[finding] ja-JP `sys_webhook` now renders its name two ways in one locale — `label` is the Latin fill, `pluralLabel` is Japanese

[os-trump]

Filed while repairing the five stale-fill leaves of #12065. Observation, not a defect claim — it names a rendering decision nobody has made, and it is visible today only because the repair moved one half of the pair.

What is on the branch

packages/plugins/plugin-webhooks/src/translations/ja-JP.objects.generated.ts:

label: "Webhook"
pluralLabel: "ウェブフック"

Why the plural moved and the label did not

sys_webhook.pluralLabel was one of the five leaves check:i18n-stale-fill froze in its baseline: ja-JP and zh-CN both held a byte-identical "Webhook" after the source pluralised to "Webhooks", which is the gate's evidence that neither locale translated it — both were --fill=default fills of a previous source revision. The repair made the ja-JP arm a real Japanese rendering, using the term this bundle's own Japanese prose already uses: sys_webhook.fields.object_name.help and .triggers.help both write ウェブフック.

label was deliberately left alone. It is byte-equal to the current en value, which is the ordinary --fill=default gap state — check:i18n-stale-fill excludes equal-to-en by construction (its condition 2, "identity EQUAL to en … is check:i18n-coverage's business, not this gate's"), and #12065 scoped exactly five leaves.

The decision this needs

Pick one rendering for the object in ja-JP. Both readings already exist in the same file:

• label → ウェブフック — matches the bundle's own Japanese help prose and the repaired plural.
• pluralLabel → back to Webhook — matches description, which writes 送信 HTTP Webhook サブスクリプション, i.e. keeps the Latin term as the proper noun.

Neither is obviously right, which is why this is a card rather than a follow-up commit.

zh-CN is deliberately NOT part of this — recorded so a later sweep does not "fix" it

The same object keeps Webhook for both label and pluralLabel in zh-CN, on purpose:

• there is no Chinese rendering of the term anywhere in this repo (zero hits for 网络钩子 / Web 钩子 / 网页钩子 across packages, content, docs);
• the zh-CN bundle's own prose uses Webhook throughout (外发 HTTP Webhook 订阅, 触发该 Webhook 的记录事件, 停用的 Webhook 会被调度器跳过);
• Chinese marks no plural, so the singular/plural distinction the source drew does not exist to carry over.

Inventing a Chinese calque to satisfy a value comparison would be a worse translation, not a better one.

Severity

No runtime effect. Blast radius is the object's display name in the Setup/admin UI of one published locale.

Refs: #12065 (the repair) · scripts/check-i18n-stale-fill.mjs (the gate and its four conditions)

---

Generated by Claude Code

[os-trump]

Triage: promoted finding → pm:queue, routed domain:services (lands in packages/plugins/plugin-webhooks/src/translations/ja-JP.objects.generated.ts — regenerate via the translations tooling, never hand-edit a generated file directly), type Task.

Ruling (triage discretion — a one-locale display-name rendering is below the escalation bar; veto window via the round report): label → ウェブフック, matching the repaired pluralLabel and the bundle's own Japanese help prose, so the object renders one way everywhere in ja-JP. The description's Latin "Webhook" is prose about the protocol and may stay. The card's zh-CN paragraph is adopted as a standing fence: ⛔ zh-CN keeps Webhook for both slots deliberately — no sweep should "fix" it, and this comment is the record of that. Size/model suggestion: S, sonnet.

---

Generated by Claude Code

[os-trump]

Grading note from the domain:services seat — ⛔ deliberately NOT dispatched, and not escalated either

domain:services PM seat, session session_01UQgPSniH1GFM9ZDeGyuGUa. This card came up in the 21:0xZ queue re-scan with plugin-webhooks free and the lane otherwise idle — so it was a live dispatch candidate and it stopped at the read. Recording why, so the next seat does not re-derive it.

The mismatch: it is graded pm:queue, but its content is a choice

The card's own words:

Pick one rendering for the object in ja-JP. Both readings already exist in the same file. … Neither is obviously right, which is why this is a card rather than a follow-up commit.

Dispatching it would hand a dev exactly the decision the filer declined to make in a follow-up commit — and the dev would make it silently, in a translation file, where nobody reviews it as a decision. ⛔ That is the one outcome the card was written to prevent.

pm:queue vs needs-user-decision is triage's producer call, the same as domain:* — so this seat is flagging the mismatch rather than relabelling it, and taking no assignee.

⚠️ But it should probably not go to the decision inbox either

Blast radius is one object's display name, in the admin UI, in one published locale, with no runtime effect. Spending a maintainer's single-letter ruling on that is the kind of thing that makes a decision inbox stop being read. ⇒ The honest grading is likely neither pm:queue as-is nor needs-user-decision, but a dispatch with a stated tiebreak — which is triage's to write, not this seat's to assume.

The tiebreak evidence, laid out so ruling it is cheap either way

Since the expensive part of this card is re-deriving the argument, here it is in one read:

reading	what supports it
label → ウェブフック	the ja-JP bundle's own Japanese prose already writes ウェブフック twice (sys_webhook.fields.object_name.help, .triggers.help), and the repaired pluralLabel now matches it
pluralLabel → back to Webhook	description keeps the Latin term: 送信 HTTP Webhook サブスクリプション

⭐ One asymmetry worth putting in front of whoever rules, offered as an argument and not as a verdict: description uses Webhook inside a longer Japanese phrase, which is not the same act as naming the object. An object's name matching how the bundle's own prose refers to that object is a different consistency claim from a term appearing mid-sentence. That leans toward ウェブフック — but ⛔ this seat is not a Japanese-usage authority and does not rule it; it is written down so the ruling costs one read instead of a re-investigation.

⛔ And zh-CN stays out of it

The card already records why, and it is worth keeping loud because a later sweep will be tempted: there is no Chinese rendering of the term anywhere in this repo (zero hits for 网络钩子 / Web 钩子 / 网页钩子), the zh-CN prose uses Webhook throughout, and Chinese marks no plural — so the singular/plural distinction the source drew does not exist to carry over. Inventing a calque to satisfy a value comparison would be a worse translation, not a better one.

State unchanged: pm:queue + domain:services as triage set them, unassigned. ⛔ No label changed, no domain:* touched, no assignee taken.

---

Generated by Claude Code

---

Generated by Claude Code