Add a web-scraping API to Claude Code. One command.
Remote MCP over HTTPS. Not an official Claude plugin, not an npx package. Paste the command, restart, ask a URL a question.
Claude Code talks to Plainscrape the same way Cursor does: HTTP MCP, header x-api-key. Nothing to install: the server is already up.
Paste this
Set me up on Plainscrape: read https://www.plainscrape.com/skills/onboard.md (curl -fsSL it, or open it) and follow it. Do the steps yourself, and only ask me what it tells you to ask.
Or add it yourself, then type /mcp, pick plainscrape and choose Authenticate — a browser tab opens; click Allow:
claude mcp add --transport http plainscrape https://www.plainscrape.com/mcp
Already have a key?
claude mcp add --transport http plainscrape \
https://www.plainscrape.com/mcp \
--header "x-api-key: ps_live_YOUR_KEY"
Restart Claude Code once. Then: “Use Plainscrape ask on this URL — title, price, rating. Confirm the shape is record and coverage shows it complete.”
Three tools, not forty
The agent gets ask, answer and account. It reads the pages you point it at. budget defaults to 10 pages for rows and 5 for a reply (2–20); trial asks use up to 10 pages, paid plans any number. Long answers stop at 12 minutes and return what they have.
Output is JSON, with prices and ratings as numbers.
FAQ
claude mcp add --transport http plainscrape https://www.plainscrape.com/mcp --header "x-api-key: ps_live_YOUR_KEY". Restart Claude Code. This is HTTP MCP, not an official Claude plugin.
Three: scrape (get data — page, list or answer, picked from your words), results and account. It reads the pages you point it at.
Yes. The tool returns the fields you asked for, not the page, so your agent's context stays small. budget caps each ask: 10 pages by default and on trial, no per-ask cap on paid plans.
Per successful page fetched. Failed pages are free. Same key and limits as REST. 100-page trial, 14 days, no card.