Agent Web-Access

Web scraping API for AI agents: read any live web page as clean markdown, text or HTML. Renders JavaScript in our own headless browser, then strips navigation, ads and boilerplate. Use it when a plain HTTP GET returns empty or garbled HTML: single-page apps, dynamic feeds, or content that only appears after scripts run. Also returns the page's HTTP status. Page text is flagged untrusted so an agent treats it as input, never as instructions.

Targeted web scrape: pass a CSS selector and get back only the matching elements from a fully JavaScript-rendered page. Each element returns its tag, text content and every HTML attribute. Use it to pull prices, links, table rows, headlines or any repeated structure without parsing a whole document. Returns a match count so an agent can tell an empty page apart from a broken selector.

Extract text from an image by URL (OCR): PNG/JPG/WEBP/TIFF/BMP of a screenshot, scan, receipt, chart or photo -> the readable text, via the Tesseract engine. Keyless, no LLM. Use it to read text an agent can only see as an image. SSRF-guarded; max ~10MB.

Render any live web page to a portable PDF, returned as base64. JavaScript runs first, so single-page apps and dynamic content render correctly instead of coming out blank. Use it to archive a page, capture a receipt, invoice or report, snapshot terms or pricing that may change later, or hand a human a fixed copy of a URL.

Company profile from a domain, keyless: registrar and domain age (RDAP), email provider from MX, SPF senders (SendGrid, HubSpot...), DMARC policy, verified SaaS from TXT records, plus homepage name, description, social links, public contact emails and tech stack. Use it to qualify a lead, vet a vendor or spot a days-old phishing domain. No LLM; unregistered domain = not charged.

Pixel-accurate PNG screenshot of any live web page, rendered in a real browser at a chosen viewport width, viewport-only or full-page. Returns base64 PNG. Use it when an agent needs to SEE a page: visual verification of a deploy, layout and design capture, confirming a rendered chart or map, or feeding an image to a vision model that cannot read HTML.