Cloudflare
Cloudflare
Browser Run: a route per output, crawl jobs, links and the accessibility tree.
- Supports
- 10 of 13 operations
- Scrape
- one call
- Needs
- CF_API_TOKEN and CF_ACCOUNT_ID
- Endpoint
- api.cloudflare.com
Operations
- scrapeone call
- screenshotone call
- elementone call
- pdfone call
- crawlone call
- linksone call
- a11y treeone call
- extractone call
- searchno
- sessionsone call
- cdpone call
- navigateno
- evaluateno
Recorded
- Came back
- markdown
- Size
- 886 characters
- Time
- 1,065 ms
Thisdomainisforuseindocumentationexampleswithoutneedingpermission.Thisisnotaservice,avoidrelyingonitfortestingandmonitoringpurposes.
Access
- Create
create("cloudflare") - CLI
browsers scrape https://example.com -p cloudflare - Tool
provider: "cloudflare"
Cloudflare's Browser Run has a REST route for nearly everything: /markdown, /content, /screenshot, /pdf, /links, /json for extract, /accessibilityTree, and /crawl for jobs. No session needed for any of them. The operations band above adds them up.
It wants two values: CF_API_TOKEN and CF_ACCOUNT_ID. CLOUDFLARE_API_TOKEN and CLOUDFLARE_ACCOUNT_ID work too.
Ask for more than one output and it renders once: formats: ["markdown", "html"], or screenshot: true, go through /snapshot and come back together. A scrape also fills title and statusCode.
The markdown in the recording has the words of example.com glued together, Thisdomainisforuse…. Jina Reader spaced the same page out letter by letter, so the page is doing something odd with its markup and two converters read it two ways. Other pages come out normal.
browser: "kitesurf" switches to Cloudflare's Kitesurf engine. Without it you're on Chromium.