docs(skills): objectstack-i18n names the action-entry keys os validate judges (#21656)

Fixes #21583
Clause-②: no

Tier H (`skills/**`): left as a draft; lands only by the maintainer's
hand or an authorized APPROVED review, then the owning seat lands it.
Not readied, no auto-merge armed.

## What changed

One file, `skills/objectstack-i18n/SKILL.md`, three places, mirroring
`content/docs/protocol/kernel/i18n-standard.mdx` :1039–:1058 (the
sentence at :1039 and the table rows at :1053, :1055–:1058). No second
vocabulary: keys and conditions are spelled the way the page and the
lint spell them; the skill names shapes and outcomes, never the gate's
internals.

### 1. The `_actions.{action_name}` row of the bundle-shape table (was
:164)

Before:

| `_actions.{action_name}` | `label`, `description`, `confirmText`,
`successMessage`, `params.{param_name}`, `resultDialog` |

After:

| `_actions.{action_name}` | `label`, `description`, `confirmText`,
`successMessage`, `params.{param_name}` (and `.options.{value}` beneath
it — an option's stored `value` in that param's inline `options`),
`outcomeMessages.{outcome}` (an outcome the action's own
`outcomeMessages` declares), `resultDialog.title` / `.description` /
`.acknowledge` (only under an action that declares a `resultDialog`),
`resultDialog.fields.{path}` (the literal `path` of a field in
`resultDialog.fields[]`, dots included — `client.secret` is one key);
the same keys hold under `globalActions.{action_name}` |

Reference rows: :1055 (param options), :1056 (`outcomeMessages`), :1057
(`resultDialog.fields.{path}`), :1058 (dialog leaves), :1039 (a
`{action}.` row holds at both action addresses).

### 2. The `globalActions` line (was :170)

Before: `globalActions` (object-less actions)

After: `globalActions` (actions with no `objectName` — a key naming a
bound action is an **error** whose message names the
`objects.{owner}._actions.{action_name}` key to write instead)

Reference row: :1053.

### 3. The `translation-target-unknown` paragraph (was :198–:203)

Before, after the enumeration sentence:

that do. A bundle keyed to something since renamed still parses — the
label just renders
    silently in its source locale while every neighbouring one resolves.

After:

that do. Three more shapes under an action entry are the error: an
outcome the action does
not declare in `outcomeMessages`; a `resultDialog.fields` path it does
not declare; and
`resultDialog` `title` / `description` / `acknowledge` under an action
that declares no
`resultDialog` — a bundle translates a dialog the action declares and
cannot add one. A
param's inline option keys share the option warning with a field's
`options`; a param that
references a `field` and declares no inline `options` is not judged. A
bundle keyed to
something since renamed still parses — the label just renders silently
in its source locale
    while every neighbouring one resolves.

Reference: rows :1055–:1058 and the paragraph under the table (the
error, and the option warning as the deliberate exception).

### Gate-side reading (read only, not changed)

`packages/lint/src/validate-translation-references.ts` judges every rule
the page states: `checkActionEntry` is called from both the `_actions`
loop and the `globalActions` loop; `checkActionParamOptionKeys` is the
warning rule and returns without judging a field-backed param that
declares no inline `options`; `checkActionOutcomeMessages`,
`checkActionResultDialogLeaves` (which returns when the action declares
a `resultDialog`) and `checkActionResultDialogFields` are the error
rule; a `globalActions` key naming a bound action is an orphan whose
hint reads "Move these keys under `objects.{owner}._actions.{action}`".
Page and gate agree, so the page was not touched.

Placeholder spelling: the skill's own `{action_name}` / `{param_name}`
are kept in the table (the row already used them); `{value}`,
`{outcome}`, `{path}` and `{owner}` are the page's.

## Readings

- `skills/objectstack-i18n/SKILL.md`: 496 → 503 lines (net +7, budget at
most +8).
- All `skills/**/SKILL.md` together: 4397 → 4404 lines (base
`origin/main` 759dbe9).
- Token ratchet (`node scripts/check-skills-token-ratchet.mjs`): 4752 →
5032 tokens against the unchanged ceiling 6338 (headroom 1586 → 1306).

