Oxylabs AI Studio JavaScript SDK

A JavaScript SDK for seamlessly interacting with Oxylabs AI Studio API services, including AI-Scraper, AI-Crawler, AI-Browser-Agent and other data extraction tools.

Installation

npm install oxylabs-ai-studio

Quick Start

1. Environment Setup

Either add OXYLABS_AI_STUDIO_API_URL and OXYLABS_AI_STUDIO_API_KEY values to the .env file, or as your environment variables:

export OXYLABS_AI_STUDIO_API_KEY=your_api_key_here

AI-Scraper

Generate Schema

import { 
  OxylabsAIStudioSDK
} from 'oxylabs-ai-studio';

const sdk = new OxylabsAIStudioSDK({
  apiKey: 'your_api_key_here',
  timeout: 120000,
  retryAttempts: 3,
});

async function testGenerateSchema() {
  try {
    console.log('Testing schema generation...');
    const schema = await sdk.aiScraper.generateSchema({
      user_prompt: 'Extract the title of the page'
    });
    console.log('Schema:', schema);
  } catch (error) {
    console.error('Schema generation error:', error.message);
  }
}

testGenerateSchema();

Basic usage

import { 
  OxylabsAIStudioSDK, 
  OutputFormat
} from 'oxylabs-ai-studio';

const sdk = new OxylabsAIStudioSDK({
  apiKey: 'your_api_key_here',
  timeout: 120000,
  retryAttempts: 3,
});

async function testScrapeOutputJson() {
  try {
    console.log('Testing synchronous scraping with JSON output...');
    
    const options = {
      url: 'https://www.freelancer.com',
      output_format: OutputFormat.JSON,
      geo_location: "US",
      schema: {
        type: 'object',
        properties: {
          links: { type: 'array', items: { type: 'string' } }
        }
      }
    };
    
    const results = await sdk.aiScraper.scrape(options);
    console.log('Sync scraping results:', results);
  } catch (error) {
    console.error('Sync scraping error:', error.message);
  }
}

testScrapeOutputJson();

Available Parameters

• url (string): The target URL to process.
• output_format (string): The desired format for the output. Can be either markdown, json, screenshot, csv or toon. Defaults to markdown.
• render_javascript (boolean | "auto"): Whether to render JavaScript before extraction. Can be "auto" to auto-detect if rendering is needed. Defaults to false.
• schema (Record<string, any>): A JSON Schema object that defines the structure of the output data. This is required when output_format is json, csv or toon.
• geo_location (string): Specifies the geographic location (ISO2 format) or country canonical name from which the request should be simulated.
• user_agent (string): User-Agent request header. See available values.

AI-Crawler

Basic usage

import { 
  OxylabsAIStudioSDK, 
  OutputFormat
} from 'oxylabs-ai-studio';

const sdk = new OxylabsAIStudioSDK({
  apiKey: 'your_api_key_here',
  timeout: 120000,
  retryAttempts: 3,
});

async function testCrawlOutputJson() {
  try {
    console.log('Testing crawling with JSON output...');
    
    const options = {
      url: 'https://www.freelancer.com',
      output_format: OutputFormat.JSON,
      user_prompt: 'Get job ad pages',
      return_sources_limit: 3,
      geo_location: "Germany",
      schema: {
        type: "object",
        properties: {
          jobAd: {
            type: "object",
            properties: {
              position_title: {
                type: "string"
              },
              salary: {
                type: "string"
              }
            }
          }
        }
      }
    };
    
    const results = await sdk.aiCrawler.crawl(options);
    console.log('Crawling results:', JSON.stringify(results, null, 2));      
  } catch (error) {
    console.error('Crawling error:', error.message);
  }
}

testCrawlOutputJson();

Available Parameters

• url (string): The starting URL for the crawl.
• user_prompt (string): Crawling instructions.
  • For auto-schema flow, use parse_prompt in crawlWithAutoSchema() to generate the schema.
• output_format (string): The desired format for the output. Can be either markdown, json, csv or toon. Defaults to markdown.
• return_sources_limit (integer): The maximum number of pages/sources to return. Defaults to 25.
• render_javascript (boolean): Whether to render JavaScript on pages before extraction. Defaults to false.
• schema (Record<string, any>): A JSON Schema object that defines the structure of the output data. Required when output_format is json, csv or toon.
• geo_location (string): Specifies the geographic location (ISO2 format) or country canonical name from which the request should be simulated.
• max_credits (integer | null): Optional cap on credits to spend for the run.

