Skip to content

Configuration Guide

LoopLM provides a straightforward way to configure different LLM providers. All configuration is handled through the CLI, with credentials stored securely using encryption.

Initial Setup

To start configuring LoopLM, run:

looplm --configure

This will launch an interactive setup process where you can configure one or more providers.

Supported Providers

LoopLM supports the following providers:

Anthropic

Required environment variables: - ANTHROPIC_API_KEY

Example model: claude-3-5-sonnet-20240620

OpenAI

Required environment variables: - OPENAI_API_KEY

Example model: gpt-4o

Google Gemini

Required environment variables: - GEMINI_API_KEY

Example model: gemini/gemini-pro

Azure OpenAI

Required environment variables: - AZURE_API_KEY - AZURE_API_BASE - AZURE_API_VERSION

Example model: azure/gpt-4o

AWS Bedrock

Required environment variables: - AWS_ACCESS_KEY_ID - AWS_SECRET_ACCESS_KEY - AWS_REGION_NAME

Example model: anthropic.claude-3-5-sonnet-20240620-v1:0

Other Providers

LoopLM supports any provider that's compatible with LiteLLM. When configuring other providers, you'll need to:

  1. Specify the provider name
  2. Enter the required environment variables
  3. Specify the model name according to LiteLLM's documentation

Configuration Commands

View Current Configuration

looplm --status

This shows: - Configured providers - Default provider and model - Provider status

Set Default Provider

looplm --set-default <provider>

Example: 

looplm --set-default anthropic

Reset Configuration

Reset all configuration: 

looplm --reset

Reset specific provider: 

looplm --reset-provider anthropic

Using Different Providers

In Direct Prompts

# Use specific provider
looplm --provider anthropic "Explain quantum computing"

# Use specific model
looplm --provider openai --model gpt-4 "Write a regex pattern"

In Chat Mode

# Start chat with specific provider
looplm chat --provider anthropic

# Start chat with specific model
looplm chat --provider openai --model gpt-4o-mini

Configuration Storage

LoopLM stores configuration in two locations in your home directory:

  1. .looplm/config.json: General configuration (non-sensitive)
  2. .looplm/secrets.enc: Encrypted API keys and credentials
  3. .looplm/sessions/: Saved chat sessions
  4. .looplm/prompts/: System prompts

The configuration is encrypted using Fernet symmetric encryption, ensuring your API keys remain secure.

Additional Environment Variables

When configuring a provider, you can set additional environment variables that might be required for your specific use case. These will be stored securely with your other credentials.

Managing System Prompts

In chat mode, you can manage system prompts that define the assistant's behavior:

  1. View current prompt: /system
  2. Create new prompt: Use the system command menu
  3. Save prompts for reuse: Save option in system menu
  4. Load saved prompts: Load option in system menu

This allows you to customize the assistant's behavior for different use cases.