Check Reference

kapi check runs a set of checks over source content and reports each finding under a stable rule id, such as hygiene.doubled-word. The list below gives every check with the rule ids it reports; each check's page says what each rule reports, at which severity, and what fixes it.

The rule id's first segment is the rule family, which differs from the name the code registers the check under: the content-lint check reports under hygiene. A finding names the family, so the family is what to look up here.

The list is generated from the code: the checks, their families and their rule ids come from the check package, and a CI drift gate fails the build when the committed dataset and the code disagree.

Check	Rule ids	What it checks
Content Lint	hygiene.empty hygiene.leading-whitespace hygiene.trailing-whitespace hygiene.double-spaces hygiene.doubled-word hygiene.control-char	Report the text-hygiene defects of a single source text, covering emptiness, stray whitespace, spacing and repetition.
Length Check	length.max-chars-exceeded length.max-words-exceeded	Hold each source block to an absolute character or word limit.
Pattern Check	pattern.forbidden-pattern pattern.pattern-missing	Require a pattern to appear in every source block, or forbid one from appearing at all.
Comment Style	comment.sentence-length comment.length comment.density	Hold the comments in source files to the sentence and comment word limits a voice profile sets for them.

The voice and terminology checks are configured by the project's context, not by a flag, and their rules are the ones a voice profile or a terms store declares. Checks describes the report they share with the checks above, the verdict, and what fails.