Better socket life time handling in core for HTTP connections

Plus some other minor goodies

Significantly better backend codegen

Which combined with the new Hyperlambda LLM creates stunning results!

Using GPT-5.4 in default sw-gen model

The primary vibe coding model is now using GPT-5.4 by default. Notice, you'll need to have access to it or downgrade your model during setup.

Minor changes

v22.7.6

Update Version.cs

Support for [io.file.patch]

Which of course drastically reduces token consumption during vibe coding, if you can get the LLM to invoke patch function instead of save function.

More stable headless browser support

Due to checking operating system before trying to launch one!

Better Terminal and Python support

Better Terminal and Python support

Test release

v22.7.2

Update HttpFileExecutorAsync.cs

ARM64 installation of chromium

v22.7.1

Fixing headless chromium installation for ARM64

Fixing .Net Core 10 chromium bug

v22.7.0

Update Version.cs