Type
Enhancement (test cleanup, non-blocking)
Problem / Value
The tests covering the built-in model table carry one check that no longer adds protection, and one note that states a cause the evidence does not establish. Neither affects how the product behaves, but both make the tests harder to trust and maintain.
Context
Deferred from the review of #19 (companion documentation: https://github.com/simurg79/roo-vault/pull/648). Both items were recorded as non-blocking, so they are tracked here rather than changed inside that pull request.
Two items, suitable to fix together:
- One check confirms that none of the refreshed model families still carries the old borrowed placeholder figure. Each of those families is already pinned to its exact expected value by a later check, so the earlier one cannot fail unless the exact pin fails first. It can be removed.
- A pre-existing note on one model row describes its smaller input limit as an enforced ceiling and the larger figure as an advertised window. The two numbers genuinely differ, but that explanation was never measured. The wording should be softened to state only what is known.
Constraints/Preferences
- Test-only cleanup; the published model figures must not change.
- Keep the exact pins, which provide the real coverage.
- Low priority.
Type
Enhancement (test cleanup, non-blocking)
Problem / Value
The tests covering the built-in model table carry one check that no longer adds protection, and one note that states a cause the evidence does not establish. Neither affects how the product behaves, but both make the tests harder to trust and maintain.
Context
Deferred from the review of #19 (companion documentation: https://github.com/simurg79/roo-vault/pull/648). Both items were recorded as non-blocking, so they are tracked here rather than changed inside that pull request.
Two items, suitable to fix together:
Constraints/Preferences