General copy review
Purpose
Review user-facing prose embedded in TypeScript, JavaScript, JSON, YAML, or another non-MDX source
file. This is a small adapter to the shared version 2 review model in
content-quality-rubric.md.
Rubric version
The current version is 2.0.
Extract the review scope
Before reviewing, enumerate every string value written for a person to read: headings, body copy, labels, help text, and similar output. Record each field key or path with its value. This extraction list is the complete review scope.
Exclude:
- object, property, variable, function, and type names;
- CSS classes, asset names, file paths, URLs, and slugs;
- enum-like values that select a program state rather than communicate with a reader;
- implementation comments that are not themselves published copy.
Short labels remain reviewable. The test is who reads the string, not its length.
Apply the shared review model
Apply these four lenses from the MDX rubric to the extracted copy:
- Clarity
- Ambiguity control (stored as
ambiguity) - Terminology
- Actionability
Structure is not applicable to isolated strings in an arbitrary source file. Do not manufacture a structure finding or substitute another lens.
Assess each applicable lens as Strong, Mixed, or Weak. Omit structure from lensAssessments;
do not store a not-applicable value. Use the same Finding standard, holistic Quality rating,
editorial guidance, and reporting contract as MDX.
Purely descriptive or evocative copy is not defective for lacking an instruction it never promised
to provide.
General-copy reviews do not use MDX mechanical checks, MDX content expectations, Coverage, or pipeline-draft exceptions. Their version 3 snapshot uses the same schema as an MDX review.