MCP server
Give your AIeyes on the live web.
Ask your AI assistant to research live pages, extract data, compare content, and track changes through your ScrapeNest account. Add one URL, sign in, and get 20 tools for the live web.
20 tools/3 engines/no install
I created a monitor. It will check the page every hour.
monitor created · next run 10:00
Latest change
- Annual plan · €49 / month
+ Annual plan · €59 / month
Connect once. Then ask.
Add the server to your agent, sign in with your ScrapeNest account, and start with a live page.
Add the server
Paste https://mcp.scrapenest.com/mcp into your client as a custom connector, or run one command in Claude Code.
Sign in
Your client opens a browser. Sign in with your ScrapeNest account and approve the connection. No API key changes hands.
Ask
"Scrape this page and summarize it." Your assistant picks the right engine, runs the job, and returns the result as Markdown.
What you can do
Give an agent a job. Get an answer with evidence.
Research a live page
Ask a question about a page today. ScrapeNest returns readable Markdown instead of navigation, scripts, and ads.
“Compare this page with a competitor and show me what changed.”
Extract usable data
Turn product pages, listings, and search results into structured JSON the agent can analyse or pass into your workflow.
“Extract every plan's name, price, and limits from this page.”
Track a change
Create a monitor in chat, then come back to the actual diff instead of another vague alert.
“Watch this pricing page every hour and show me the diff.”
Not just a scrape command
The full ScrapeNest platform, through a chat window. Same engines, same billing, same extraction hooks.
Markdown, not HTML
Raw HTML wastes context tokens on nav, scripts, and ads. The MCP server strips it down to readable content, capped at 40k characters so a big page cannot swallow your window.
Monitoring from chat
Ask your assistant to watch a page on a schedule. It creates the monitor. Next session, ask for changes and get a diff.
Three engines
Light (HTTP, 1 credit), Standard (Chromium, 5 credits), Stealth (anti-bot browser, 30 credits). Engine access follows your plan.
Extraction in the tool call
Pass CSS selectors to the extract tool and get structured JSON back. No second pipeline to maintain.
Browser actions
Click, fill, scroll, wait. Standard and Stealth engines run the page's JavaScript and drive interactions before capture.
RAG-ready
The crawl tool ingests many pages concurrently and returns chunked Markdown. Feed it straight to your vector DB.
Side-by-side compare
Scrape two URLs and get a unified diff of their content. Compare competitors, track deployments, or spot regressions across environments.
Usage analytics
Ask for your success rate, job counts, and average duration. The usage tool pulls hourly stats so you can spot problems without leaving your editor.
Job history and retry
Search past jobs by status, engine, tag, or date. Found a failure? Resubmit it with one tool call, optionally upgrading the engine.
Works with the agents you already use
One hosted server. One account-bound sign-in.
Use the hosted server from CLI agents and web agents alike. Add the endpoint, sign in with your ScrapeNest account, and give your agent a live-web task. Claude is one setup path, not the limit.
Security
Your account, your org, no shared secrets
The server is an OAuth 2.1 resource server. Every caller signs in with their own ScrapeNest account, and the token is verified before a single job is queued. No long-lived ScrapeNest API key sits in a client config file.
- Sign in with the ScrapeNest account you already have - no key to copy
- Every tool call follows the caller's organization, role, and plan
- Organizations are isolated; access inside an organization follows its roles
- Access follows the account, so there is no copied API key to chase down later
- EU-only data residency applies to MCP jobs exactly like API jobs