One-time migration: transfer non-seam repo:objectui / repo:cloud cards from the objectstack backlog to their target repos

[os-zhuang]

Maintainer ruling (2026-08-10)

「同意」— ruling on the PM's proposal that execution cards live in the repo where the fix lands, with the existing transfer-lane stock migrated via GitHub issue transfer. Full ruling context and the standing-rule amendment are in #7165. This card does NOT wait for #7165's SKILL PR to merge — its authority is the maintainer ruling directly (same pattern as #6904); #7165 is the durable record.

Scope

All OPEN issues in objectstack-ai/objectstack labeled repo:objectui or repo:cloud, EXCEPT seam cards. Snapshot 2026-08-10: 47 repo:objectui + 2 repo:cloud (#6954, #5852) — re-enumerate at execution time; the two label queries are the authoritative worklist.

Seam criterion (stay in objectstack): the card's content is the cross-repo ordering itself — contract-first chains (a spec/contract change whose objectui/cloud adaptation must wait on an objectstack PR), console-bump linkage chores (#6162 family), anything whose Blocked-by line against an objectstack artifact is the substance. Test per card: would the body still make sense if objectstack did not exist? Yes = transfer; no = seam, stays. Blocked cards #7100 / #6955 / #6954 and any card carrying a Blocked-by on an objectstack issue/PR need this judgment individually; when in doubt, leave in place and note why.

Mechanics

• GitHub issue transfer within the org (GraphQL transferIssue mutation, e.g. via gh api graphql; the MCP toolset has no transfer tool). Requires issues enabled on target — they are.
• Thread, assignees and same-named labels carry over (pm:queue / finding / needs-user-decision / pm:blocked / target:v17 all exist in objectui and cloud). The repo:objectui / repo:cloud labels do not exist in the targets and are dropped automatically — correct, they are meaningless there.
• Transferred issues get NEW numbers; old URLs redirect permanently, so existing references keep resolving.
• Post-transfer checks: (a) spot-check open PRs whose bodies reference transferred numbers with closing keywords (Fixes #N) — repoint to the full new reference if the keyword must still auto-close; (b) verify no transferred card kept a dangling Blocked-by pointing at a same-repo shorthand number that now means something else — rewrite such lines to full owner/repo#number form.

Post-transfer grading

The objectui and cloud whole-repo seats grade the arrivals into their own backlogs — including the target:v17 four-class check per #6903's one-producer-per-backlog clause. Coordination with #6904 (objectui one-time v17 clearing): whichever of the two runs second must cover the transferred cards, and should say so in its completion comment.

Executor

Triage seat (transfer is a routing action — single-producer territory), with the sibling seats doing post-transfer grading. Audit comment on each transferred issue is not required (the transfer event itself is the audit trail); one summary comment on this card listing old-number → new-number pairs is.

No domain:* label on this card — routing is the triage seat's single-producer territory.

[os-zhuang]

Routing: domain:devx applied directly by the PM session under the maintainer's standing directive (this session, 2026-08-10):「以后你应该直接路由」. Lane precedent: #6903.

Note for the devx lane PM: per the body's executor clause, the transfer operations themselves are routing actions executed by the triage seat — this card sits in the devx lane for inventory and coordination. Do not dispatch a dev agent onto the transfers; the dispatchable remainder (if any) is the post-transfer reference-repair checks. The body's closing line ("No domain:* label applied") predates the directive and is superseded by this comment.

Session: 01JaVVMrSxt7Tgi1uwEuDtH7

---

Generated by Claude Code

[os-zhuang]

Claim. Taken by the maintainer-facing PM session under the maintainer's direct-dispatch directive (2026-08-10):「7165、#7167 你直接帮我派发处理」.

• Session: 01JaVVMrSxt7Tgi1uwEuDtH7
• Execution shape: API operations only (REST issue-transfer endpoint; the session's GraphQL surface is restricted, REST is authenticated) — no code branch.
• Label: pm:queue → pm:dispatched (this comment is the pair)
• Next: per-card seam judgment over the current repo:objectui / repo:cloud stock, then batch transfer; the summary comment here will list every old number and its new home.

---

Generated by Claude Code

[os-zhuang]

Migration complete — 42/42 cards moved, 6 stayed by design. Closing as completed.

Deviation from the body's mechanics (recorded)

Native GitHub issue transfer is unavailable to this session's credential — the REST transfer endpoint returns 404 for fine-grained/app tokens (measured; other REST writes succeed). Executed as recreate-and-close instead: destination card = verbatim title + provenance header + original body with bare #NNN references rewritten to objectstack-ai/objectstack#NNN; every created body was byte-diffed against the transformed raw-API original (the MCP write path sanitizes angle-bracket fragments — five cards were repaired from raw-API ground truth and re-verified); source closed not-planned with a crosslink comment, thread retained there as authoritative history. No labels were dropped; target:v17 survived on objectui#4044, objectui#4051 and cloud#1222.

