Configuration Options

Complete reference for all Anouk configuration options.

---

Core Options

provider

Type: string Default: 'together' Required: No

The AI provider to use. Determines which preset configuration to apply.

{ provider: 'openai' }

Valid values: 'openai', 'together', 'anthropic', 'ollama', 'huggingface', or any custom string.

---

apiKey

Type: string Default: '' Required: Yes (for cloud providers)

API authentication key for the provider.

{ apiKey: 'sk-your-api-key-here' }

Security

Never commit API keys to version control. Use environment variables or runtime configuration.

---

model

Type: string Default: Varies by provider Required: No

The model identifier to use for API requests.

{ model: 'gpt-4' }

Common values by provider:

Provider	Models
OpenAI	gpt-4, gpt-4-turbo, gpt-3.5-turbo
Together	meta-llama/Llama-3-70b-chat-hf, mistralai/Mixtral-8x7B-Instruct-v0.1
Anthropic	claude-3-opus-20240229, claude-3-sonnet-20240229
Ollama	llama2, mistral, codellama

---

baseUrl

Type: string Default: Varies by provider Required: No

The API endpoint URL.

{ baseUrl: 'https://api.openai.com/v1' }

Default values by provider:

Provider	Base URL
OpenAI	https://api.openai.com/v1
Together	https://api.together.xyz/v1
Anthropic	https://api.anthropic.com/v1
Ollama	http://localhost:11434/v1
Hugging Face	https://api-inference.huggingface.co/v1

---

Generation Options

maxTokens

Type: number Default: 2000 Required: No

Maximum number of tokens in the response.

{ maxTokens: 4000 }

Considerations: - Higher values allow longer responses - May increase API costs - Must be within model's context limit

---

temperature

Type: number Default: 0.7 Range: 0 to 1 (some models support up to 2) Required: No

Controls response randomness/creativity.

{ temperature: 0.3 }

Guidelines:

Value	Effect	Use Case
0.0-0.3	More deterministic	Factual, consistent outputs
0.4-0.7	Balanced	General use
0.8-1.0	More creative	Creative writing, brainstorming

---

systemPrompt

Type: string Default: '' Required: No

Default system instruction prepended to all requests.

{
  systemPrompt: 'You are a helpful email assistant. Be concise and professional.'
}

Best practices: - Keep it concise but specific - Define the assistant's role and behavior - Include any constraints or formatting requirements

---

Provider Presets

Presets provide sensible defaults for each provider. Access via:

ConfigManager.getPresetConfigs()

OpenAI Preset

{
  provider: 'openai',
  baseUrl: 'https://api.openai.com/v1',
  model: 'gpt-4',
  maxTokens: 2000,
  temperature: 0.7
}

Together Preset

{
  provider: 'together',
  baseUrl: 'https://api.together.xyz/v1',
  model: 'meta-llama/Llama-3-70b-chat-hf',
  maxTokens: 2000,
  temperature: 0.7
}

Anthropic Preset

{
  provider: 'anthropic',
  baseUrl: 'https://api.anthropic.com/v1',
  model: 'claude-3-opus-20240229',
  maxTokens: 4000,
  temperature: 0.7
}

Ollama Preset

{
  provider: 'ollama',
  baseUrl: 'http://localhost:11434/v1',
  model: 'llama2',
  maxTokens: 2000,
  temperature: 0.7
}

Hugging Face Preset

{
  provider: 'huggingface',
  baseUrl: 'https://api-inference.huggingface.co/v1',
  model: 'meta-llama/Llama-2-70b-chat-hf',
  maxTokens: 2000,
  temperature: 0.7
}

---

Storage Keys

Anouk uses localStorage with these keys:

Key	Purpose
anouk_config	Persisted configuration
anouk_cache_*	Cached API responses

Clear All Data

// Clear configuration
localStorage.removeItem('anouk_config');

// Clear all cached responses
Object.keys(localStorage)
  .filter(key => key.startsWith('anouk_cache_'))
  .forEach(key => localStorage.removeItem(key));

---

Environment Variables

For build-time configuration:

// In your extension code
const config = {
  apiKey: process.env.ANOUK_API_KEY,
  provider: process.env.ANOUK_PROVIDER || 'together'
};

With esbuild:

esbuild src/extension.js --bundle --define:process.env.ANOUK_API_KEY=\"your-key\"

---

Configuration Priority

Configuration is resolved in this order (highest priority first):

1. Runtime updates - aiService.updateConfig()
2. Constructor config - new AIService(config)
3. Stored config - From localStorage
4. Provider preset - Based on provider name
5. Default config - Built-in defaults

Example:

// Default: temperature = 0.7
// Stored config: temperature = 0.5
// Constructor: temperature not specified
// Runtime: aiService.updateConfig({ temperature: 0.3 })

// Result: temperature = 0.3

---

Complete Configuration Example

import { AIService } from 'anouk';

const aiService = new AIService({
  // Provider identification
  provider: 'openai',

  // Authentication
  apiKey: 'sk-your-api-key',

  // API endpoint
  baseUrl: 'https://api.openai.com/v1',

  // Model selection
  model: 'gpt-4',

  // Generation parameters
  maxTokens: 2000,
  temperature: 0.7,

  // System behavior
  systemPrompt: `You are a helpful assistant that specializes in
email analysis. Always respond in a professional, concise manner.
Format lists using bullet points.`
});

---

Validation

Anouk performs basic validation on configuration:

Option	Validation
temperature	Must be between 0 and 2
maxTokens	Must be positive integer
baseUrl	Must be valid URL format

Invalid configurations may result in API errors or unexpected behavior.