# Changelog

## Score

- CAI 63 → 63 (-0.1)
- Rubric changed (rubric-2026.10.5 → rubric-2026.11.2) — scores are not directly comparable.

## Lenses

- Code Health 67 → 67 (+0.0)
- Architecture 100 → 100 (+0.0)
- Maturity 49 → 49 (+0.0)
- Readiness 74 → 74 (+0.0)
- Security 99 → 98 (-0.7)

## Resolved (5)

- Inconsistent logging signatures. The `Einhorn` module methods require a `tag` argument, while the `AbstractTextDescriptor` instance methods do not. This forces callers to know the context (module vs. object) to determine if a tag is required, breaking consistency in logging API usage.
- Inconsistent naming strategy for related operations. `replenish_immediately` and `replenish_gradually` are explicit variants, but `replenish` (presumably the default behavior) lacks a qualifier. This makes the API surface feel disjointed compared to the explicit naming of the other two.
- Inconsistent parameterization for upgrade strategies. `full_upgrade_smooth` and `full_upgrade_fleet` are zero-argument methods implying fixed strategies, while `full_upgrade` takes an `options` hash. This creates confusion about whether `full_upgrade` is a generic dispatcher or a specific strategy, and why some strategies are separate methods while others are options.
- Off-boarding risk: anonymized user #1
- Redundant factory methods across state types. All three state classes expose a class method `default_state()` with identical intent (creating a default instance). This suggests a shared interface or base class method should be used instead of duplicating the signature.

## New (1)

- Off-boarding risk: anonymized user #1
