Repository navigation
A flow notice cannot name a select field's option by its label in the recipient's language: the mail renderer has no label formatter and the flow passes no option labels to CEL label() #30172
Description
Source
Measured by HotCRM on platform 18.3.0 while fixing hotcrm#2391 (the stalled-deal notice printed the stage as needs_analysis). The implementer's report is hotcrm#2391 comment 6640505913; the HotCRM side is hotcrm#2406.
What happens
A notify node's templateData is evaluated once per run, before the recipients fan out, and each recipient is then sent the template of their own locale. A select field's value reaches the template as its stored value. None of the platform's routes turns it into the option label in the recipient's language:
| # | Route tried on 18.3.0 | Result |
|---|---|---|
| 1 | A template formatter, {{stage | label}} |
Refused as an unknown formatter (known-formatter). The refusal text says to format the value in the caller and pass finished text. |
| 2 | CEL label(currentOpp.stage) in templateData |
Refused at parse: label() takes record.FIELD only. |
| 3 | CEL label(record.stage) in a flow |
Refused: "this evaluation context carries no option labels". |
| 4 | A formula field label(record.stage) read by the flow |
Resolves, but in the reader's locale. A schedule run reads as the system, so a zh-CN recipient got "Needs Analysis". |
Seam: NotifyConfigSchema.templateData → the service-automation notify executor → the plugin-email renderer.
What HotCRM does meanwhile
hotcrm#2406 builds, at flow-build time, a CEL map of every shipped locale's option labels from the package's translation bundle (stage_label: { en, zh_CN, es_ES, ja_JP }), and each locale's template reads its own key ({{stage_label.zh_CN}}). It works, but every app has to write it, and it covers only the locales the app ships.
Suggestion
One of these, at the platform's choice:
- a renderer formatter that labels an option in the template's locale, given the object and field (for example
{{stage | option:'crm_opportunity.stage'}}); or - the notify executor resolves option labels per recipient when
templateDatanames a field by reference rather than by value.
Either lets an app pass the stored value and get the label each recipient reads.
Filing gate ① a reproduced gap, class b. Filed by the repo:hotcrm seat (huangyiirene, session session_01J8VVAnPkgK6CWrJa6t7Fk5).
Grade: priority:p3. Apps have a workaround; the cost is a hand-built label map per notice. Size: M.
Dedupe: searched open objectstack issues for notify option label locale, templateData label recipient, label formatter email template and known-formatter label. #9205 (closed) localized a notify node's own title and message, not a field's option label; the rest are QA run logs and unrelated mail cards.
Filed by the repo:hotcrm lane seat from hotcrm#2391's out-of-scope finding.
Activity
objectstack-fleet commentedon Nov 4, 2026
triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U
Path: none
Triage: first-touch audit · p3 kept ·
domain:server·pm:queue· 2026-11-04T20:55Z⛔ Not a claim.
templateDatais evaluated once, before the recipients fan out, and neither a template formatter nor CELlabel()reaches it. HotCRM measured it, and a stored value prints instead. p3 kept.service-automation'snotifyand the template renderer. The option's label belongs to the recipient's locale, so it resolves per recipient.Generated by Claude Code
objectstack-fleet commentedon Nov 5, 2026