## Verification (tree at afecd20)

Derived with `node scripts/pm/dispatch-gates.mjs --commands --repo
objectstack-ai/objectstack` (24 commands, the same 24 the dispatch
named) and reconciled with `--ran`. Every exit code was captured before
any pipe, from a redirected log.

- Direct node gates, each exit 0: `check-ci-filter-parity`;
`check-closing-keyword-parity` and its `--self-test` (40 assertions);
`check-comment-mask-corpus` (8122 files, 0 disagree);
`check-doc-route-spelling --advisory` ("population clean") and its
`--self-test`; `check-skills-token-ratchet` ("54 authored bundle file(s)
within their ceilings") and its `--self-test` (65 cases).
- `pnpm check:*`, each exit 0: `agent-test-spelling` (0 violations),
`corpus-claim-drift` (251 files, 53 under skills),
`cross-package-test-inputs`, `doc-authoring` ("53 published skill files
clean — no internal issue-id references"), `driver-memory-census`,
`gitlink-declared`, `nul-bytes` ("10027 text file(s) … no raw ASCII
control bytes"), `pm-governed-merges`, `refd-timer-probe`, `role-word`,
`skill-compatibility` ("10 SKILL.md file(s) reconciled"),
`skill-frame-sync`, `skill-identifier-liveness` ("Leg 1: 457 citation(s)
over 53 published file(s) … Leg 2: 8 registered exhaustive section(s), 0
ledgered gap(s)"), `watch-hint-literal`.
- `pnpm --filter @objectstack/spec run check:skill-docs` exit 0 ("Skill
docs in sync"); `check:skill-refs` exit 0 ("9 generated files in sync
with packages/spec") — the second added because a SKILL.md body edit
names it in AGENTS.md. Both read SKILL.md frontmatter and
`packages/spec/src` through tsx, so no build was needed for them.
- `pnpm --filter @objectstack/lint run check:doc-formula-expressions`
exit 0 ("22 record-scoped formula example(s) across 460 files / 1381 TS
blocks judged clean"), after `turbo run build
--filter=@objectstack/lint` under `scripts/pm/os-verify-lock.sh` (4
tasks, all cache hits; no tracked file changed by the build).
- Not measured locally, declared to CI: the whole-tree families, the
type-check lanes and the Test Core shards — the diff touches no
TypeScript and no package.

Byte hygiene: no raw control bytes in the file, no tracker-shaped ids,
no new code fence, no model identifier.

## 维护者速读(草稿)

**改了什么。** 只改 `skills/objectstack-i18n/SKILL.md` 一个文件、三处:bundle 形状表的
`_actions.{action_name}` 行补齐 `os validate`
实际判定的动作子键(参数选项、`outcomeMessages`、`resultDialog` 的三个文案叶子与
`resultDialog.fields.{path}`),并注明同样的键在 `globalActions`
下同样成立;`globalActions` 一句写明:键名指向已绑定对象的动作是错误,错误消息会指出应改写到
`objects.{owner}._actions.{action_name}`;`translation-target-unknown`
段落列出三种新的错误形状,并说明参数选项键与字段选项键共用同一条警告。措辞与 `i18n-standard.mdx` :1039–:1058
一致,不另造词汇,不复述门禁内部实现。

**为什么改。** 这是您在会话中点名、但当时没有建卡的技能文档修改。技能是 AI 作者写翻译包时读的文本;原文落后于门禁,作者无从得知
`outcomeMessages`、`resultDialog.fields`
可译,也不会被告知在没有对话框的动作下写对话框文案、或把已绑定对象的动作写到 `globalActions` 下会被 `os validate`
以错误拒绝。参考页已经写对,技能只是对齐。

**风险与代价(含回滚)。** 纯技能文本,不发布任何 npm 包,无 changeset(打 `skip-changeset`)。净增 7
行(预算不超过 8),token 棘轮 4752 → 5032 / 6338,上限未动。派生出的 24 项门禁本地全绿,另加
`check:skill-refs` 也绿。回滚即 revert 这一个提交。

**席位意见。** (留空,席位定稿)

**你要做的。** 这是 Tier H(`skills/**`):请审阅后给出 APPROVED 评审;之后由席位落地。PR 保持
draft,未开 auto-merge。

## Acceptance notes

- The `translation-target-unknown` paragraph now carries the conditions
once and the table row names the keys; the page says "takes the field's
list when the dialog renders" as the reason a field-backed param is not
judged — the skill states the consequence only, to stay inside the line
budget.
- `references/_index.md` is generator-owned and unchanged: the body edit
touches no frontmatter, which is all `gen:skill-refs` reads from this
file.
- `content/docs/**` is the source copied from, not an edit target;
nothing else in the tree was touched.

