F8 Computation or token-level error

Arithmetic, counting or date maths went wrong.

How counterfix finds it

Judge writes one expression for the claimed result; counterfix evaluates it with a whitelist-only evaluator and compares, so the judge's own arithmetic is never trusted.

Fixes it tries

In this order (a model profile can reorder them). Each one is kept only if it passes 2 of 3 re-runs.

Intervention	Change
work_it_out	Ask for the calculation step by step, checked, before the final answer

When no fix verifies

• Route the calculation to a code or calculator tool