This guide covers all configuration options available in Prometheus Swarm.

Environment Variables

Prometheus Swarm uses environment variables for sensitive configuration. You can set these using a .env file or system environment variables.

Required Variables

# AI Model API Keys
ANTHROPIC_API_KEY=your_anthropic_key
OPENAI_API_KEY=your_openai_key
GOOGLE_API_KEY=your_google_key

# Database Configuration (optional)
DATABASE_URL=your_database_url

WorkflowConfig

The WorkflowConfig class is the main configuration object for workflows:

from prometheus_swarm.types import WorkflowConfig

config = WorkflowConfig(
    model="claude-3-opus-20240229",  # AI model to use
    temperature=0.7,                  # Response randomness (0.0-1.0)
    max_tokens=1000,                 # Maximum response length
    database_url="sqlite:///db.sqlite3",  # Database connection string
    retry_attempts=3,                # Number of retry attempts
    timeout=30,                      # Request timeout in seconds
)

Available Models

Anthropic Models

• claude-3-opus-20240229
• claude-3-sonnet-20240229
• claude-2.1
• claude-2.0

OpenAI Models

• gpt-4-turbo-preview
• gpt-4
• gpt-3.5-turbo

Google Models

• gemini-pro
• gemini-pro-vision

Database Configuration

SQLite (Default)

from prometheus_swarm.database import Database

db = Database("sqlite:///local.db")

PostgreSQL

db = Database("postgresql://user:password@localhost:5432/dbname")

MongoDB

db = Database("mongodb://localhost:27017/")

Client Configuration

Anthropic Client

from prometheus_swarm.clients import AnthropicClient

client = AnthropicClient(
    model="claude-3-opus-20240229",
    temperature=0.7,
    max_tokens=1000,
)

OpenAI Client

from prometheus_swarm.clients import OpenAIClient

client = OpenAIClient(
    model="gpt-4-turbo-preview",
    temperature=0.7,
    max_tokens=1000,
)

Google Client

from prometheus_swarm.clients import GoogleClient

client = GoogleClient(
    model="gemini-pro",
    temperature=0.7,
    max_tokens=1000,
)

Logging Configuration

Prometheus Swarm uses Python’s built-in logging module. You can configure it like this:

import logging

logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)

Advanced Configuration

Custom Model Configuration

from prometheus_swarm.types import ModelConfig

model_config = ModelConfig(
    name="custom-model",
    api_base="https://api.custom.com",
    api_key="your-key",
    max_tokens=2000,
)

Retry Configuration

from prometheus_swarm.types import RetryConfig

retry_config = RetryConfig(
    max_attempts=3,
    initial_delay=1,
    max_delay=10,
    exponential_base=2,
)

Workflow Timeout Configuration

from prometheus_swarm.types import TimeoutConfig

timeout_config = TimeoutConfig(
    request_timeout=30,
    workflow_timeout=300,
)