Old → new mapping (42)

objectstack	new home		objectstack	new home
#7148	objectui#4022		#5446	objectui#4021
#7147	objectui#4025		#5432	objectui#4026
#7146	objectui#4029		#5428	objectui#4032
#7138	objectui#4034		#5419	objectui#4035
#7120	objectui#4038		#5418	objectui#4039
#7119	objectui#4041		#5175	objectui#4044
#7074	objectui#4045		#5150	objectui#4047
#7066	objectui#4049		#5149	objectui#4051
#6975	objectui#4050		#5144	objectui#4053
#6968	objectui#4052		#5142	objectui#4055
#6950	objectui#4054		#5141	objectui#4020
#6669	objectui#4016		#5084	objectui#4024
#6653	objectui#4017		#5083	objectui#4028
#6639	objectui#4018		#5076	objectui#4030
#6411	objectui#4019		#5067	objectui#4033
#6337	objectui#4023		#5019	objectui#4037
#6325	objectui#4027		#4118	objectui#4040
#5968	objectui#4031		#4115	objectui#4043
#5768	objectui#4036		#4075	objectui#4046
#5751	objectui#4042		#3083	objectui#4048
#5666	objectui#4056		#5852	cloud#1222

Stayed in objectstack (6, seam criterion / in-flight guard)

• finding: permission.rowLevelSecurity.priority is a retiredKey tombstone, but objectui's PermissionAdvancedFacets still types it and seeds priority: 0 on every newly added RLS policy #7130, finding: liveness dashboard.json widgets.actionUrl closure claim ("Nothing anywhere read widget.actionUrl") no longer measures true — objectui's widget config panel reads and authors the widget-level key #7129 — pm:dispatched, in flight; never transfer mid-dispatch, they exit via their PRs.
• Studio: remove the global_nav mock command-palette preview, inspector and palette entries (objectui half of #6888) #7100, objectui: the Unpublished banner and per-app publish read _unpublished; the App Switcher keeps filtering on hidden (#4829 A1 objectui half) #6955, cloud: the AI materialization path stamps _unpublished: true instead of hidden: true (#4829 A1 cloud half) #6954 — pm:blocked seam cards; their objectstack halves (global_nav action location renders nowhere in the running app — the ⌘K palette never reads action metadata, but Studio previews a command-palette frame for it #6888, rest: filterAppForUser treats app hidden flag as builder-only access gate — built-in account app returns 404 for all normal users #4829-A1) are live cross-repo ordering.
• objectui 配套:#5055 退役 spec 的 WidgetManifest / WidgetSource 后,parity tripwire 按设计打红,packages/types/src/widget.ts 的锚点注释指向不存在的名字 #6275 — pm:on-hold; restart condition sits on the objectstack side (ADR-0049 enforce-or-remove: ui/widget.zod.ts 全文件 + ui/i18n.zod.ts 五个形状实测无门(#4001 批 16 改判) #5055 parity tripwire).
• console 的内嵌索引编辑器与 IndexSchema 漂移:where vs partial、枚举里的 brin —— 阻塞 #4001 对 IndexSchema 的收紧 #5247 — reverse seam: it blocks in-flight objectstack#4001; moving it would break that Blocked-by shorthand.

Notes for follow-up (small, non-blocking)

• objectstack#5149's source carried assignee hotlong; the recreated objectui#4051 has no assignee — re-claim if that assignment was live.
• A handful of bodies whose prose reads "objectui PR #N" had those bare shorthands expanded to objectstack full form (the source's own auto-links pointed there too); the surrounding prose disambiguates, flagged per card in the batch reports.
• No inbound closing-keyword repairs were needed: all migrated cards were queue/hold/decision state, none referenced by an open PR's Fixes line (in-flight cards were excluded from migration).

Post-transfer grading of arrivals now falls to the objectui / cloud whole-repo seats per #6903's one-producer-per-backlog clause (labels carried over, so the cards are already in their local state machines). Standing rule lands via #7165 / PR #7185.

Session: 01JaVVMrSxt7Tgi1uwEuDtH7

---

Generated by Claude Code

[os-zhuang]

Arrival-grading coverage statement (per this card's coordination clause with #6904 — the migration ran second, so the transferred cards' target:v17 four-class check is declared here).

