# Changelog

## Score

- CAI 57 → 58 (+0.7)
- Rubric changed (rubric-2026.10.5 → rubric-2026.11.2) — scores are not directly comparable.

## Lenses

- Code Health 74 → 74 (+0.0)
- Architecture 100 → 100 (+0.0)
- Maturity 49 → 54 (+5.3)
- Readiness 50 → 50 (+0.0)
- Security 77 → 69 (-7.4)

## Resolved (4)

- Duplicate error recording logic across `Attribute` and `ErrorRecorder`. `Attribute.validation_error` seems to be a low-level helper that might duplicate the work of `ErrorRecorder.validation_error`. The existence of both suggests a split in responsibility that is not clearly defined.
- Inconsistent method naming for URI operations. Some methods use the object as the first argument (`parse(uri)`), while others appear to be instance methods or have ambiguous signatures in the list (e.g., `URI.unescaped_path()` listed without args in the instance context). The distinction between `normalized_uri`, `normalize_ref`, and `absolutize_ref` is unclear without documentation, suggesting potential overlap in intent (resolving relative URIs).
- Inconsistent serialization/naming conventions for error output. `to_string` is non-standard (usually `to_s` or `to_str`), `to_hash` is standard, but `message_with_schema` is a specific accessor rather than a serialization method. This mixes serialization intent with data access.
- Severe naming inconsistency and redundancy in validation entry points. There are 10 different methods for essentially the same operation (validating data against a schema), differentiated by obscure suffixes (!, 2, _json, _uri) and prefixes (fully_). It is unclear what the semantic difference is between `validate`, `validate_json`, `validate_uri`, `validate!`, `validate2`, `fully_validate`, etc. This creates a high cognitive load and risk of calling the wrong variant.
