setup

Overview ¶

Package setup provides an interactive first-time setup wizard for AINative Code.

The setup wizard guides new users through the initial configuration process, including:

• LLM provider selection (Anthropic, OpenAI, Google Gemini, Ollama)
• API key input and validation
• Model selection with recommendations
• Optional AINative platform integration
• Extended thinking configuration
• Color scheme and UI preferences

The wizard creates a configuration file at ~/.ainative-code.yaml and an initialization marker at ~/.ainative-code-initialized to track first-time setup.

Interactive Mode ¶

By default, the wizard runs in interactive mode with a rich TUI powered by Bubble Tea. Users can navigate through steps using arrow keys and make selections interactively:

wizard := setup.NewWizard(ctx, setup.WizardConfig{
    InteractiveMode: true,
    SkipValidation:  false,
})
result, err := wizard.Run()

Non-Interactive Mode ¶

For CI/CD environments or automated setups, the wizard supports non-interactive mode using environment variables:

export AINATIVE_PROVIDER=anthropic
export AINATIVE_ANTHROPIC_API_KEY=sk-ant-...
export AINATIVE_ANTHROPIC_MODEL=claude-3-5-sonnet-20241022
ainative-code setup --non-interactive

Validation ¶

The wizard validates API keys and credentials by default:

• Anthropic: Tests API key format and authentication
• OpenAI: Validates key format and tests API access
• Google: Checks API key validity
• Ollama: Verifies server connectivity and model availability

Validation can be skipped for faster setup:

ainative-code setup --skip-validation

First-Run Detection ¶

The wizard automatically detects first-time runs by checking for the initialization marker file. Users can force re-run setup:

ainative-code setup --force

Configuration Structure ¶

The generated configuration includes:

• App metadata (name, version, environment)
• LLM provider settings with defaults
• Platform authentication (optional)
• Performance tuning (cache, concurrency)
• Logging configuration
• Security settings

Provider-Specific Configuration ¶

Anthropic (Claude):

• API key validation (must start with "sk-ant-")
• Model selection (Sonnet 3.5, Opus 3, etc.)
• Extended thinking mode
• Prompt caching preferences

OpenAI (GPT):

• API key validation (must start with "sk-")
• Model selection (GPT-4 Turbo, GPT-4, GPT-3.5)
• Organization ID (optional)

Google (Gemini):

• API key validation
• Model selection (Gemini Pro, Gemini Pro Vision)
• Project ID and location (optional)

Ollama (Local):

• Server URL (default: http://localhost:11434)
• Model name (must be pre-downloaded)
• Connection validation

Error Handling ¶

The wizard handles errors gracefully:

• Invalid API keys: Clear error messages with help links
• Network failures: Distinguishes from validation failures
• User cancellation: Clean exit without partial state
• File system errors: Directory creation and permissions

Testing ¶

The package includes comprehensive tests for:

• Configuration building for all providers
• API key validation
• File operations (writing config, creating markers)
• First-run detection
• Non-interactive mode
• Error cases

Example usage in tests:

wizard := setup.NewWizard(ctx, setup.WizardConfig{
    InteractiveMode: false,
    SkipValidation:  true,
})
wizard.SetSelections(map[string]interface{}{
    "provider":          "anthropic",
    "anthropic_api_key": "sk-ant-test",
})
result, err := wizard.Run()