Browser-Agent

Basic usage

import { 
  OxylabsAIStudioSDK, 
  OutputFormat
} from 'oxylabs-ai-studio';

const sdk = new OxylabsAIStudioSDK({
  apiKey: 'your_api_key_here',
  timeout: 120000,
  retryAttempts: 3,
});

async function testBrowseOutputJson() {
  try {
    console.log('Testing synchronous browsing with JSON output...');
    
    const options = {
      url: 'https://www.freelancer.com',
      output_format: OutputFormat.JSON,
      user_prompt: 'Navigate to the first job ad you can find.',
      geo_location: "US",
      schema: {
        type: 'object',
        properties: {
          job_title: { type: 'string' }
        }
      }
    };
    
    const results = await sdk.browserAgent.browse(options);
    console.log('Sync browsing results:', JSON.stringify(results, null, 2));
  } catch (error) {
    console.error('Sync browsing error:', error.message);
  }
}

testBrowseOutputJson();

Available Parameters

• url (string): The target URL for the browser agent to start at.
• user_prompt (string): Instructions describing what actions to perform and data to extract.
  • For auto-schema flow, use parse_prompt in browseWithAutoSchema() to generate the schema.
• output_format (string): The desired format for the output. Can be markdown, html, json, csv, toon, or screenshot. Defaults to markdown.
• schema (Record<string, any>): A JSON Schema object that defines the structure of the output data. This is required when output_format is json, csv or toon.
• geo_location (string): Specifies the geographic location (ISO2 format) or country canonical name from which the request should be simulated.

AI-Search

Basic usage

import {
  OxylabsAIStudioSDK,
} from 'oxylabs-ai-studio';

const sdk = new OxylabsAIStudioSDK({
  apiKey: 'your_api_key_here',
  timeout: 120000,
  retryAttempts: 3,
});

async function testSearch() {
  try {
    console.log('Testing search...');

    const options = {
      query: 'weather in London',
      limit: 3,
      return_content: true,
      render_javascript: false,
      geo_location: "IT",
    };

    const results = await sdk.aiSearch.search(options);
    console.log('Search results:', JSON.stringify(results, null, 2));
  } catch (error) {
    console.error('Search error:', error.message);
  }
}

testSearch();

Available Parameters

• query (string): The search query.
• limit (integer): The maximum number of search results to return. Maximum: 50.
• geo_location (string): ISO 2-letter format, country name, coordinate formats are supported. See more at SERP Localization.

Performance Optimization

The search() method automatically optimizes performance by intelligently choosing between two endpoints:

Instant Endpoint (used automatically when):

• limit is ≤ 10 (or undefined)
• AND return_content is false (or undefined)

This provides real-time results without polling for faster response times.

Polling-based Endpoint (used when):

• limit > 10
• OR return_content is true

You can also directly use searchInstant() method if you want to explicitly use the instant endpoint:

const results = await sdk.aiSearch.searchInstant({
  query: 'weather today',
  geo_location: 'United States'
});

• query (string): The search query.
• limit (integer): The maximum number of search results to return. Maximum: 10.
• geo_location (string): Google's canonical name of the location. See more at Google Ads GeoTargets.

AI-Map

Basic usage

import { 
  OxylabsAIStudioSDK
} from 'oxylabs-ai-studio';

const sdk = new OxylabsAIStudioSDK({
  apiKey: 'your_api_key_here',
  timeout: 120000,
  retryAttempts: 3,
});

async function testMap() {
  try {
    console.log('Testing map...');
    
    const options = {
      url: 'https://www.freelancer.com/jobs',
      user_prompt: 'Extract tech job ads',
      return_sources_limit: 10,
      geo_location: 'US',
      render_javascript: false
    };
    
    const results = await sdk.aiMap.map(options);
    console.log('Map results:', JSON.stringify(results, null, 2));
  } catch (error) {
    console.error('Map error:', error.message);
  }
}

testMap();

Available Parameters

• url (string): The target URL to map and discover URLs from.
• search_keywords (string[]): Optional keywords to bias the mapping.
• user_prompt (string | null): Optional instructions to focus the mapping on relevant areas.
• limit (integer): The maximum number of URLs to return (default 50, max 10000).
• max_crawl_depth (integer): Maximum depth to crawl within the site (default 1, max 5).
• include_sitemap (boolean): Whether to include sitemap URLs (default true).
• allow_subdomains (boolean): Whether to include subdomains (default false).
• allow_external_domains (boolean): Whether to include external domains (default false).
• geo_location (string | null): The location to use (ISO2 or country canonical name).
• render_javascript (boolean): Whether to render JavaScript when mapping (default false).
• max_credits (integer | null): Optional cap on credits to spend for the run.

Running Examples

You can find more examples of each application here:

• Browser-agent Example
• AI-Crawler Example
• AI-Scraper Example
• AI-Search Example
• AI-Map Example