[twoimo]

Why

The next protected develop → data promotion has conflicting historical merge ancestry. The two conflicts are in the nightly source-contract test and local Supabase baseline provisioning.

Change

• Join the exact reviewed data head 20fd13c9ff4026eff82f87a62983e258d91a4300 into the develop candidate based on 8a75965476b9d70084a56e72fe8d514e8020c8a8.
• Preserve both branches' intended behavior. The resolved runtime, tests, and configuration are byte-identical to the already-verified develop tree; the classifier and platform-preserving baseline behavior remain intact.
• Update only docs/operations/cicd-simplification.md to record PRs perf(ci): scope PR checks and require complete results #3810–chore(release): retire bypassing sync helper and prepare staged promotion #3813, saved required checks, completed post-merge checks, observed job selection, and the remaining production/Mac boundaries.
• Use this useful documentation update to observe the hosted documentation-only CI lane before the later branch-structure transition.

Verification

• Before the documentation update, the resolved complete tree was exactly 8b7d254b91cfc850587958541a27ea76bea64264, matching the tested develop head.
• Final candidate tree: f6853410d1b49deca57de9310281dfa470ba5bf3.
• Only one file differs from develop: the rollout runbook. Relative links and git diff --check pass.
• Existing passing code suites were not rerun locally because no runtime or test bytes changed. Hosted checks for this exact candidate will be verified before merge.
• Phase4 post-merge CI, Release, and CodeQL completed successfully.

Release boundary

This is protected source preparation. The subsequent develop → data PR will retain full promotion checks. No production deployment, main-only activation, provider mutation, or Mac service cleanup is performed here. The active Vercel connection still cannot identify the project/current rollback target.

Hosted documentation-only observation

Exact head: 592e713c6968e07c254aaf30a12787e173d02609.

• Web CI: success; the scope and final CI jobs ran, with eight skipped API job records.
• Security: success; scope and final Security ran, with seven skipped jobs.
• The skipped web unit matrix is one unexpanded API record, not eight executed jobs. These counts must not be described as fifteen actual skipped web jobs.
• Release, Promotion Path, and applicable CodeQL remain independent checks; the documentation scope does not replace them.

Observed web CI timings

All measurements use each named run's exact head and completed attempt 1.

Input	PR / run	Executed / skipped API records	Elapsed seconds	Initial wait seconds	Sum of executed job seconds
Backend	#3811 / 43254101716	11 / 6	310	4	1581
Full compatibility	#3813 / 43256948902	17 / 0	905	3	4568
Documentation	#3814 / 43266364981	2 / 8	32	4	26

Elapsed is run creation to the latest job completion. Initial wait is run creation to the earliest non-skipped job start; it is not total queue time. Summed job duration includes parallel setup/cleanup and result jobs. These different changes select different workloads, so this is an observation table, not a controlled speedup, billing, CPU, or whole-PR readiness measurement.