About
One person, one API, kept free on purpose.
webscrapingapi.dev is built and run by Jaeyoung, a developer who kept rewriting the same fetch-and-parse code for side projects and decided to run the version everyone could share.
What this is
A web scraping API for developers and AI agents. Send a URL and get the page back as HTML, text, Markdown or JSON; or call an endpoint that already knows a site (Amazon, YouTube, the app stores, Hacker News, Wikipedia, GitHub) and returns clean JSON. The documentation describes exactly what each endpoint does today, not a roadmap.
Why it is free
Because the cost is small when it is capped. Every account gets 50,000 credits a day, 60 requests a minute and 5 keys, and the service as a whole has a daily budget it will not exceed. Structured results are stored and shared, so the second person asking for the same page costs nothing. There are no plans, no card and no trial that ends; the cap is what keeps it that way.
How site endpoints get built
People post the site they need on the board. Each request is checked by hand: can the page be fetched reliably from a cloud address, and does it need anything the service refuses to do, such as solving CAPTCHAs or signing in? If it can be built, it becomes a /v1/structured/… endpoint that takes only the part that changes. If it cannot, the request is closed as Not possible with the reason. Both outcomes are public unless the request was private.
What is logged
For every request: the target hostname, the status, the timing, the credits used and which key made it, kept for 30 days so you can see your own usage. Raw pages are never stored. Parsed results from site endpoints are stored so they can be shared, without the page they came from. The privacy policy has the full list.
What it will not do
- Solve CAPTCHAs or bypass dedicated anti-bot vendors. When a site blocks cloud traffic, the API says so (SITE_BLOCKED) and does not charge for it.
- Sign in anywhere on your behalf. Only public pages are reachable.
- Pretend a broken parser is working. Layout changes surface as SITE_PARSE_FAILED, are refunded, and get fixed once for everyone.
Contact
The fastest route is the board: questions and site requests are answered there, in public, so the next person finds the answer too.