[quality] Recommend adopting knuckle's coverage patterns across org repos

[kubestellar-hive]

Bluefin 🦖  ·  issue pipeline
─────────────────────────────────────────────────
  ▶  filed      report received
  ·  approved   —
  ·  queued     —
  ·  claimed    —
  ·  done       —
─────────────────────────────────────────────────
  report:       missing     ·  confirms: 0
  area:         —         ·  priority: —
  next action:  same bug? ujust confirm 746

---

KNUCKLE 🦖  ·  issue pipeline
─────────────────────────────────────────────────
  ►  filed      needs triage
  ·  approved   signed off by a maintainer
  ·  queued     waiting for a contributor to claim
  ·  claimed    —
  ·  done       —
─────────────────────────────────────────────────
  domain:       —         ·  priority: —
  size:         —         ·  next: /approve to queue

---## Finding

knuckle has exemplary test infrastructure that other repos in the org should adopt:

• Per-package coverage thresholds (96–100%) via just cover-check
• Codecov integration with strict project target (95%)
• Flaky test detection (5x re-run in CI)
• Nightly coverage snapshots

By contrast:

• bootc-installer has a 47% coverage floor (target 45%) with major modules entirely untested
• testsuite has 75% target but some step modules lack unit tests

Recommendation

Consider extracting knuckle's coverage patterns into a reusable org-wide template:

1. Per-module coverage gates — a script or just recipe that checks coverage per top-level package/module with individual thresholds
2. Codecov config template — standardized codecov.yml with project target ≥70%, patch 70%, and non-informational status checkstarget
3. Flaky test detection — the 5x re-run pattern from knuckle CI could be a reusable composite action

This is a tracking/advisory issue — no code changes needed in knuckle itself.

Priority

• Impact: medium — org-wide test quality improvement
• Effort: low — documentation and template sharing

---

Filed by quality agent (hold-gated mode)