RCrewAI::Configuration

Configuration system for LLM providers and settings

RCrewAI::Configuration

The Configuration class manages settings for various LLM providers and global options for RCrewAI.

Supported Providers

RCrewAI supports the following LLM providers:

OpenAI (GPT-4, GPT-3.5-turbo, etc.)
Anthropic (Claude 3, Claude 2, etc.)
Google (Gemini Pro, PaLM, etc.)
Azure OpenAI (Azure-hosted OpenAI models)
Ollama (Local/self-hosted models)

Basic Configuration

Using Environment Variables (Recommended)

Set your API keys as environment variables:

export OPENAI_API_KEY="your-openai-key"
export ANTHROPIC_API_KEY="your-anthropic-key"
export GOOGLE_API_KEY="your-google-key"
export AZURE_OPENAI_API_KEY="your-azure-key"

Then configure RCrewAI:

RCrewAI.configure do |config|
  config.llm_provider = :openai  # or :anthropic, :google, :azure, :ollama
end

Direct Configuration

RCrewAI.configure do |config|
  config.llm_provider = :openai
  config.openai_api_key = "your-openai-key"
  config.openai_model = "gpt-4"
  config.temperature = 0.1
  config.max_tokens = 4000
end

Provider-Specific Configuration

OpenAI

RCrewAI.configure do |config|
  config.llm_provider = :openai
  config.openai_api_key = ENV['OPENAI_API_KEY']
  config.openai_model = 'gpt-4'  # or 'gpt-3.5-turbo', 'gpt-4-turbo'
  config.temperature = 0.1
  config.max_tokens = 4000
end

Available Models:

gpt-4
gpt-4-turbo
gpt-3.5-turbo
gpt-3.5-turbo-16k

Anthropic (Claude)

RCrewAI.configure do |config|
  config.llm_provider = :anthropic
  config.anthropic_api_key = ENV['ANTHROPIC_API_KEY']
  config.anthropic_model = 'claude-3-sonnet-20240229'
  config.temperature = 0.1
  config.max_tokens = 4000
end

Available Models:

claude-3-opus-20240229
claude-3-sonnet-20240229
claude-3-haiku-20240307
claude-2.1
claude-2.0

Google (Gemini)

RCrewAI.configure do |config|
  config.llm_provider = :google
  config.google_api_key = ENV['GOOGLE_API_KEY']
  config.google_model = 'gemini-pro'
  config.temperature = 0.1
  config.max_tokens = 4000
end

Available Models:

gemini-pro
gemini-pro-vision
text-bison-001

Azure OpenAI

RCrewAI.configure do |config|
  config.llm_provider = :azure
  config.azure_api_key = ENV['AZURE_OPENAI_API_KEY']
  config.base_url = "https://your-resource.openai.azure.com/"
  config.api_version = "2023-05-15"
  config.deployment_name = "your-deployment-name"
  config.azure_model = 'gpt-4'
end

Ollama (Local Models)

RCrewAI.configure do |config|
  config.llm_provider = :ollama
  config.base_url = "http://localhost:11434"  # Default Ollama URL
  config.model = "llama2"  # or any installed Ollama model
  config.temperature = 0.1
end

Configuration Methods

`RCrewAI.configure`

Configure RCrewAI with a block:

RCrewAI.configure do |config|
  config.llm_provider = :openai
  config.temperature = 0.2
  config.max_tokens = 2000
end

`RCrewAI.configuration`

Access the current configuration:

config = RCrewAI.configuration
puts config.llm_provider  # => :openai
puts config.model         # => "gpt-4"

`RCrewAI.reset_configuration!`

Reset configuration to defaults:

RCrewAI.reset_configuration!

