What it is

coorl is a curl-compatible command-line HTTP client that mimics Chrome’s TLS fingerprint instead of advertising the standard OpenSSL signature that scripted tools send. It was written by Konstantin Lebedev. The scope is deliberately narrow: produce a ClientHello that hashes to a Chrome JA3/JA4 fingerprint, so the connection looks like a real browser at the layer anti-bot systems inspect first.

How it works

A normal HTTP client gives itself away during the TLS handshake, because the shape of its ClientHello is set by the TLS library it was built on, not by anything in the request code. See tls-fingerprinting for the mechanism. coorl performs the handshake the way Chrome does, so the JA3/JA4 fingerprint on the wire is a Chrome fingerprint rather than an OpenSSL one.

It does not run a headless browser, execute JavaScript, or carry Chromium. It mirrors curl’s flags (-X, -H, -d, -b, and so on), so adopting it is mostly a matter of swapping the command name. By design it implements only the most-used subset of curl’s flags over HTTP and HTTPS, not curl’s full protocol and flag surface.

TWSC experience

We have not tested coorl directly. It surfaced through Lebedev’s June 2026 write-up on TLS-layer bot detection. Conceptually it sits in the same niche as curl-cffi: swap the TLS stack so the handshake matches a browser. The difference is scope and form. curl-cffi is a Python library (and the engine under scrapy-impersonate and hrequests) with selectable Chrome and Firefox impersonation profiles. coorl is a standalone CLI focused on a Chrome handshake and a curl-compatible flag set. For pipelines already in Python, curl-cffi remains the documented TWSC choice. coorl is relevant when what you want is a drop-in curl replacement on the command line.

Known limitations

  • It addresses only TLS-layer detection, the JA3/JA4 check that blocks a request before it reaches the origin. It is not a CAPTCHA solver and does not execute JavaScript challenges, so any target that gates content behind a full browser challenge still needs a real browser.
  • It implements a subset of curl’s flags, so it is not a complete drop-in for every existing curl invocation.
  • Untested by TWSC. The claims here come from the author’s documentation, not our own benchmarks.

Sources


The Web Scraping Wiki is compiled from the tests and research published on The Web Scraping Club, the newsletter for web scraping professionals. Subscribe to get the next article in your inbox.