with CrewAI+

| Feature | Open-Source Deployment | CrewAI+ |
|---------|------------------------|---------|
| Requires CrewAI+ account | No | Yes |
| Self-hosted | Yes | No |
| Managed infrastructure | No | Yes |
| Scaling | Manual | Automatic |
| Monitoring | Basic logs | Advanced monitoring |