Pagewise turns any URL into clean markdown or schema-valid JSON, through a simple API or straight inside your agent.
Cleaned 1 page. Removed the menu, a promo and the cookie banner.
# Introducing Arm-7, our fastest picker yet Arm-7 sorts up to 1,400 parcels an hour and ships to customers in November. | Spec | Arm-7 | |------------|-----------------| | Throughput | 1,400 parcels/h | | Payload | 7 kg |
One API key for scraping, searching and extracting. Every response tells you what it cost.
Done. I cleaned up the Arm-7 launch post: removed the menu, a promo and the cookie banner, and kept the specs table.
Introducing Arm-7, our fastest picker yet
acme-robotics.example/blog/arm-7
Markdown · 1 creditScrape
Navigation, footers, cookie banners and ads are removed. Tables, code blocks and links survive as clean markdown.
This page needed JavaScript
RenderedRendering
Headless Chromium when a page needs it, at the same 1 credit.
warehouse picking robots
Introducing Arm-7, our fastest picker yet
acme-robotics.example
The state of warehouse picking in 2026
logistics-weekly.example
Choosing a robot for small warehouses
parcelhub.example
Search
Turn on scrape_results and every result comes back as markdown.
Extract
Send a URL and a JSON Schema. The response validates against your schema, so your code gets data instead of prose.
Failed requests are never billed. Every call shows up in your usage log.
The things you would otherwise build, debug and babysit yourself. Pick one to see it from your side of the API.
/v1/scrape400// request { "url": "http://169.254.169.254/latest/meta-data" } // response { "error": { "code": "url_not_allowed", "message": "Private and reserved IP addresses are not allowed" } }
/v1/scrape403// request { "url": "https://acme-robotics.example/private/", "respect_robots": true } // response { "error": { "code": "blocked_by_robots", "message": "The site's robots.txt disallows fetching this URL" } }
/v1/usage200extractCredits reserved−5extractUpstream returned 502failedextractCredits refunded+5scrape200 · 412 ms−1pw_live_••••3f9aRESTMCP# REST curl https://{{API_HOST}}/v1/scrape \ -H "Authorization: Bearer pw_live_••••3f9a" \ -d '{"url": "https://acme-robotics.example"}' # MCP, in Claude Code claude mcp add --transport http pagewise \ https://{{API_HOST}}/mcp \ --header "Authorization: Bearer pw_live_••••3f9a"
/v1/scrape200"metadata": { "title": "Introducing Arm-7", "site_name": "Acme Robotics", "language": "en", "author": "Maya Chen", "published_time": "2026-09-14T09:00:00Z", "canonical_url": "https://acme-robotics.example/blog/arm-7" }
Checkout summarypaidUsage log
Each request lands in your log with its status and its cost. When one fails, the credits come straight back.
Calls per hour
scrapeacme-robotics.example/blog/arm-7−1search“warehouse robot pricing”−4extractshop.northwind.example/p/arm-7−5extractslow-store.example/item/88refundedscrapedocs.fabrikam.example/api−1Start free. Upgrade when your agents need more pages. Cancel anytime.
About 2,500 credits a month. Hobby covers it.
Free
Fits your estimate$0 /month
500 credits every month
No card required
Request accessHobby
Fits your estimate$16 /month
5,000 credits every month
$3.20 per 1,000 credits
Request accessStandard
Most popularFits your estimate$79 /month
100,000 credits every month
$0.79 per 1,000 credits
Request accessWhat a credit buys
/v1/scrape1 credit per page/v1/search2 credits, +1 per scraped result/v1/extract5 credits per page/v1/usage0 always freeNeed more than 100,000 credits a month? We'll set up a custom plan with volume pricing.
Talk to usThe short answers to what developers ask us most.
Didn't find your answer? Write to {{CONTACT_EMAIL}}
The credits reserved for it go back to your balance automatically, and the failure appears in your usage log with its error code.
Your balance is topped up to your plan's amount each month. Unused credits stay, but the top-up never takes you above the plan amount.
Yes. Pages that need JavaScript are rendered in headless Chromium automatically, or you can force rendering on any request.
Requests to each domain are rate-limited across all customers, robots.txt can be enforced per request, and site owners can opt out. See PagewiseBot.
Any MCP client that supports remote or local stdio servers, including Claude Desktop, Claude Code and Cursor.
Paddle is our merchant of record. It handles checkout, invoices, sales tax and VAT, and accepts cards and PayPal.
We're onboarding developers in small batches. Request access and we'll send your API key with 500 free credits.