---
_Generated by [Claude
Code](https://claude.ai/code/session_01CB6W87z22K2yjUCDyVrJRk)_

Co-authored-by: objectstack-fleet[bot] <objectstack-fleet[bot]@users.noreply.github.com>
Co-authored-by: Claude <noreply@anthropic.com>

Author: objectstack-fleet[bot]

skills/objectstack-i18n/SKILL.md

@@ -161,16 +161,17 @@ All translatable content for a single object is aggregated under
 | `label` / `pluralLabel` / `description` | Object-level text (every key optional) |
 | `fields.{field_name}` | `label`, `help`, `placeholder`, `options` (option value → label) per field |
 | `_views.{view_name}` | `label`, `description`, `emptyState.title` / `emptyState.message`, `bulkActions.{def_name}` (`label`, `confirmText`, `confirmLabel`, `params`) |
-| `_actions.{action_name}` | `label`, `description`, `confirmText`, `successMessage`, `params.{param_name}`, `resultDialog` |
+| `_actions.{action_name}` | `label`, `description`, `confirmText`, `successMessage`, `params.{param_name}` (and `.options.{value}` beneath it — an option's stored `value` in that param's inline `options`), `outcomeMessages.{outcome}` (an outcome the action's own `outcomeMessages` declares), `resultDialog.title` / `.description` / `.acknowledge` (only under an action that declares a `resultDialog`), `resultDialog.fields.{path}` (the literal `path` of a field in `resultDialog.fields[]`, dots included — `client.secret` is one key); the same keys hold under `globalActions.{action_name}` |
 | `_sections.{section_name}` | Form section `label`, `description` |
 | `_tabs.{tab_name}` | Filter-preset tab `label` (`ViewTabSchema.name`) |
 | `_validations.{rule_name}` | `message` only — the sentence a rejected write returns (replaces the retired `validationMessages`) |
 
 Top-level groups alongside `objects`: `apps` (label, description, navigation),
-`messages`, `globalActions` (object-less actions), `dashboards`, `datasets`, `pages`,
-`flows`, `metadataForms`, `settingsCommon`. `settings` is **platform-only** — it is
-keyed by a settings manifest's namespace and only the platform declares one, so an
-app bundle carrying it is refused by name.
+`messages`, `globalActions` (actions with no `objectName` — a key naming a bound action is
+an **error** whose message names the `objects.{owner}._actions.{action_name}` key to write
+instead), `dashboards`, `datasets`, `pages`, `flows`, `metadataForms`, `settingsCommon`.
+`settings` is **platform-only** — it is keyed by a settings manifest's namespace and only
+the platform declares one, so an app bundle carrying it is refused by name.
 
 For the exact Zod shape (and any field that may have been added since), read
 `node_modules/@objectstack/spec/src/system/translation.zod.ts` —
@@ -199,8 +200,14 @@ parse, ship, and resolve to nothing.
 is an **error** and fails the run; `translation-option-key-unknown` is a warning.
 A key naming an object, field, view, tab, action, param, section, validation rule, app,
 nav item, dashboard, widget or flow screen that does not exist is listed alongside the names
-that do. A bundle keyed to something since renamed still parses — the label just renders
-silently in its source locale while every neighbouring one resolves.
+that do. Three more shapes under an action entry are the error: an outcome the action does
+not declare in `outcomeMessages`; a `resultDialog.fields` path it does not declare; and
+`resultDialog` `title` / `description` / `acknowledge` under an action that declares no
+`resultDialog` — a bundle translates a dialog the action declares and cannot add one. A
+param's inline option keys share the option warning with a field's `options`; a param that
+references a `field` and declares no inline `options` is not judged. A bundle keyed to
+something since renamed still parses — the label just renders silently in its source locale
+while every neighbouring one resolves.
 
 ---