domain:server#6 · os-bill · session_01Utp6CXCS98uhsFQYk64epE
Claim: PM loop round 1
Session:
session_01Utp6CXCS98uhsFQYk64epE(seat) · dev = project card threadsession_01Utp6CbWG4NyLTjvVq3tE4UAccount:
os-billBranch:
claude/project-thread-0q8hex(harness-pinned thread branch; platform exception toclaude/issue-<n>-<slug>)Worktree:
objectstack-issue-30172Domain:
domain:serverSeat:
domain:server#6File surface:
packages/services/service-automation/src/**, and only if the label lookup belongs therepackages/plugin-email/src/**orpackages/formula/src/**(CEL evaluation context, no grammar change), with tests;content/docs/**only for sentences this PR makes false (notreleases/**); one changeset. ⛔packages/spec/**,ui/**,skills/**,docs/adr/**,.claude/**,AGENTS.md(stop on breach; explain in the report)Container & model:
M,mode:thread,model: default tier (opus; dispatch-gates --tier: no path-derived mandate),effort: highClause-②: yes (widening) — provisional on route P: a flow that is refused today (an option label in a notify) would evaluate. The dev corrects the arm from the diff
Metadata-protocol: semantic (provisional) — what a notify's select value renders as per recipient; the dev confirms or corrects
Responsibility:
service-automation's notify evaluates templateData once before fan-out, and neither the renderer nor CEL label() reaches option labels there (measured by HotCRM on 18.3.0) | none: apps hand-build a per-locale label map (hotcrm#2406) | every app that sends a notice naming a select field to recipients of more than one languageThread-read: 6641888761
Premise: holds per the HotCRM measurement; the dev measures route P first and reports the route before writing code. A
packages/specneed (template grammar, NotifyConfigSchema key, CEL label() contract) stops as needs_decision.Serial constraints cleared:
none (no open PR or claim on service-automation notify at claim time; dev re-checks)Order: project file
server-seat6/dispatch-30172.md, delivered to the card thread.objectstack-fleet commentedon Nov 5, 2026
domain:server#6 · os-bill · session_01Utp6CXCS98uhsFQYk64epE
os-dev-report
决策： 流程通知里的选项值，按每个收件人的语言显示标签，落在哪条路。状态
needs_decision，无 PR，未写代码。dev = 卡线程session_01Utp6CbWG4NyLTjvVq3tE4U，分支claude/project-thread-0q8hex（零提交）。一句话问题
同一封流程通知发给中文和英文同事，选项字段（如商机阶段）印出的是存储值
needs_analysis，平台没有一条路能让每个人读到自己语言的标签；派单令首选的「仅 server 内」路线覆盖不了卡上复现的场景。三条读数（origin/main f273c9f）
role:/team:选择器），templateData在packages/services/service-automation/src/builtin/notify-node.tsL988 求值一次。收件人的 locale 第一次出现在packages/services/service-messaging/src/email-channel.tsL383（resolveRecipient返回locale）和inbox-channel.tsL198（resolveRecipientLocale），两者都把data+locale交给 plugin-email 的sendTemplate/renderTemplate。执行器里拿不到任何一位收件人的语言。label(record.FIELD)。 宏只认record作接收者（packages/formula/src/stdlib.tsoptionLabelFieldOfL429–434），对象由record所属对象隐含。写进templateData能过 parse，运行时因流程的 CEL 上下文不带optionLabels被拒（stdlib.ts L502–510，即卡上第 3 行）。按语言构造标签的机制已存在：packages/objectql/src/formula-option-labels.ts（readerOptionLabels，与/meta/object同一条翻译路径）。label(currentOpp.stage)，parse 即拒（卡上第 2 行）。定时 tick 没有触发记录，record不存在。要让它工作，必须改label()的契约（允许非record接收者并说明对象），派单令把这一条列为停手条件。Governing text
packages/spec/src/system/email-template.zod.tsL107–123MAIL_TEMPLATE_HOLES：「A formatter … is one the template engine knows (an unknown name … is refused …)」。新增 formatter 即改此句，属 spec 车道。packages/formula/src/types.tsL200–214EvalContext.optionLabels：「The option labels of the record's object, in the reader's locale」，label(record.FIELD)只读它。packages/spec/src/contracts/email-service.tsSendTemplateInput.data：「Render context — placeholders in subject/body are resolved against this object.」本报告的选项均不改这句。协议声明：P1 不改 spec 文字；F 改模板文法（spec）；L 改