All 42 arrivals graded on 2026-08-10 under the maintainer's directive 「objectui 任务你帮我现在执行 v17 补判」:

• Carried marks kept (3): objectui#4044 (ex DashboardWidget.chartConfig 在 dashboard 元数据面只有 showLegend 落地 —— 12 个已声明键里 11 个惰性,liveness 台账那一行高估了证据 #5175), objectui#4051 (ex Conditional-visibility predicates (visibleWhen / visibleOn) fail OPEN and silently — a broken predicate is indistinguishable from no predicate #5149), cloud#1222 (ex ADR-0057 hierarchy DEPTH: the resolver's tenant isolation never engages — plugin-sharing passes organizationId: null while the active org rides in tenantId #5852) — board labels survived the migration.
• New marks (2, both class ② declared-vs-enforced on a shipped surface): objectui#4027 (ex objectui 跟进(#5775):page:card 的 registry inputs 仍把已退役的 body 发布为设计器槽;page:section/footer/sidebar 三个注册项没有 inputs,设计器无法授权 children #6325 — designer registry still publishes the retired body slot) and objectui#4041 (ex object-grid 的 inputs 声明 filters(复数),渲染器只读 filter(单数):按已发布词汇写筛选条件静默无效 #7119 — published filters vocabulary silently ineffective; also a decision card, so the ruling is board-priority).
• Off-board (37): consistent with the 2026-08-06 full-clearing calibration these cards already passed through as objectstack backlog members — the i18n/display family (ex Console: list-view chrome (Grid / Gallery switcher, Avg: / Sum: aggregate footer) and the form dialog's a11y fallback text are hardcoded English #5084/Console: AddressField sub-labels and placeholders are hardcoded English literals — an address field is untranslatable on any non-English console #5083/Analytics surfaces resolve select-option labels but never apply the i18n bundle — reports and dashboards show English while lists and forms show the translation #5076/Console: number cell/display renderers hardcode a grouping Intl.NumberFormat('en-US') — an ordinal Field.number (a year) always renders 2,026, and no field property can turn it off #5067/Record activity feed prints the untranslated English field label and the raw lookup ID (Rating Owner: ∅ → oBK25BVuz…) #5142/Report charts render the raw dataset measure name as the axis title, ignoring the measure's label — the same measure on a dashboard chart resolves correctly #5141 …), tooling/lint gates, observations/findings, adaptation and feature cards all stay off by the standing default (improvements, refactors, observations, internal tooling, cosmetic polish do not block). No full objectui rescan was run — incremental-only posture respected.

objectui local board after this pass: 8 open target:v17; cloud: 1.

Session: 01JaVVMrSxt7Tgi1uwEuDtH7

---

Generated by Claude Code

[os-zhuang]

Correction to the arrival-grading comment above: the board figures I quoted were an estimate, not a measurement.

I wrote "objectui local board after this pass: 8 open target:v17; cloud: 1". The objectui figure was never measured — I asserted it from arithmetic on the pass rather than querying the label. Measured at 2026-08-10 ~05:00Z, label:target:v17 is:open per repo:

• objectui: 4 — [P3] objectui ActionDef's [key: string]: any absorbs every retired and misspelled action key — removing a spec key produces no compile error #4075, fix(driver-sql,service-analytics): bare-day upper bounds cover the whole day on Field.datetime (#3777) #4041, feat(runtime): route ledger + conformance guard for the dispatcher↔client surface (#3563) #3569, fix(create-objectstack): ship a .gitignore in scaffolded projects #3129
• objectstack: 6 — Studio: remove the global_nav mock command-palette preview, inspector and palette entries (objectui half of #6888) #7100, objectui: the Unpublished banner and per-app publish read _unpublished; the App Switcher keeps filtering on hidden (#4829 A1 objectui half) #6955, cloud: the AI materialization path stamps _unpublished: true instead of hidden: true (#4829 A1 cloud half) #6954, global_nav action location renders nowhere in the running app — the ⌘K palette never reads action metadata, but Studio previews a command-palette frame for it #6888, finding: /meta/_drafts serves DRAFT object schemas unmasked — the one ADR-0106 outlet left uncovered #6599, objectui 配套:#5055 退役 spec 的 WidgetManifest / WidgetSource 后,parity tripwire 按设计打红,packages/types/src/widget.ts 的锚点注释指向不存在的名字 #6275
• cloud: 1 — chore(deps)(deps-dev): bump the development-dependencies group across 1 directory with 11 updates #1222

Three of the arrival-pass marks are legitimately no longer on the open board, and none of it is a labeling failure:

• objectui#4027 (marked in this pass) was claimed and fixed within ~2 hours — closed completed at 04:11Z. The board mark did exactly what it is for.
• objectui#4051 was closed completed at 03:06Z.
• objectui#4044 had its target:v17 removed by the objectui whole-repo seat and was moved to pm:on-hold — that seat is this backlog's board producer per pm-dispatch SKILL: give the objectui local backlog its own target:v17 producer (release-board coverage gap) #6903, so this is its call, correctly exercised.

Also note the objectstack board fell 14 → 6 over the same window: the five finding-triage promotions marked this morning (#7130, #7129, #7057, #7049, #6964) all landed and closed.

The grading itself stands as reported; only the summary count was wrong. Recording the correction here rather than editing the comment, so the error stays visible.

Session: 01JaVVMrSxt7Tgi1uwEuDtH7

---

Generated by Claude Code