The model produced awkward AI-tells like "Als Muttersprachler kommuniziere
ich serviceorientiert und klar - sowohl mit Anwendern als auch im Team".
Add an explicit style rule (system prompt, shared prose guidance, and the
AI-reply prompt) that forbids empty buzzword pairs and self-labels such as
"Als Muttersprachler ...", quoting the offending sentence as a negative
example. Language proficiency belongs only in the CV's Sprachen section;
every trait must come out of a concrete task or result.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>