label()契约（formula，影响 lint 与技能所教）。选项
label(record.F)的templateData项，对 i18n 服务持有的每个 locale 各求值一次（无 locale 用声明标签），payload 带每个 locale 的变体；email / inbox 两个 channel 按收件人 locale 用resolveBundleLocale挑一份packages/services/service-messaging/src/**；无实测先例{{stage | option:'crm_opportunity.stage'}}，plugin-email 渲染时（已持收件人 locale）按对象元数据加翻译包取标签currency:@path参数，行数未实测label()：如label(VAR.FIELD, 'object')，再加 P1 的 channel 改动业务含义直译：P1 像只给「从某张单据发起」的通知配翻译，批量巡检发出的仍是原始代码；F 像在邮件模板里放一个「按读者语言显示这一项」的占位符，谁发都一样；L 像让每个发信流程自己声明「这是哪张表的哪一列」再翻译。
四轴分析
renderTemplate）。P1 在执行器把每个 locale 预算一遍再让投递端挑，多一条两包之间的隐式 payload 约定，且只覆盖一半。L 让 CEL 与模板两处都背契约。label-field-no-options的拒法）。P1 下 AI 写label(currentOpp.stage)仍被拒，但写在定时流程里的label(record.stage)会在运行时失败，作者在保存时看不到。L 让 AI 必须写对两个参数，错拼对象名是运行时问题。os-decision-facets
Prior rulings read: "option label" formatter → 16 hits, "label(record" → 919 hits (top 6 read), known-formatter → 42 hits; none is a ruling on this question; ADR none; thread: none
推荐
推荐 F。只看①选 F：两年后的样子是模板占位符本身声明「这是哪个对象字段的选项」，渲染器按读者语言出标签，主流 CRM 的邮件模板合并字段同样以对象.字段引用写成（如 Salesforce 的
{!Opportunity.StageName}；其是否按收件人语言出标签未实测）。②③④ 是否翻转：否。置信缺口：F 的具体拼写（option:'obj.field'还是别的词）未与 spec 席位对齐，属 spec 车道裁定。P1 不取：覆盖不到真实场景。L 不取：两处契约同时扩。裁后执行
MAIL_TEMPLATE_HOLES加 formatter 句，formula 模板引擎认名，lint 作者规则），server 半卡挂Blocked-by:该卡，做 plugin-email 渲染 + 两收件人 en/zh-CN pin + 定时流 pin + 非选项值对照 pin。本线程可接 server 半。packages/services/service-messaging/src/**，本线程按派单令先红后绿开草稿 PR，卡用Part of，循环场景留卡。label()契约与技能，server 再按 P1 做。{ "issue": 30172, "status": "needs_decision", "branch": "claude/project-thread-0q8hex", "pr": null, "session": "session_01Utp6CbWG4NyLTjvVq3tE4U", "premise_still_valid": true, "summary": "Measured route P on origin/main f273c9fe before any code. The recipient locale first exists in service-messaging's email and inbox channels after fan-out, never in the notify executor, and the only declared field reference is CEL label(record.FIELD), whose receiver must be record; the card's reproduced case is a schedule loop variable (label(currentOpp.stage)), refused at parse. A per-recipient route P needs the service-messaging surface and still misses that case; covering it needs the template grammar (spec) or the label() contract, both stop conditions.", "tests": "none run: measurement only, no code written", "gates": "none run: no diff", "dogfood_battery": "n/a: no PR", "mcp_calls": "0", "api_writes": "1: POST issues/30172/comments (this report, via post-stamped)", "open_questions": [ { "question": "Where does a select option label in a flow notice resolve in each recipient's language?", "options": [ "P1: notify executor evaluates label(record.F) entries once per held locale, channels pick by recipient locale; needs service-messaging surface; record-triggered flows only", "F: a template formatter naming object.field, resolved by the plugin-email renderer in the recipient's locale; spec grammar change; covers every flow", "L: widen label() to name a non-record receiver and its object, plus P1's channel change" ], "recommendation": "F, because it is the only option that covers the reproduced schedule-loop notice and resolves at the one render point that already holds the recipient locale" } ], "security_findings": "none", "out_of_scope_findings": [] }Report taken 2026-11-05T12:41Z.
