ci(fleet-patrols): prose and comments stop naming the four retired patrol workflow files

.github/workflows/fleet-patrols.yml

@@ -22,7 +22,7 @@ name: Fleet Patrols
 # ⛔ A cron string under `schedule:` and the same string in its job's `if:` are
 # ONE fact spelled twice: change both in the same edit. A cron no `if:` names
 # fires a run that does nothing, and a job whose `if:` names a cron the
-# schedule does not declare never runs — neither shape is red anywhere.
+# schedule does not declare never runs — clause 7 of the watchdog pin refuses both.
 #
 # `workflow_dispatch` takes one input, `patrol`: `all` (the default) runs every
 # job, a job id runs that one alone — the manual fire and the smoke test each
@@ -757,7 +757,7 @@ jobs:
   #
   # ## Why a workflow, when the gate ALREADY runs on every PR
   #
-  # Unlike `half-state-patrol.yml` and `required-set-patrol.yml`, this check does
+  # Unlike `half-state-patrol.yml` and the `required-set-patrol` job, this check does
   # not need a runner for TRANSPORT reasons. It is fully offline — it reads
   # `packages/spec/CHANGELOG.md` and `content/docs/releases/**` straight out of the
   # tree, opens no socket, and runs inside `lint.yml` on every pull request today.

.github/workflows/lint.yml

@@ -147,7 +147,7 @@ concurrency:
 #     self-test half of a gate whose scan runs elsewhere). An alias is a copy
 #     only when it is the same command, and a copy is what this rule removes;
 #   * its job sets up no pnpm (an install-free job elsewhere, such as
-#     check-links.yml or board-snapshot.yml, keeps the direct form);
+#     check-links.yml or fleet-patrols.yml's board-snapshot job, keeps it);
 #   * it predates this rule and its script has no alias yet. Those steps are
 #     not swept (#25919 「存量不横扫」); each gains an alias when its step is
 #     next rewired. Their comments say "invoked as `node` rather than through a
@@ -2539,21 +2539,21 @@ jobs:
       # per-PR path means an unrelated PR is never blocked by checklist drift.
       # Run it by hand: `pnpm check:platform-checklist`. See that dir's README.
       # Its red does have a reporting channel now, and it is NOT this file:
-      # `.github/workflows/platform-checklist-watchdog.yml` runs the gate on `main`
-      # daily and files or refreshes one issue when it is red.
+      # the `platform-checklist-watchdog` job of `.github/workflows/fleet-patrols.yml`
+      # runs the gate on `main` daily and files or refreshes one issue when it is red.
 
       # ⛔ …and the step below is NOT that wiring. What runs here is the STATIC PIN
       # over the watchdog workflow: it asserts the file exists, fires on
       # `schedule:` + `workflow_dispatch:`, carries NO `merge_group:` /
       # `pull_request_target:` trigger, that any `pull_request:` trigger is
-      # `paths`-filtered to the watchdog file ALONE, that no board write is
+      # `paths`-filtered to a frozen set of the four patrols' inputs, that no board write is
       # reachable from a pull_request run, and that the gate is invoked through its
       # package script rather than an inlined copy. Two file reads, no socket.
       #
-      # The watchdog's own `pull_request:` trigger names only itself, so it fires on
-      # no unrelated PR — that filter is what keeps the decision above intact, and
-      # pinning it is most of this step's job. It has to live HERE rather than
-      # inside the watchdog for the other half: a PR DELETING the watchdog would
+      # That trigger names only the patrols' inputs and no job there is required or
+      # reds on drift, so no unrelated PR is blocked — that filter keeps the decision
+      # above intact, and pinning it is most of this step's job. It has to live HERE
+      # rather than inside the watchdog for the other half: a PR DELETING it would
       # silence any check that lived inside it, and this job is the only place where
       # the absence of that file is observable before it merges.
       #

.github/workflows/merged-branch-reaper.yml

@@ -234,8 +234,8 @@ on:
         type: boolean
         required: false
         default: true
-  # Exercise the sweep on changes to itself, the same posture as
-  # required-set-patrol.yml. This is what makes the FIRST delivery of this PR a
+  # Exercise the sweep on changes to itself, the same posture as fleet-patrols.yml's
+  # `required-set-patrol` job. This is what makes the FIRST delivery of this PR a
   # real would-delete list produced by a real runner rather than a claim about
   # one. Publishes no required context, so a red here blocks nothing.
   pull_request:

.github/workflows/shard-timings-refresh.yml

@@ -214,7 +214,7 @@ on:
     - cron: '30 5 * * 1'
   workflow_dispatch:
   # Changes to this lane get exercised before they merge — the same posture
-  # half-state-patrol.yml and required-set-patrol.yml keep, and the reason a
+  # half-state-patrol.yml and fleet-patrols.yml's required-set-patrol keep, and the reason a
   # scheduled-only lane is not an option: `dispatch-gates.mjs` reds when a gate
   # family is reachable ONLY on a schedule, because a family no PR-time event
   # reaches appears on no card's gate list and is therefore graded by nobody

