Built by Metorial, the integration platform for agentic AI.
Run an AI browser agent that autonomously operates a browser to complete a natural language task. Supports multiple agent types: **Browser Use** (fast, lightweight), **Claude Computer Use** (complex tasks), and **OpenAI CUA** (general-purpose). Starts the agent task, waits for completion, and returns the final result and step-by-step history.
Scrape content from a single webpage and return it in structured formats (markdown, HTML, links, screenshot). Starts a scrape job, waits for completion, and returns the extracted page content. Supports configurable session options like proxy, stealth mode, and CAPTCHA solving.
Retrieve a saved browser profile by ID, including its name and timestamps. Call list_profiles to discover profile IDs.
Retrieve details of an existing browser session including status, connection endpoints, and live view URL.
Search the web using Bing and get structured search results. Returns titles, URLs, and descriptions for matching web pages. Supports advanced filtering by file type, site, date, and geographic location.
Use AI to extract specific structured data from one or more webpages. Provide URLs and a natural language prompt describing what data to extract, and optionally a JSON schema for the output structure. The AI transforms unstructured web content into structured data matching your requirements.
List browser sessions with pagination. Returns session details including status and timestamps.
Create a new browser profile for persisting login state, cookies, and browser data across sessions. Attach the created profile to sessions to maintain authenticated state.
Retrieve the current status, final result, and execution history of an existing browser agent task. Use the same agent type that started the task.
Scrape content from multiple webpages simultaneously in a single batch job. Returns results for each URL in the same order as provided. Each page's content is available in markdown, HTML, links, or screenshot format.
List browser profiles with optional filtering by name and pagination.
Permanently delete a browser profile. This removes all stored cookies, login state, and browser data associated with the profile.
Crawl an entire website starting from a URL, following links to subpages and extracting content from each page. Starts the crawl job, waits for completion, and returns all crawled page content. Useful for gathering data across an entire site.
Create a new cloud browser session with configurable options. Returns connection endpoints (WebSocket, WebDriver) and a live view URL for real-time observation. The session can be connected to via Playwright, Puppeteer, or Selenium.
Retrieve recording URLs for a stopped browser session. Can get rrweb DOM-based replay recording URLs and MP4 video recording URLs. Video recordings must be explicitly enabled when creating the session.
Stop an existing browser agent task to halt its actions. Use the same agent type that started the task. Call get_browser_agent_task afterward to inspect the final state.
Stop a running browser session. This terminates the browser and releases resources. Session recordings and downloads become available after stopping.