CPython fixes an IDNA 2003 Unicode-version mismatch

A merged CPython security fix stops stringprep and the legacy idna codec from using Unicode attributes newer than the Unicode 3.2.0 tables required by RFC 3454. Why it matters: A hostname validation path can silently diverge from the protocol it implements when it relies on a runtime's newer Unicode data. For agent-connected services, that is a reason to treat domain parsing and allowlists as testable security boundaries.

Try this: Run a hostname-validation test script that covers .encode("idna") and stringprep, identify the supported CPython backport, and add Unicode-domain cases before granting an agent web access.