[Model behavior] Contributor claims used as provider evidence without saying so at verdict time

The procedure says contributor captures/tests are claims, never ground truth, and every component needs its own normative citation. For several components (error header values, error messages, a replacement behaviour that contradicted the official reference) the verdict rested only on the contributor's capture; this was disclosed only later, when the user asked if everything was per the provider's behaviour.

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.