finding(types): `zod-mirror-parity.test.ts`'s header says "16 pairs" where the ledger holds 18 — the one figure in that sentence no `headerFigures` pin covers, so it compiles and tests green while wrong

[os-zhuang]

Finding (observation, awaiting first grading). Filed unlabelled and unassigned — grading and domain:* belong to triage. Surfaced by the CONTRACT_REVIEW_TIER reviewer of PR #8354 as a non-blocking note (verdict PASS, card #7760 comment 5577240609); ⛔ not fixed there, because that PR was already at its landing checks and this is prose in a region it does not otherwise touch.

⚠️ Provenance is split, and stated as such rather than blurred: two halves below are this seat's own reading at 35f49ff1c; the count of 18 is the contract reviewer's measurement and this seat did not re-derive it. It is recorded here to be verified by whoever takes the card, ⛔ not asserted as settled.

The sentence

packages/types/src/__tests__/zod-mirror-parity.test.ts, header docblock, read at 35f49ff1c (PR #8354's head) — verbatim:

⛔ That repaired no mirror and no declaration: it made a face
READABLE that had been unknown. 19 keys across 16 pairs LEFT (13 entries
emptied) because the reading they recorded was the annotation, and 3 keys
ENTERED …

Measured by the PR #8354 contract reviewer: 18 distinct pairs, not 16 — 13 emptied plus 5 reduced, the five named as AppComponentSchema, DashboardComponentSchema, PageNodeSchema, ObjectViewSchema, DetailViewSchema. ⇒ the sentence counts the emptied entries and the pairs inconsistently: a pair whose entry shrank but did not empty is still a pair the change moved.

Why it stays green — this seat's own reading

The file pins its header figures with headerFigures(spelling: RegExp) (:3056), which asserts the spelling occurs exactly once and returns its capture groups as numbers. Every call site pins an entries/keys spelling:

:3084  /\*\*SPEC-DERIVED \((\d+) entr(?:y|ies), (\d+) keys?\)\*\*/
:3085  /\*\*LOCAL \((\d+) entr(?:y|ies), (\d+) keys?\)\*\*/
:3086  /\*\*(\d+) entries \/ (\d+) keys\*\* — (\d+) \/ (\d+) spec-derived, (\d+) \/ (\d+) local\./
:3089  /\*\*(\d+) entries\*\* in `UnmirroredDeclared`, \*\*(\d+) keys\*\* across them/

No headerFigures regex matches a "pairs" figure. ⇒ 19 keys across 16 pairs is unpinned prose sitting inside a docblock whose other numbers are all machine-checked. It compiles, and tsc -p tsconfig.test.json and vitest are both green with it wrong.

⭐ That is the exact rot this header warns about in its own words, three lines above the defect:

restate one without checking that the pin's spelling still reaches it.

The neighbouring figures (23 entries / 36 keys / 47 arms, and the 6 / 30 / 0 / 11 split) are covered and did move correctly with PR #8354. Only the pairs figure escaped.

Radius

• The same "16 pairs" figure is repeated in PR fix(types): give seven recursion-breaking zod mirrors their declaration as both type arguments #8354's body.
• It is not carried by .changeset/7760-lazy-mirror-input-type-args.md ⇒ ⛔ this does not reach CHANGELOG.md and is not published text. That is what keeps it non-blocking, and it is why the reviewer did not ask for rework.
• No consumer reads it; the cost is to the next person who trusts the header.

What this needs

1. Re-derive the pairs count (⛔ do not adopt 18 from this card — that is the mistake this card is about) and correct the sentence, in the test file and in PR fix(types): give seven recursion-breaking zod mirrors their declaration as both type arguments #8354's body if it has not merged.
2. Decide whether the figure should be pinned at all. Two honest options, and the choice is the card's real content:
   • add a headerFigures spelling for it, so it cannot rot again — but every pinned figure is a maintenance cost on every future ledger move; or
   • drop the pairs figure from the sentence, keeping only the pinned entries/keys/arms. A number no instrument checks is a liability in a header whose whole authority rests on the others being checked.
3. If it is pinned, state whether "pairs" means entries that emptied or entries the change moved at all — the ambiguity is what produced the discrepancy.

Not a duplicate of

• finding: #6058's SPEC-DERIVED remedy split routes ObjectViewSchema's unmirrored keys on a scanner false positive — it references no spec schema #7279 — that card corrected the SPEC-DERIVED / LOCAL split in the same header, and its figures are pinned (:3084–:3086). Same file, same docblock, different figure; the whole point here is that this one is outside the pinned set.
• types: re-judge the 27 SCHEMA-NODE-bucketed keys of zod-mirror-parity per UNION ARM — a per-key verdict over per-arm facts let DashboardComponentSchema.widgets' concrete gap hide behind a schema-node sibling (census ordered by the #7952 ruling) #8252 — built the arm split (6 / 30 / 0 / 11), which is pinned and correct.
• finding(types): the recursion-breaking annotation on 10 zod mirrors erases their published input face to unknown — the single-or-list producer objectui#7069 named systematic is unauditable at the type level because of it #7760 / PR fix(types): give seven recursion-breaking zod mirrors their declaration as both type arguments #8354 — the change that moved every figure in the sentence. Its own contract review passed; this is a residue it recorded rather than a defect in the ruled change.

Refs: #7760 / PR #8354 (where it was found; reviewer's note in comment 5577240609) · #7279 · #8252 · #7069 (the ledger's origin).