Generated by Claude Code
objectstack-fleet commentedon Nov 5, 2026
domain:server#6 · os-bill · session_01Utp6CXCS98uhsFQYk64epE
Release:
session_01Utp6CXCS98uhsFQYk64epE· reason: seat clock-out · next:needs-user-decisionResume-from: none
The server lane closes this seat on the maintainer's word 「当前任务处理完，合并后只留一个 server 席位」 (2026-11-05T16:41Z); the assignee is cleared and
pm:dispatchedremoved 2026-11-05T18:52Z. The card waits on the maintainer's letter (decision analysis 6658266694; the seat recommended F, which is spec-lane work first). No code was written on the card branch. On the ruling,domain:server#2(the remaining server seat) claims it afresh, or routes F's spec half todomain:spec.objectstack-fleet commentedon Nov 5, 2026
domain:server#2 · os-bill · session_01Utp6Cbvcmv8YWFhFJwzjKL
Ruling confirmed: letter F · recorded by
domain:server#2; routed todomain:specfirst (seat's split: the spec half is the gate); the server half is cut as its own cardBlocked-bythis one when the spec half lands. The card branchclaude/project-thread-0q8hexhas no commits and is dropped.Maintainer: 「同意」 to the recommendation, in the decision thread, 2026-11-05T21:43Z (project files
decisions/30172-流程通知选项标签按收件人语言-呈批-复审.md, seven fixed parameters).{{ path | option:'object.field' }}: the argument is a literal object.field (spelling is the spec seat's call); no@path.currency:@path, mail templates only; in-app title / message / confirmText refuse it on save as an unknown formatter.plugin-emailbuilds a label table per object.field in the template with the recipient's locale (thereaderOptionLabelstranslation path) before a synchronous render; no locale ⇒ declared labels; a value outside the options prints as is; a missing object/field or one with no options ⇒TEMPLATE_INVALID, the whole send refused.MAIL_TEMPLATE_HOLESgrammar sentence + the formula allow-list + the lint rule); the server card isBlocked-byit (plugin-email prefetch + render; pins: en and zh-CN recipients each see their own labels; a scheduled flow running as system still renders per recipient; a non-option value prints as is).objectstack-automationskill gains one line: option labels in notices use the formatter, notlabel()intemplateData.label()'s contract does not move; no P1.State change:
needs-user-decisionanddomain:serverremoved;domain:specandpm:queueadded.11 remaining items
objectstack-fleet commentedon Nov 6, 2026
domain:spec#2 · os-sales · session_01L6bhgUtH2CXA8Qb8bDBRom
os-dev-report
PR #30727 patch round after the delta review (draft, head
030a4930): the ladder's rung four. Anoptionhole naming a platform-prefixed object that no rung above carries is advised atwarning, its field unjudged, at every door; formula and lint answer it alike.56928a5bbefore it changed only the three changesets'Skills-docs:line tofollow-up #30731. Revised body at/mnt/project-files/briefs/30727-body-r3.md; the PR body was not patched.{ "issue": 30172, "round": "patch round 2 (delta review PASS with one should-fix, comment 6673435245)", "status": "done", "branch": "claude/project-thread-d8v0qv", "pr": "https://github.com/objectstack-ai/objectstack/pull/30727", "head": "030a4930", "session": "session_01L6bhgUtH2CXA8Qb8bDBRom", "premise_still_valid": true, "summary": "SHOULD-FIX (rung four): formula mail-hole-grammar gains optionObjectUnregisteredPlatform(ref, objects), true when the objects do not carry the name and it has a platform prefix (hasPlatformObjectPrefix: sys_, cloud_, ai_); optionFieldRefusal returns null for it, so judgeMailHoleGrammar neither refuses nor judges it. email-template-hole-grammar adds a warning for each such hole under the same rule id, naming the hole and object, the hint pointing at os explain object-reference-unregistered-platform and quoting OPTION_ARGUMENT_RULE; EmailTemplateHoleGrammarFinding.severity widens to error or warning. The sys_userz pin moved from error to warning, with an unprefixed userz control refused at error, and os validate and the publish door both answer it as a warning (publish errors empty, advisories carry it). A send that reaches such a hole with no label table still throws optionTableRefusal. Changesets (formula, lint) and email-templates.mdx updated. Item 2 (Skills-docs follow-up #30731 in all three changesets, word for word with the body) landed in 56928a5b.", "tests": "head 030a4930, formula rebuilt: formula test exit 0 (65 files, 1701 tests), formula typecheck 0, lint test 0 (196 files, 8198 tests), lint typecheck 0; eslint --no-inline-config on the 9 changed ts files exit 0. Ablation via ablation-replace: optionObjectUnregisteredPlatform gated on a __ablated_rung4 globalThis marker (anchor 1 to 0, blob 9d18678f to 25b7acc6), formula rebuilt, dist-preflight marker present in 2 built files; formula option test 1 failed / 48 passed (rung four pin), lint option test 2 failed / 22 passed (sys_userz warning pin, validate and publish agreement pin); restored blob 9d18678f equals HEAD, git diff HEAD empty, rebuilt, preflight --absent exit 0, whole tree clean, 49/49 and 24/24.", "gates": "head 030a4930: check-changeset-no-major --base origin/main 0, check-empty-changeset --base origin/main 0, pnpm check:changeset-gate-self-tests 0, check-changeset-fixed 0. The full derived set (102 families, 0 NOT-MEASURED) was reconciled on d73ae610; this round touches formula and lint source, tests, two changesets and one docs page, and was not re-derived, per the seat's scope. main not merged (21 ahead at 05a30e6e, merge-tree clean at d73ae610).", "dogfood_battery": "n/a — no ! in the PR title and Clause-② is yes (widening), no (narrowing) arm", "mcp_calls": "4 reads in this card thread so far — mcp__github__issue_read x2 on 30172, mcp__github__pull_request_read get_comments x1 and get_status x1 on 30727; none this round; no MCP write tool", "api_writes": "this report comment (POST issues/30172/comments via post-stamped relay); plus git push of 56928a5b and 030a4930 (not REST). PR body not patched", "open_questions": [], "security_findings": "none", "out_of_scope_findings": [] }Report taken 2026-11-06T00:50Z.
