Rulvar

Appearance

Rulvar API reference

Rulvar API reference / @rulvar/core / ClaimCoverageGrade

Type Alias: ClaimCoverageGrade ​

ts
type ClaimCoverageGrade = 
  | "full"
  | "vacuous"
  | "partial"
  | "coverage-capped"
  | "critical-uncovered"
  | "judge-declined"
  | "judge-failed";

Defined in: packages/core/src/orchestrator/consistency.ts:667

The claim-coverage grade (RV1702): one closed vocabulary a consumer reads INSTEAD of inferring semantic health from an empty findings array. The eighteenth comparison benchmark's run reported completion: 'complete' with contradictions: [] while the judge had seen 40 of 144 citing sentences and said so only in counts a reader had to interpret; three material falsehoods rode that gap. The grade names the verification posture outright:

  • 'full': every citing sentence the draft carries had at least one judged pair, nothing was cut by a bound, no declared critical anchor was missed, and the judge (when needed) settled ok.
  • 'vacuous' (RV2508): the draft carried NO citing sentence, so the configured pass verified nothing. This used to grade 'full' on the reasoning that saying 'partial' would imply a subset was chosen, which is true and beside the point: 'full' is the strongest word in the vocabulary and it was standing over a denominator of zero, the same silent green the grade exists to abolish, at its extreme.
  • 'partial': the pass verified a strict subset: the pair bound truncated the fold, a run-facts bound truncated the run-claim pairs, or citing sentences exist that no judged pair covers.
  • 'coverage-capped' (RV4404): the pass ran under a DECLARED coverage target and the hard pair ceiling still cut selection the target wanted. Distinct from 'partial' because the cause is the CONFIGURED max, not the pool: the seventh comparison run declared full coverage, folded its pairs truncated at the ceiling, and reported 23 uncovered citing sentences as if the text were the problem. The honest grade names the ceiling so the refusal (and the operator) fix the config, not the document.
  • 'critical-uncovered': at least one DECLARED critical anchor got no judged pair; stronger than 'partial' because the caller named exactly these claims as the ones that must not go unverified.
  • 'judge-declined' (RV2508): the judge invocation was refused ADMISSION and never dispatched (RV2106), so nothing was judged at all. It ranks with a failed judge and above everything the counts could say, because those counts describe a pass that did not happen; before this the flag was invisible to the grade and a declined judge over a citation-free draft graded 'full'.
  • 'judge-failed': the judge invocation did not settle ok, so nothing was judged at all; every other reading of the meta is moot.

Precedence is the order above, strongest last. The helper is pure and total over metas written BEFORE the grade shipped, so a consumer can grade a persisted outcome from an older engine.