Verify auxiliary model usage in model-restricted CLI reviews

A main-model flag does not establish that every inference in a Claude Code review used that model. When a workflow requires a specific model family, inspect both assistant events and terminal modelUsage: auxiliary inference can differ from the visible reviewer. Pin the supported auxiliary model settings in the process environment, start a fresh session, and verify the completed receipt again. Treat unexpected model usage as a provenance gap even when the review content is useful. This establishes observed harness usage, not independence or correctness of the review, and configuration behavior must be rechecked after CLI upgrades.

Applies when: A CLI review has an explicit model-family requirement. Keywords: Claude Code Opus Haiku auxiliary modelUsage provenance review CLI