.github/workflows/test-nightly-tiers.yml

@@ -49,7 +49,7 @@ name: Nightly Tiers
 # and installs Chromium for it (≈ 5 min, the "Prepare the showcase browser
 # smoke" step), which no other shard pays.
 #
-# ## What lands where — the house idiom of `platform-checklist-watchdog.yml`
+# ## What lands where — the house idiom of `fleet-patrols.yml`'s `platform-checklist-watchdog` job
 #
 # On green this workflow files nothing, edits nothing and closes nothing. On
 # red the `report` job — ONE writer, after both shards, never the shards

scripts/check-platform-checklist.mjs

@@ -85,8 +85,8 @@
 //      NOT wired into per-PR CI (the maintainer decision recorded in
 //      `.github/workflows/lint.yml` — it runs by hand before a release or after
 //      a large platform surface lands, with
-//      `.github/workflows/platform-checklist-watchdog.yml` sweeping `main`
-//      daily as its only standing caller), so a future out-of-vocabulary anchor
+//      the `platform-checklist-watchdog` job of `.github/workflows/fleet-patrols.yml`
+//      sweeping `main` daily as its only standing caller), so a future out-of-vocabulary anchor
 //      sits unreported until somebody runs this gate. The exit is the ruling's
 //      own: widen the shared vocabulary in `scripts/symbol-anchors.mjs`, and
 //      ⛔ never re-fork a private extension set here.
@@ -485,8 +485,8 @@ const err = (file, id, msg) => errors.push(`${file}${id ? ` · ${id}` : ''}: ${m
 // on a PR would ever reach a `--self-test` leg.
 //
 // ⚠️ CORRECTED (#11730): the second half of that sentence no longer holds, and
-// the first half is unaffected. `.github/workflows/platform-checklist-watchdog.yml`
-// now runs this gate on `main` daily through its package script, so a
+// the first half is unaffected. The `platform-checklist-watchdog` job of
+// `.github/workflows/fleet-patrols.yml` now runs this gate on `main` daily through its package script, so a
 // `--self-test` leg DOES execute somewhere — and the root alias
 // `check:platform-checklist` carries one, which is what `check:self-test-wired`
 // requires of every script CI runs. The gate is still NOT wired into per-PR CI;

scripts/check-stall-guard-budget.mjs

@@ -456,9 +456,9 @@ export function scan(root, defaults, parseYaml) {
  *
  * NOT the job id alone. A job id is unique only within its own workflow file,
  * and this tree really does reuse three of them across files (`publish` in
- * docker-publish.yml and release.yml, `patrol` in half-state-patrol.yml and
- * release-coverage-patrol.yml, `registry-canary` in publish-smoke.yml and
- * scaffold-e2e.yml). Keyed on the id alone, two single-guard jobs in different
+ * checklist-status.yml, docker-publish.yml and release.yml, `pack-smoke` in
+ * pack-smoke-optin.yml and publish-smoke.yml, `registry-canary` in
+ * publish-smoke.yml and scaffold-e2e.yml). Keyed on the id alone, two single-guard jobs in different
  * files would be printed as siblings -- inventing a shared clock that does not
  * exist, which is the more dangerous direction to be wrong in.
  *

scripts/ci/board-snapshot-r2-upload.sh

@@ -1,8 +1,8 @@
 #!/usr/bin/env bash
 # Copyright (c) 2026 ObjectStack. Licensed under the Apache-2.0 license.
 #
-# board-snapshot-r2-upload -- the body of board-snapshot.yml's "Upload the
-# archive to R2" step (#26086), moved out of an inline `run:` block.
+# board-snapshot-r2-upload -- the body of fleet-patrols.yml's `board-snapshot` job's
+# "Upload the archive to R2" step (#26086), moved out of an inline `run:` block.
 #
 #   bash scripts/ci/board-snapshot-r2-upload.sh          # in the step, env from its `env:`
 #   bash scripts/ci/board-snapshot-r2-upload.selftest.sh # offline: stub aws, throwaway archive

scripts/import-prerequisite.mjs

@@ -232,8 +232,8 @@ export const INSTALL_FIX = 'pnpm install';
  *   • The half that HOLDS: nothing mechanical changes. Every consumer of the
  *     gates that import this frame treats any non-zero as failure — measured,
  *     not assumed: the `&&` chains in the root `package.json`, the bare `run:`
- *     steps in `lint.yml`/`ci.yml`, and `required-set-patrol.yml`, which
- *     branches on `== '0'` / `!= '0'` and nothing finer. ⛔ This change buys
+ *     steps in `lint.yml`/`ci.yml`, and fleet-patrols.yml's `required-set-patrol`
+ *     job, which branches on `== '0'` / `!= '0'` and nothing finer. ⛔ This change buys
  *     zero CI benefit today and must not be sold as if it did.
  *   • The half that does NOT: it is not a new contract. The contract was
  *     already written, by the five sites above. What this frame was doing was

