SDK Reference

Complete reference for LaikaTest SDKs

Getting Your API Token

Generating API Tokens

1. Go to Dashboard → Settings → API Keys
2. Click Create New API Key
3. Give your key a descriptive name
4. Select the project to associate with the key
5. Copy the generated token (displayed only once)

Environment Variables

Store your API key in environment variables for security:

# .envLAIKATEST_API_KEY=your_api_token_here

Base URL

https://api.laikatest.com

Authentication Methods

LaikaTest supports two authentication methods:

• API Tokens: For external integrations and programmatic access (used by SDK)
• JWT Tokens: For dashboard and session-based authentication

Access Control

Installation

npm install @laikatest/js-client

Client Initialization

Basic Setup

import { LaikaTest } from '@laikatest/js-client';const client = new LaikaTest(apiKey);

With Configuration Options

const client = new LaikaTest(apiKey, {  baseUrl: 'https://api.laikatest.com',  timeout: 10000,      // Request timeout in milliseconds  cacheTTL: 1800000,   // Cache time-to-live (30 minutes)  cacheEnabled: true   // Enable/disable caching});

Core Methods

getPRomt()

Retrieve a prompt template by name.

const prompt = await client.getPRomt('my-prompt-name');// Compile the prompt with variablesconst compiledPrompt = prompt.compile({  username: 'Alice',  role: 'developer'});// With optionsconst prompt = await client.getPRomt('my-prompt-name', {  versionId: '10',        // or 'v10'  bypassCache: true       // Skip cache for this request});

Parameters:

• promptName (string) - The name of the prompt template
• options (optional):
  • versionId / version_id - Specific version to fetch (numeric or 'v' prefixed)
  • bypassCache / bypass_cache - Skip cache for this request

getExperimentPrompt()

Evaluate an experiment and retrieve the assigned prompt variant. The returned prompt includes embedded experiment metadata for tracking and score submission.

const prompt = await client.getExperimentPrompt(  'experiment-name',  {    user_id: 'user123',    session_id: 'session456',    custom_param: 'value'  });// Compile the prompt with variablesconst compiledPrompt = prompt.compile({ username: 'Alice' });// Submit scores using the prompt object (metadata is embedded)await client.pushScore(prompt, {  scores: [{ name: 'rating', type: 'int', value: 5 }],  userId: 'user123'});

Parameters:

• experimentName (string) - The experiment identifier
• variables (object) - Context variables for experiment evaluation

Returns: A Prompt instance with experiment metadata embedded.

pushScore()

Submit performance metrics and scores for experiment tracking and analytics. Pass the prompt object from getExperimentPrompt() to automatically include experiment metadata.

// Get experiment promptconst prompt = await client.getExperimentPrompt('my-experiment', {  userId: 'user123'});// Submit scores with prompt object (recommended)await client.pushScore(prompt, {  scores: [    { name: 'rating', type: 'int', value: 5 },    { name: 'helpful', type: 'bool', value: true },    { name: 'comment', type: 'string', value: 'Great response!' }  ],  sessionId: 'session-456',  // At least one of sessionId or userId required  userId: 'user-123'});

Parameters:

• prompt (Prompt object, required) - The prompt object returned from getExperimentPrompt()
• options (object, required):
  • scores (array, required) - Array of score objects:
    • name - Metric name (e.g., 'rating', 'helpful')
    • type - Data type: 'int', 'bool', or 'string'
    • value - Metric value matching the specified type
  • sessionId (string, conditional) - Session identifier for grouping interactions. At least one of sessionId or userId must be provided.
  • userId (string, conditional) - User identifier for tracking individual users. At least one of sessionId or userId must be provided.

prompt.compile()

Compile a prompt template with variable substitution. Returns the compiled content as a string.

const prompt = await client.getPRomt('greeting');const compiled = prompt.compile({  name: 'Alice',  role: 'developer'});console.log(compiled); // Compiled prompt string

client.destroy()

Clean up resources and close connections.

try {  const prompt = await client.getPRomt('my-prompt');  // Use prompt...} finally {  client.destroy();}

Error Handling

The SDK provides specialized error classes for different failure scenarios:

import {  LaikaTest,  ValidationError,  AuthenticationError,  NetworkError,  LaikaServiceError} from '@laikatest/js-client';try {  const prompt = await client.getPRomt('my-prompt');} catch (error) {  if (error instanceof AuthenticationError) {    console.error('Invalid API key');  } else if (error instanceof NetworkError) {    console.error('Network request failed');  } else if (error instanceof LaikaServiceError) {    console.error('Service error:', error.message);  }}

Error Types

• ValidationError: Invalid input parameters
• AuthenticationError: Invalid or missing API key
• NetworkError: Network request failures
• LaikaServiceError: Server-side errors

Best Practices

• Store API keys securely: Use environment variables, never hardcode keys
• Reuse client instances: Create one client instance and reuse it across requests
• Implement error handling: Always wrap SDK calls in try-catch blocks
• Use caching wisely: Keep caching enabled for better performance, bypass only when necessary
• Clean up resources: Call destroy() when done, especially in serverless environments

Complete Example

import { LaikaTest } from '@laikatest/js-client';const client = new LaikaTest(process.env.LAIKATEST_API_KEY, {  timeout: 5000,  cacheTTL: 600000  // 10 minutes});try {  // Fetch prompt  const prompt = await client.getPRomt('customer-greeting');  // Compile with variables  const message = prompt.compile({    customerName: 'John Doe',    accountType: 'premium'  });  console.log(message);} catch (error) {  console.error('Failed to fetch prompt:', error);} finally {  client.destroy();}