When PowerShell quietly corrupts your source files

While building One Ocean I was writing files from PowerShell scripts and here-strings. Some files came out with mangled characters, such as ₦ where ₦ should be. The next build would fail, or worse, still pass and ship broken copy.

Cause

Windows PowerShell 5.1 does not write plain UTF-8 by default. Depending on the cmdlet you get UTF-16 or the system ANSI code page, and -Encoding UTF8 adds a byte-order mark. Tools that expect clean UTF-8 then misread the file.

Fix

Write through .NET with an explicit UTF-8 encoder that has the BOM turned off:

$utf8 = New-Object System.Text.UTF8Encoding $false
[System.IO.File]::WriteAllText($path, $content, $utf8)