Configuration Options

Core Settings

Option	Type	Default	Description
`llm_provider`	Symbol	`:openai`	LLM provider to use
`temperature`	Float	`0.1`	Sampling temperature (0.0 - 2.0)
`max_tokens`	Integer	`4000`	Maximum tokens in response
`timeout`	Integer	`120`	Request timeout in seconds

Provider-Specific Keys

Option	Environment Variable	Description
`openai_api_key`	`OPENAI_API_KEY`	OpenAI API key
`anthropic_api_key`	`ANTHROPIC_API_KEY`	Anthropic API key
`google_api_key`	`GOOGLE_API_KEY`	Google API key
`azure_api_key`	`AZURE_OPENAI_API_KEY`	Azure OpenAI API key

Provider-Specific Models

Option	Default	Description
`openai_model`	`gpt-4`	OpenAI model name
`anthropic_model`	`claude-3-sonnet-20240229`	Anthropic model name
`google_model`	`gemini-pro`	Google model name
`azure_model`	`gpt-4`	Azure OpenAI model name

Azure-Specific Settings

Option	Environment Variable	Description
`base_url`	`LLM_BASE_URL`	Azure OpenAI endpoint URL
`api_version`	`AZURE_API_VERSION`	Azure API version
`deployment_name`	`AZURE_DEPLOYMENT_NAME`	Azure deployment name

Dynamic Provider Switching

You can switch providers at runtime:

# Start with OpenAI
RCrewAI.configure do |config|
  config.llm_provider = :openai
  config.openai_api_key = ENV['OPENAI_API_KEY']
end

crew = RCrewAI::Crew.new("test_crew")
# ... create agents and tasks

# Switch to Anthropic for specific tasks
RCrewAI.configure do |config|
  config.llm_provider = :anthropic
  config.anthropic_api_key = ENV['ANTHROPIC_API_KEY']
end

different_crew = RCrewAI::Crew.new("claude_crew")
# This crew will use Claude models

Environment Variables

RCrewAI automatically loads configuration from environment variables:

# Provider API Keys
export OPENAI_API_KEY="sk-..."
export ANTHROPIC_API_KEY="sk-ant-..."
export GOOGLE_API_KEY="AIza..."
export AZURE_OPENAI_API_KEY="..."

# Azure Specific
export AZURE_API_VERSION="2023-05-15"
export AZURE_DEPLOYMENT_NAME="gpt-4-deployment"
export LLM_BASE_URL="https://your-resource.openai.azure.com/"

# General fallback
export LLM_API_KEY="fallback-key"

Configuration Validation

RCrewAI validates configuration when configure is called:

begin
  RCrewAI.configure do |config|
    config.llm_provider = :invalid_provider
  end
rescue RCrewAI::ConfigurationError => e
  puts "Configuration error: #{e.message}"
end

Multiple Configurations

For applications that need multiple configurations:

# Save current config
old_config = RCrewAI.configuration.dup

# Configure for specific task
RCrewAI.configure do |config|
  config.llm_provider = :anthropic
  config.temperature = 0.8
end

# Do work with Claude...

# Restore original config
RCrewAI.reset_configuration!
# Re-apply old config...

Best Practices

Use Environment Variables: Store API keys in environment variables, not in code
Provider Selection: Choose providers based on task requirements:
- OpenAI GPT-4: General-purpose, high quality
- Claude: Great for analysis and reasoning
- Gemini: Good for creative tasks
- Azure: Enterprise requirements
- Ollama: Privacy-sensitive or offline use
Temperature Settings:
- Low (0.1-0.3): Factual, consistent outputs
- Medium (0.5-0.7): Balanced creativity and consistency
- High (0.8-1.0): Creative, varied outputs
Model Selection: Use the most capable model your budget allows
Timeout Settings: Set appropriate timeouts for your use case

Troubleshooting

Common Issues

“API key must be set” Error

# Ensure API key is set
RCrewAI.configure do |config|
  config.llm_provider = :openai
  config.openai_api_key = ENV['OPENAI_API_KEY']  # Make sure this env var exists
end

“Unsupported provider” Error

# Check supported providers
puts RCrewAI.configuration.supported_providers
# => [:openai, :anthropic, :google, :azure, :ollama]

Azure Configuration Issues

RCrewAI.configure do |config|
  config.llm_provider = :azure
  config.azure_api_key = ENV['AZURE_OPENAI_API_KEY']
  config.base_url = "https://your-resource.openai.azure.com/"  # Include trailing slash
  config.api_version = "2023-05-15"  # Use supported API version
  config.deployment_name = "your-deployment-name"  # Exact deployment name
end