Description
N1: gauge group populations and seven-day histogram observations do not establish a matched acknowledgement-share denominator. Existing ratios may imply an unsupported cohort or causal outcome. No new live research or semantic change is granted by this task.
Acceptance Criteria
- #1 Owner decides separately qualified counts versus matched-cohort research; any retained quantitative ratio has independent population/window evidence, otherwise it is withheld without calling unknown zero.
Definition of Done
- #1 just test
- #2 just tf-validate
- #3 just check-identifiers and just no-em-dashes both return clean
Implementation Plan
D-DOC23 build: retain separately qualified OnCall group counts and 7d response bands, withhold unmatched acknowledgement ratios with reason. Implement rendered dashboard contract, gate just check, independent review and root-only land.
Implementation Notes
loop23: implementation attempts1, review-repair rounds0. Exact candidate78daf4b accepted after discriminating fail-first, real CLI-built dashboard readbacks, independent reviewer PASS and CodeRabbit complete0/all4files. Root landed4310826aabb10f4c78e56f58e73611aa2fc0ea25; composed just check0 (2614 passed,2 existing skips not passes,37Tofu) and exact required CI38254439412 all leaves success. Four unsupported shares are query-free withheld explanations, counts and7d bands separately qualified; layout unchanged. Evidence /Users/rob/repos/grafana-cloud-org-insights/codex/loop23-evidence/oncall/.
Final Summary
D-DOC23 applied: unmatched OnCall acknowledgement shares withheld rather than numeric zero or unqualified ratio. Count/window contract exercised through actual built artifacts; exact integrated local and required CI proof retained. No source, new series or live deployment change.