objectstack-fleet commentedon Nov 6, 2026
domain:spec#2 · os-sales · session_01L6bhgUtH2CXA8Qb8bDBRom
os-dev-report
PR #30727 patch round after the second delta review (draft, head
7cea4186): the publish door judges anoptionhole's object against the live universe alone and refuses a name it does not carry, whatever its prefix;os validate/os build/os lintkeep the ladder and rung ④'s warning. Revised body at/mnt/project-files/briefs/30727-body-r4.md; the PR body was not patched.{ "issue": 30172, "round": "patch round 3 (delta review PASS with one should-fix, comment 6673674554)", "status": "done", "branch": "claude/project-thread-d8v0qv", "pr": "https://github.com/objectstack-ai/objectstack/pull/30727", "head": "7cea4186", "session": "session_01L6bhgUtH2CXA8Qb8bDBRom", "premise_still_valid": true, "summary": "SHOULD-FIX (publish door over-relaxed): the registry entry passes validateEmailTemplateHoleGrammar(stack, { liveUniverse: ctx.runtimeWriteType !== undefined }). With liveUniverse the option index is objectsAsServed alone (no packages[] rung, no platform-name null rung), no rung-four warning is emitted, and judgeMailHoleGrammar gets liveUniverse: true (new formula option), under which optionFieldRefusal skips rung four, so sys_userz and a platform name the live universe does not serve are refused option-object-unknown at error, the list-view-object-unknown reading. The CLI doors are unchanged (sys_userz a warning, sys_user resolves). AuthoringRuleContext.runtimeWriteType's doc names this reader; the formula and lint changesets and email-templates.mdx state the split. Pins: side by side per hole (sys_userz, sys_user) os validate and os build from the ladder, the publish door error against a live universe of [opportunity]; control: publish accepts sys_user.status once the live universe serves sys_user with a select status; formula: liveUniverse true refuses sys_userz, cloud_tenantz, ai_agentz and still accepts a carried field.", "tests": "head 7cea4186, formula rebuilt: formula test exit 0 (65 files, 1702 tests), formula typecheck 0, lint test 0 (196 files, 8200 tests), lint typecheck 0; eslint --no-inline-config on the 9 changed ts files exit 0. Ablation via ablation-replace: the liveUniverse expression in authoring-rules.ts gated on a __ablated_live globalThis marker (anchor 1 to 0, blob 3cb1d3c9 to baa96acd); lint option test 2 failed / 24 passed (the sys_userz and sys_user door pins); restored blob 3cb1d3c9 equals HEAD, git diff HEAD empty, tree clean, 26/26.", "gates": "head 7cea4186: check-changeset-no-major --base origin/main 0, check-empty-changeset --base origin/main 0, pnpm check:changeset-gate-self-tests 0, check-changeset-fixed 0. Full derived set last reconciled on d73ae610 (102, 0 NOT-MEASURED); not re-derived, per the seat's scope. main not merged.", "dogfood_battery": "n/a — no ! in the PR title and Clause-② is yes (widening), no (narrowing) arm", "mcp_calls": "4 reads in this card thread so far — mcp__github__issue_read x2 on 30172, mcp__github__pull_request_read get_comments x1 and get_status x1 on 30727; none this round; no MCP write tool", "api_writes": "this report comment (POST issues/30172/comments via post-stamped relay); plus git push of 7cea4186 (not REST). PR body not patched", "open_questions": [], "security_findings": "none", "out_of_scope_findings": [] }Report taken 2026-11-06T01:01Z.
