The unblocker is the real thing. No tuning, no fingerprint games — it just passes.

The js_scenario engine paired with XHR capture is excellent design. I need a couple of clicks per page to surface the data; Scrapfly runs the scenario and hands back every network call the page made, so you just read the site's own API responses as JSON instead of a rendered DOM. One request per page, no selectors to maintain.

Billing is transparent: every response carries its exact credit cost and failed requests aren't charged. The free tier was enough to validate the entire integration end to end before committing to a plan.

There's also an MCP server for AI, which is a nice touch — haven't needed it yet, but good to know it's there. Review collected by and hosted on G2.com.

Two small things.

Sessions stay locked for a few seconds after a request fails on Scrapfly's side (a 503 "no backend available"), so an immediate retry on the same session comes back as ERR::SESSION::CONCURRENT_ACCESS. Easy to work around — retry on a fresh session name — but it cost me a few minutes of head-scratching, and the docs could call it out.

And the unblocker's per-target cost isn't knowable up front. The docs say "30+ credits, scales with the target", which is honest, but you only learn your real number by running requests. A cost=true dry-run parameter exists, but an estimate on the dashboard per domain would be nicer.

Neither is a deal-breaker; the failed requests weren't billed either way. Review collected by and hosted on G2.com.