Phase 2b (after Phase 2, before Phase 3): instrument per-phase timings, baseline a
representative recipe set (cold vs warm), attribute where time goes (Pareto), then try
improvements as controlled before/after experiments and keep measured winners — image
pull cache/pre-pull, readiness-wait tuning, dedup deploy cycles, warm/shared infra
(isolation-proven), runner caching, concurrency sizing, vCPU. Speed never weakens tests
or isolation (Adversary re-measures + re-verifies). Phase 3 now follows 2b. Linked in README.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phase 3 (after Phase-2 DONE, manual transition): compute a per-run quality LEVEL, post an
image-forward Gitea PR comment in the YunoHost shape (marker + status/level badge + a
rendered summary card containing a real app screenshot, linking to the run), and polish the
overview dashboard to a ci-apps.yunohost.org look/feel with per-recipe level badges +
screenshots. Reuses the Phase-1 dashboard/bridge/Playwright; presentation never changes the
verdict; no secrets in any artifact; cosmetics never block the pipeline. Linked from README.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
