note: -Yforce-sbt-phases is a private flag and thus gives no guarantees.

That being said, it seems the error pops up independently from the option being passed when compiled with Scala CLI + Bloop... so I'm guessing it may be passed implicitly somewhere 🤔

scala-cli compile repro.scala -S 3.nightly
# Compiling project (Scala 3.7.1-RC1-bin-20250425-fb6cc9b-NIGHTLY, JVM (23))
# Warning: No class, trait or object is defined in the compilation unit.
# The incremental compiler cannot record the dependency information in such case.
# Some errors like unused import referring to a non-existent class might not be reported.
# Compiled project (Scala 3.7.1-RC1-bin-20250425-fb6cc9b-NIGHTLY, JVM (23))

the warning goes away when compiled with --server=false

 scala-cli compile repro.scala -S 3.nightly --server=false

and comes back when -Yforce-sbt-phases is passed alongside it.

scala-cli compile repro.scala -S 3.7.nightly --server=false -Yforce-sbt-phases
# No class, trait or object is defined in the compilation unit.
# The incremental compiler cannot record the dependency information in such case.
# Some errors like unused import referring to a non-existent class might not be reported.
# 
# 1 warning found

cc @tgodzik

note: -Yforce-sbt-phases is a private flag and thus gives no guarantees.

I discovered the issue in sbt project (without explicitly using that flag) so my understanding the sbt enable that Phase as part of enabling incremental compilation. I only used that flag to reproduce the issue in a simple single scalafile without involving sbt.

Right, so the flag enables the warning (and is an easy way to reproduce with just the compiler script), as it runs the ExtractApi and ExtractDependencies sections... which are also run by zinc via SBT/Bloop.
but the core issue is that this warning shouldn't pop up in those phases in the first place.
so everything's fine regarding the -Yforce-sbt-phases flag, got confused there.