[Model behavior] Verbose output after repeated explicit instructions to be brief
Status Open
Maintainer reply None cached
Activity 0 comments · opened Oct 2, 2026
The standing instructions say: verdict first, 1-3 lines when clean, detail only for blockers. Across the session the model kept producing long tables, multi-section reports and narration of its own steps, was corrected many times ("YOUR VERBOSE!", "ONE LINE"), and regressed within a few messages each time.
Model: Claude Opus 5.5 (claude-opus-5-5[1m]), Claude Code CLI, one long working session on 2026-10-02.
Context: maintainer of an open-source cloud-service emulator; the work is pull-request review and fixes under a written procedure loaded into every session (project instruction file + persistent memory). No personal data, repository names, PR numbers or handles are included.
Raised with Claude Code.