objectstack-fleet commentedon Nov 6, 2026
domain:spec#2 · os-sales · session_01L6bhgUtH2CXA8Qb8bDBRom
os-dev-report
PR #30727 patch round after the fourth delta review (draft, head
98f61ec6): a write into a package judges anoptionhole's object against that package's closure, and an object outside it gets the closure verdict and the declare-the-dependency remedy, not the unknown-object sentence;EmailTemplateHoleGrammarOptionsis exported. Revised body at/mnt/project-files/briefs/30727-body-r5.md; the PR body was not patched.{ "issue": 30172, "round": "patch round 4 (delta review PASS with two should-fixes, comment 6673851181)", "status": "done", "branch": "claude/project-thread-d8v0qv", "pr": "https://github.com/objectstack-ai/objectstack/pull/30727", "head": "98f61ec6", "session": "session_01L6bhgUtH2CXA8Qb8bDBRom", "premise_still_valid": true, "summary": "SHOULD-FIX 1 (package-scoped write): the registry entry passes packageScope beside liveUniverse when ctx.packageScope is set. formula MailHoleGrammarOptions gains packageScope { packageId } (implies liveUniverse) and the refusal rule option-object-outside-closure: names the object as outside package X's closure (its own objects, its declared dependencies' objects, platform objects), quotes OPTION_ARGUMENT_RULE and names the remedy, declaring the dependency in X's manifest. lint passes it and, for that rule, the hint is closureDependencyRemedy(object, packageId, 'name an object inside that closure') plus OPTION_ARGUMENT_RULE, the list-view-object-outside-closure reading. AuthoringRuleContext.packageScope's doc names the new reader. SHOULD-FIX 2: lint index exports EmailTemplateHoleGrammarOptions beside EmailTemplateHoleGrammarFinding. Pins: runtime gate write into com.acme.crm (dependency com.acme.core) with opportunity owned by an undeclared package gets the closure message (not 'resolves to no object') and the declare-the-dependency hint; an object of com.acme.core accepted; the same write naming no package accepted. formula: narrowed objects refuse deal and sys_userz as outside the closure and accept a carried field. Changesets and email-templates.mdx updated.", "tests": "head 98f61ec6, formula rebuilt: formula test exit 0 (65 files, 1703 tests), formula typecheck 0, lint test 0 (196 files, 8203 tests), lint typecheck 0; eslint --no-inline-config on the 9 changed ts files exit 0. Ablation via ablation-replace: the packageScope spread in the registry entry gated on a __ablated_closure globalThis marker (anchor 1 to 0, blob eaef25cc to 9662b429); lint option test 1 failed / 28 passed (the outside-closure pin); restored blob eaef25cc equals HEAD, git diff HEAD empty, tree clean, 29/29.", "gates": "head 98f61ec6: check-changeset-no-major --base origin/main 0, check-empty-changeset --base origin/main 0, pnpm check:changeset-gate-self-tests 0, check-changeset-fixed 0. Full derived set last reconciled on d73ae610 (102, 0 NOT-MEASURED); not re-derived, per the seat's scope. main not merged.", "dogfood_battery": "n/a — no ! in the PR title and Clause-② is yes (widening), no (narrowing) arm", "mcp_calls": "4 reads in this card thread so far — mcp__github__issue_read x2 on 30172, mcp__github__pull_request_read get_comments x1 and get_status x1 on 30727; none this round; no MCP write tool", "api_writes": "this report comment (POST issues/30172/comments via post-stamped relay); plus git push of 98f61ec6 (not REST). PR body not patched", "open_questions": [], "security_findings": "none", "out_of_scope_findings": [] }Report taken 2026-11-06T01:13Z.
objectstack-fleet commentedon Nov 6, 2026
domain:spec#2 · os-sales · session_01L6bhgVugtCCwBs98FeR27P
Release: claim 6671557736 (domain:spec#2), at 2026-11-06T03:31Z.
The maintainer ordered the lanes to finish what they hold and stop once it merges: 「所有车道当前任务处理完，合并后就下班」 (this account's project, 2026-11-06T01:17Z). This card cannot merge tonight: its landing pairs with the server half #30730, which no seat has claimed.
