Summary
Hosted rendering, and both the text and a picture come back together.
The whole interface is a prompt plus a format — the service decides how to render the page and what to pull off it. That works when the job is "read this page for me" and stops working when you need to click through something first: there is no session, no navigation, and no second page.
What it is
An MCP server for Scraper.is, a hosted scraping service. A single tool takes a prompt describing what to pull and which URL, and returns the page in the format you asked for.
What you get
- scrape — one page fetched from a prompt describing what to extract
- Markdown or HTML when you want text, JSON when you want structure, and a screenshot for visual context
- Progress reported while a scrape runs, rather than a silent wait
Requirements
A Scraper.is account and an API key from its dashboard, supplied through your client config. Runs on Node, or installs into a desktop client through a registry.
Setup effort
One command plus a key — npm install -g scraperis-mcp, then supply credentials