fix(devx): classify the tree-measurement claims under scripts/ instead of sweeping them

scripts/check-pnpm-filter-targets.mjs

@@ -127,10 +127,16 @@ const SCANNED_EXTENSIONS = [...JS_EXTENSIONS, ...HASH_COMMENT_EXTENSIONS, '.json
  * ── Why `scripts/**` is honest here, with the measurement ───────────────────
  *
  * This is the `subtree` case: the walk descends the whole of scripts/ and every
- * file carrying a scanned extension is judged. Measured on this tree, the
- * declaration names 235 tracked files under scripts/ and this gate reads 228 of
- * them — 97.0%. The 7 it skips are the non-code files the extension filter
- * drops, not a subtree it never opens.
+ * file carrying a scanned extension is judged. Measured at `52a41b72e`
+ * (2026-08-23): the declaration named 235 tracked files under scripts/ and this
+ * gate read 228 of them — 97.0%, the 7 it skipped being the non-code files the
+ * extension filter drops rather than a subtree it never opens.
+ *
+ * ⛔ That reading is deliberately pinned to its commit and NOT refreshed here.
+ * Both counts move with every file added under scripts/ and nothing reprints
+ * them, so a figure restored in the present tense would read as current and go
+ * false in silence. What this paragraph argues is the SHARE, and the share is
+ * what survives the churn; for today's counts, run the walk below.
  *
  * ── Why the workspace manifests stay UNDECLARED ─────────────────────────────
  *

scripts/check-published-files.mjs

@@ -388,11 +388,17 @@ function matcher(pattern) {
  * This is the `subtree` case, not the `filtered` one check-examples-live-imports
  * refuses. `walk()` below enumerates EVERY non-build file of every publishable
  * member and MINIMAL judges each of them against FORBIDDEN, so the declaration
- * names files this gate really opens. Measured on this tree: the declaration
- * names 5263 tracked files and the gate judges 4803 of them — 91.3%. The 460 it
- * does not judge are the members whose OWN manifests this gate read in order to
- * exclude them (`private`), which is itself a read of the declared subtree, so
- * a manifest card there is a true lead rather than a fabricated one.
+ * names files this gate really opens. Measured at `52a41b72e` (2026-08-23):
+ * the declaration named 5263 tracked files and the gate judged 4803 of them —
+ * 91.3%. The 460 it did not judge are the members whose OWN manifests this gate
+ * reads in order to exclude them (`private`), which is itself a read of the
+ * declared subtree, so a manifest card there is a true lead rather than a
+ * fabricated one.
+ *
+ * ⛔ Those counts are deliberately pinned to that commit and NOT refreshed. They
+ * move with every package added to or removed from the workspace, no gate
+ * reprints them, and the claim this paragraph rests on is the SHARE — that the
+ * declaration names what the gate really opens — never the level.
  *
  * The contrast that sets the boundary is in check-published-readme-exports.mjs,
  * which enumerates the same members and scores 2.8% — its refusal docblock

scripts/check-skill-compatibility-version.mjs

@@ -97,14 +97,22 @@ const PACKAGE_ROOTS = ['packages', 'apps', 'examples'];
  * The instrument can only express a SUBTREE: `hintCovers` collapses globs, so a
  * declared hint names every tracked file beneath it and there is no way to spell
  * "the package manifests under this root". That makes the two sides of this gate
- * completely different trades, and both were measured on this tree:
+ * completely different trades, and both were measured at `f29e89717`
+ * (2026-08-22):
  *
- *   skills/**       49 of 50 tracked files are skill directories this gate
+ *   skills/**       49 of 50 tracked files were skill directories this gate
  *                   reads — 98% precision over a 50-file subtree.
  *   packages/**     73 package.json files out of 4903 tracked files — 1.5%,
  *                   pasted into every packages/** dispatch prompt in the repo.
  *   apps/**         1 of 35 (2.9%) · examples/**  4 of 238 (1.7%).
  *
+ * ⛔ Those four readings are deliberately pinned to that commit and NOT
+ * refreshed in place. Every term moves with the tree, nothing reprints them,
+ * and what decides the trade is the ORDER of magnitude between the two sides —
+ * which is why a figure restored in the present tense would be a new decaying
+ * claim rather than a better one. `node scripts/pm/bare-root-worklist.mjs`
+ * carries this gate's rows for the three package roots, with their own dates.
+ *
  * The three package roots are therefore the +139084 fabrication one level up —
  * the very measurement in `hintCovers`' docblock, which prices accepting bare
  * top-level directory words at that many fabricated (gate, file) pairs because

scripts/check-slot-lookup-ratchet.mjs

@@ -147,9 +147,17 @@ const BASELINE_PATH = 'scripts/slot-lookup-baseline.json';
  * a lead it does not need. That trade is the one `CHANGE_KIND_GATES` already
  * decided for the three structurally identical whole-tree ratchets in
  * `dispatch-gates.mjs`, and it is decided the same way here — by what the gate
- * costs to run needlessly. Measured on this tree: 46s, no build required, and a
- * failure names the offending file and line. A seat that runs it needlessly
- * loses seconds; a seat that is never prompted loses a CI round.
+ * costs to run needlessly. Measured at 46s when this was recorded
+ * (`99ca6623f`, 2026-08-19), with no build required, and a failure names the
+ * offending file and line. A seat that runs it needlessly loses seconds; a seat
+ * that is never prompted loses a CI round.
+ *
+ * ⛔ The duration is deliberately left at that commit and NOT refreshed in
+ * place. A wall-clock reading is a reading of one BOX as much as of one tree —
+ * this repo's own agent containers read several times the figures recorded for
+ * them — so a fresh number written here would decay without the tree moving at
+ * all. Time it on the box you are deciding for; what the trade needs is the
+ * ORDER (seconds against a CI round), and that is what is stated.
  *
  * Two things a seat prompted by this declaration needs to know, and neither is
  * visible from a green `pnpm lint`:

scripts/docs-audit/README.md

@@ -552,9 +552,12 @@ How often it renders, re-derived over the 40 first-parent commits ending at `e43
 3 of the 17 package-touching runs (18%) — `20a452e664`, `f213793ddb`, `dd4113ec0b` — so it
 is a rare notice rather than a per-PR banner, which is what keeps it readable.
 
-**Cost** (the card's open question): the anchor derivation reads the same 178-page corpus
-the old one did, plus the 18 route-source/ledger files (~875 KB) and one `git show`
-per changed file per side. Measured end-to-end on the ten PRs above, `node affected-docs.mjs`
+**Cost** (the card's open question): the anchor derivation reads the same hand-written
+corpus the old one did, plus the 18 route-source/ledger files (~875 KB) and one `git show`
+per changed file per side. ⛔ The corpus SIZE is deliberately absent from that sentence —
+run `check-audit-scope.mjs` (`pnpm check:docs-audit-scope`) for today's page count. A size
+written down in the present tense decays with nothing going red, which is exactly what the
+`178` that used to stand here did. Measured end-to-end on the ten PRs above, `node affected-docs.mjs`
 went from 85-195 ms to 114-582 ms. The heaviest case is the widest diff; every case stays
 well under a second, against a job that already spends seconds checking out the repo and
 setting up Node. It is the right default for every PR.