Resume-from: claude/project-thread-d8v0qv@98f61ec65b298cd6e78f38e3e5600a92e7c35861
optionformatter prints a select value as its option label in the recipient's language #30727 (Fixes #30172), body r5 patched to match the head. Contract review atCONTRACT_REVIEW_TIER: round 1 FAIL 6672632729, then delta PASS 6673435245 (d73ae610), 6673674554 (030a4930), 6673851181 (7cea4186) and 6674014531 (98f61ec6, no finding). Every round-1 must-fix and every later should-fix the seat adopted is in the head. No## 席位核对ACCEPT is posted yet.optionformatter with each recipient's option labels (server half of ruling F on #30172) #30730 lands,plugin-emailhands the renderer no label table, so anoptionhole that validates is refused at send (a bareRangeErrorfromformatValue, nothing sent). The carry note for plugin-email: render a mail template'soptionformatter with each recipient's option labels (server half of ruling F on #30172) #30730 is 6673488234: askoptionTableRefusalbeforeformatValueand prefixTEMPLATE_INVALID.optionformatter with each recipient's option labels (server half of ruling F on #30172) #30730's PR is built on top of this branch and goes green; then refresh this branch frommain(one more delta review, and a fulldispatch-gates --commands--ranrun on the refreshed head, since the last full run was ond73ae610), ACCEPT, land this PR, then land plugin-email: render a mail template'soptionformatter with each recipient's option labels (server half of ruling F on #30172) #30730 right after. ⛔ NoBlocked-by: #30730here: plugin-email: render a mail template'soptionformatter with each recipient's option labels (server half of ruling F on #30172) #30730 needs this card's exports, so that line would deadlock the pair.optionformatter, notlabel()in templateData (ruling F parameter 6, #30172) #30731 (theobjectstack-automationskill line) is the skills lane's; the PR body and changesets name it asSkills-docs: follow-up #30731.pm:queue, unassigned; any spec seat may claim it and resume from that sha.Generated by Claude Code
objectstack-fleet commentedon Nov 6, 2026
domain:server#5 · os-elon-musk · session_016WV4cyvF7SP65frVFAYLv7
Claim: PM loop round 4 · 2026-11-06T03:50Z (resuming the spec half released in 6676118909, cross-seat)
Session:
session_016WV4cyvF7SP65frVFAYLv7Account:
os-elon-musk(the seat's linked user asGET /useranswers it; the card's assignee)Branch:
claude/project-thread-d8v0qv(PR #30727's head branch, Resume-from98f61ec65b298cd6e78f38e3e5600a92e7c35861); no new push is planned: the head lands as it standsWorktree:
objectstack-issue-30172Domain:
domain:spec(carried bydomain:server#5)Seat:
domain:server#5Cross-seat declaration: no
domain:specseat has an open mailbox; triage answered that designation is not its to make and leaned that this seat carries it (6676310185 on #30661); the maintainer chose that this seat carries the spec half and lands it back to back with #30781 (the seat thread's decision card, 2026-11-06T03:49Z). The pair is #30727 (this card) then #30781 (#30730)File surface: unchanged from
domain:spec#2's claim 6671557736 (packages/spec/src/system/email-template.zod.tsand its regenerated references,packages/formula/src/**,packages/lint/src/**mail-template hole rules, the mail-template docs page, one changeset per package); ⛔plugin-email(#30730's),skills/**(#30731),ui/**,docs/adr/**,.claude/**Container & model:
M, carried by the seat (no new dev round)Clause-②: yes (widening)
Metadata-protocol: added
Responsibility: as in 6671557736
Thread-read: 6676118909
Premise: holds at PR #30727 head
98f61ec6: contract review delta PASS with no finding 6674014531 on that head; all check-runs on it green or skipped (59 success, 10 skipped); merges clean withmainas read at 2026-11-06T03:50ZSerial constraints cleared: #30781 is built on this head and lands right after; no other open PR touches
email-template.zod.ts; PR #30753 touchespackages/lint/src/index.tsand the tree merges cleanobjectstack-fleet commentedon Nov 6, 2026
domain:server#5 · os-elon-musk · session_016WV4cyvF7SP65frVFAYLv7
Landed · 2026-11-06T04:47Z · PR #30727
merged, merge commit3da9a4c1, through the merge queue (seat ACCEPT on head98f61ec6; contract review delta PASS 6674014531 on that head; not governed)Fixes #30172Shipped: the spec half of ruling F. The mail-template
optionformatter ({{ stage | option:opportunity.stage }}) is in the grammar, the lint rules and the renderer. Carried bydomain:server#5under the cross-seat claim on this card. The server half (#30730, PR #30781) lands right after it; until then a template usingoptionsaves and then fails at send.Skills-docs: follow-up #30731stays open for the skills lane.State change:
pm:dispatchedremoved, assignee cleared.