Skip to main content

Configuration

AI CLI can be configured with various API keys, AI models, and Git settings to suit your needs.

API Keys​

The CLI supports multiple AI and development services. You'll need to configure API keys for the services you want to use.

Supported Services​

  • anthropic (Claude)
  • github (for search functionality)
  • google-studio
  • openai

Setting API Keys​

Use the config command to set API keys:

ai config <service> <apiKey>

Examples​

# Configure Anthropic (Claude) API key
ai config anthropic sk-ant-xxxxx

# Configure GitHub token
ai config github ghp_xxxxx

Getting API Keys​

Use the config-help command to get instructions for obtaining API keys:

# Get help for all supported services
ai config-help

# Get help for a specific service
ai config-help anthropic
ai config-help github

Listing Configured Services​

View which services are configured:

ai config-list

Git Settings​

Configure Git-related settings for worktree management.

Default Branch​

Set your preferred default branch for merge operations:

# Set default branch
ai config-git default-branch develop

# View current configuration
ai config-git show

The default branch setting is used by:

  • worktree-merge when no target branch is specified
  • worktree-merge-all when no target branch is specified

This allows you to match your repository's branching strategy (e.g., using 'develop' instead of 'main').

AI Models​

You can configure which AI model to use for each command.

Available Models​

  • Anthropic (Claude):
    • claude-3-opus-20240229
    • claude-3-sonnet-20240229
    • claude-3-haiku-20240307
  • OpenAI:
    • gpt-4
    • gpt-4-turbo-preview
    • gpt-3.5-turbo

When you configure your first service, it will automatically become the default for all commands. Here are our recommended default models for each service:

Anthropic (Claude)​

  • Default Model: claude-3-haiku-20240307
    • Fast and efficient for most tasks
    • Good balance of speed and quality
    • Recommended for search and documentation generation

OpenAI​

  • Default Model: gpt-4-turbo-preview
    • Most capable model for complex tasks
    • Excellent for debugging and code analysis
    • Recommended for debugging and complex code analysis

Google Studio​

  • Default Model: gemini-pro
    • Good for code analysis and documentation
    • Efficient for general tasks
    • Recommended for code collection and analysis

Command-Specific Recommendations​

While you can use any service/model combination, here are our recommended pairings for each command based on their specific needs:

Search Command​

  • Recommended: Anthropic (Claude) with claude-3-haiku-20240307
    • Fast response times for quick searches
    • Good at analyzing and summarizing search results
    • Efficient for handling multiple search results
    • Alternative: OpenAI with gpt-4-turbo-preview for more detailed analysis

Debug Command​

  • Recommended: OpenAI with gpt-4-turbo-preview
    • Excellent at understanding complex error patterns
    • Strong reasoning capabilities for debugging
    • Good at suggesting specific code fixes
    • Alternative: Anthropic with claude-3-sonnet-20240229 for a good balance

Describe Command​

  • Recommended: Anthropic (Claude) with claude-3-sonnet-20240229
    • Good at generating clear, concise documentation
    • Strong understanding of code structure
    • Balanced performance for documentation tasks
    • Alternative: OpenAI with gpt-4-turbo-preview for more detailed documentation

Collect Command​

  • Recommended: Google Studio with gemini-pro
    • Efficient at analyzing large codebases
    • Good at identifying patterns across multiple files
    • Fast processing of collected code
    • Alternative: Anthropic with claude-3-haiku-20240307 for more focused analysis

Configuring Models​

Use the config-model command to set which model to use for each command:

ai config-model <command> <service> <model>

Examples​

# Configure Claude for search command
ai config-model search anthropic claude-3-haiku-20240307

# Configure GPT-4 for debug command
ai config-model debug openai gpt-4

Listing Model Configuration​

View your current model configuration:

ai config-model list

Output Directory​

By default, the CLI creates a .ai-cli directory in your project root to store:

  • Search results and analysis
  • Collected TypeScript files
  • Generated documentation
  • Worktree state tracking
  • Configuration files

You can add this to your .gitignore:

.ai-cli/

The directory structure will look like:

your-project/
├── .ai-cli/
│   ├── code_search_20240320_143045.md
│   ├── src_collection.txt
│   ├── src_collection_analysis.md
│   ├── config.json
│   └── worktrees.json
├── src/
│   ├── myfile.ts
│   └── myfile.md
└── ...