how

How AI CLI

A terminal-based AI assistant powered by various LLM providers.

Configuration

How AI CLI supports both global and local configurations, allowing you to have different settings for different projects.

Global Configuration

Global configuration applies to all projects and is stored in ~/.how/config.yaml.

1. Create a default global configuration:

how init

This creates a minimal configuration file with OpenAI as the default provider.

1. Edit the configuration file to add your API keys and customize settings:

# Open with your preferred editor
nano ~/.how/config.yaml
# or
code ~/.how/config.yaml

Local Configuration

Local configuration applies only to the current directory and is stored in ./.how/config.yaml. Local configuration takes precedence over global configuration.

1. Create a default local configuration:

how init --local

1. Edit the local configuration file:

# Open with your preferred editor
nano ./.how/config.yaml
# or
code ./.how/config.yaml

Supported Providers

The following AI providers are supported:

OpenAI

default_provider: openai
openai:
  api_key: "your-openai-api-key"
  model: "gpt-4o" # or any other OpenAI model
  timeout: 30000 # optional, in milliseconds (30 seconds)
  # Azure OpenAI specific settings (optional)
  by_azure: false
  base_url: "" # required for Azure: https://{YOUR_RESOURCE_NAME}.openai.azure.com
  api_version: "" # required for Azure

Claude (Anthropic)

default_provider: claude
claude:
  api_key: "your-anthropic-api-key"
  model: "claude-3-7-sonnet-latest" # or any other Claude model
  max_tokens: 2000 # required
  # AWS Bedrock specific settings (optional)
  by_bedrock: false
  access_key: ""
  secret_access_key: ""
  region: ""

Gemini (Google)

default_provider: gemini
gemini:
  api_key: "your-gemini-api-key"
  model: "gemini-pro" # or gemini-pro-vision, gemini-1.5-flash, etc.

Deepseek

default_provider: deepseek
deepseek:
  api_key: "your-deepseek-api-key"
  model: "deepseek-coder" # or other Deepseek models
  timeout: 60000 # optional, in milliseconds (1 minute)
  base_url: "https://api.deepseek.com/" # optional

Ollama (Local Models)

default_provider: ollama
ollama:
  base_url: "http://localhost:11434" # URL to your Ollama server
  model: "llama3" # or any other model you have installed
  timeout: 30000 # optional, in milliseconds (30 seconds)

Configuration Priority

When running the How AI CLI, it will:

1. First check for a local configuration file (./.how/config.yaml)
2. If no local configuration is found, use the global configuration (~/.how/config.yaml)
3. If no configuration is found at all, prompt you to create one with how init

This allows you to have different settings for different projects while maintaining a global default.