scripts/pm/dispatch-gates.mjs

@@ -326,8 +326,8 @@
  * `pull_request:` trigger with a `paths:` filter naming its own script, so
  * "changes to the patrol itself get exercised before they merge"
  * (`half-state-patrol.yml`'s own comment, and the same words in
- * `required-set-patrol.yml`, `release-coverage-patrol.yml` and
- * `merged-branch-reaper.yml`). The card's own specimen is one of those ten:
+ * `fleet-patrols.yml`'s `required-set-patrol` and `release-coverage-patrol`
+ * jobs, and `merged-branch-reaper.yml`). The card's own specimen is one of those ten:
  * `half-state-patrol.yml` already carried that trigger on the day the card was
  * filed. Three of the ten are `validate-deps.yml`'s, including
  * `check:override-consistency` — a gate every dependency card must run. So
@@ -1837,9 +1837,9 @@ const WORKFLOW_VALUE_SOURCE = /\$\{\{[\s\S]*?\}\}|\$\{[A-Za-z_]\w*\}|\$[A-Za-z_]
  *       MET — this gate grades a saved `turbo run test` log, and no log was
  *       named.` A dev following the list runs it and gets a non-verdict.
  *   `node scripts/check-required-contexts.mjs`  →  exit 0 — but it runs the
- *       STATIC pin check, while `required-set-patrol.yml` runs
- *       `--verify-required-set`, a live sweep of the repo's required set. Same
- *       script, different question, and the row said `[required-set-patrol.yml]`.
+ *       STATIC pin check, while fleet-patrols.yml's `required-set-patrol` job
+ *       runs `--verify-required-set`, a live sweep of the repo's required set.
+ *       Same script, different question, and the row named that patrol.
  *   `node scripts/check-shard-attestation.mjs` and
  *       `node scripts/check-cross-package-test-inputs.mjs`  →  exit 0, and both
  *       duplicate what `lint.yml` already runs under a `pnpm check:` alias

scripts/pm/dispatch-gates.self-test.mjs

@@ -672,8 +672,8 @@ export function selfTest({ tier = 'full' } = {}) {
   // invocation, so no case can pin a shape the tree does not have. Their
   // sources: `pr-automation.yml` (the `--base` gates), `ci.yml` (the shard
   // attestation), `engine-split-metric.yml` (`--days 90`),
-  // `required-set-patrol.yml` (a complete argv behind a continuation and a
-  // redirection), `prerelease-pin-watch.yml` (`--verbose 2>&1`; that workflow
+  // `fleet-patrols.yml`'s `required-set-patrol` job (a complete argv behind a
+  // continuation and a redirection), `prerelease-pin-watch.yml` (`--verbose 2>&1`; that workflow
   // left with the objectui pin, objectstack#25711, and its row stays as a
   // fixture of the shape). The live half at the end of the block re-reads the
   // others from the workflows themselves.
@@ -1046,8 +1046,8 @@ export function selfTest({ tier = 'full' } = {}) {
     );
     // A script invoked BOTH ways gets BOTH entries — the card's third clause,
     // read off the live tree. `check-release-section-coverage.mjs` is the
-    // specimen: `lint.yml` runs it bare, `release-coverage-patrol.yml` runs it
-    // bare AND `--strict`, and before the continuation join the `--strict` run
+    // specimen: `lint.yml` runs it bare, fleet-patrols.yml's
+    // `release-coverage-patrol` job runs it bare AND `--strict`, and before the continuation join the `--strict` run
     // had no entry of its own at all.
     const coverageKeys = new Set(direct.filter((i) => i.script === 'scripts/check-release-section-coverage.mjs').map((i) => i.check));
     t(

scripts/sync-release-index-currency.mjs

@@ -47,8 +47,8 @@
 // stays on 17.2.0 until someone notices. Worse here than for the sibling surfaces:
 // `.github/workflows/lint.yml` runs check-release-section-coverage WITHOUT `--strict`,
 // where a finding is advisory and the job is GREEN by design, so the staleness reaches
-// `main` green on every ordinary PR as well. Only `release-coverage-patrol.yml` passes
-// `--strict`, and a standing patrol is by construction an AFTER-THE-FACT reader.
+// `main` green on every ordinary PR as well. Only fleet-patrols.yml's `release-coverage-patrol`
+// job passes `--strict`, and a standing patrol is by construction an AFTER-THE-FACT reader.
 //
 // This repo has already chosen this exact shape for this exact drift class, three
 // times: sync-protocol-version.mjs (#2769), sync-template-versions.mjs (#2907),