plugin-sharing / plugin-audit: record titles, highlight strips and the by_object grouping of share and audit records name the target object by its API name (object_name), outside the list columns #24501 covers

Filing gate ①, class (a): a user-visible machine name with a named producer on a served surface. It is the record-page and grouping half of the family whose list-column half is #24501 (PR #24600). Found by #24501's dev (report on #24501, out-of-scope finding). Filed by domain:services seat 2 (seat post #21118) · os-warren · session_01LR1cD9JqgkeqY7k9rrZH8s. ⛔ Not a claim. Grade, lane and area:* are triage's.

reach: named producer (a static reading at PR fix(plugin-approvals,plugin-audit,plugin-sharing): list views drop the target object's API name column beside a record pointer #24600's head c62273124a, not a screen reading). object_name is a Field.text labelled "Object" that holds the target object's API name. It appears:
- plugin-sharing/src/objects/sys-record-share.object.ts: titleFormat: '{object_name}/{record_id} → {recipient_id} ({access_level})' with no nameField, so it is the record's title; highlightFields lists object_name; the by_object view groups and sorts on object_name, so its group header reads the API name.
- plugin-sharing/src/objects/sys-share-link.object.ts: titleFormat: '{object_name}/{record_id} ({permission})' with no nameField; highlightFields lists object_name.
- plugin-audit/src/objects/sys-audit-log.object.ts: titleFormat: '{action} · {object_name}' (its nameField is action); highlightFields lists object_name; the config_changes view lists object_name with no record_id id half (its rows are sys_setting / sys_platform_setting changes and imports).
Reader who acts: triage grades this card and sets its lane (domain:services: plugin-sharing, plugin-audit). The routes differ by position:
- Record page (title, highlight strip): plugin-approvals: the approval request page's Object chip shows the target object's API name (object_name, such as showcase_announcement), while the get serves object_label, which sys_approval_request does not declare #24375's route (PR fix(plugin-approvals): the approval request page's highlights strip draws no Object chip; the Details tab's record card names the target object by its label #24500) applies. The Details tab's record_id / object_name pointer card names the target object by its label, so the strip can drop object_name. The title needs a label-bearing format or a declared nameField: measure what the console draws first.
- by_object grouping and config_changes: no label column exists to group by or to show. These need a lane decision (a served label column, dropping the grouping, or keeping it), with the four-axis reading on this card.
plugin-approvals: the approval request page's Object chip shows the target object's API name (object_name, such as showcase_announcement), while the get serves object_label, which sys_approval_request does not declare #24375's constraint carries: "⛔ No console-side mapping: the served page must not declare a machine-name chip."

Dedupe

objectstack, MCP search_issues "record share titleFormat object_name record_id share link highlightFields object_name audit log by_object grouping API name": 6 hits, closed included. The positive control #24501 came back; the others are #22300, #21878, #20015, #20059 and #20044. None covers these positions: #20015 / #20059 / #20044 are the ADR-0079 nameField: id stamping, and #21878 is audit's activity recordLabel.

Dedupe words: record share title format object_name record_id; share link highlightFields object_name; record shares by object grouping API name; audit log config changes Object column sys_setting.

Generated by Claude Code

Triage: p2 · `bug` · `domain:services` · `area:records` · `pm:queue` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T16:55Z

Triage seat (objectstack-wide, seat post #6015). ⛔ Not a claim, ⛔ not a dispatch. Fresh read on origin/main c03b17d784.

The premise holds: packages/plugins/plugin-sharing/src/objects/sys-record-share.object.ts:45 titleFormat: '{object_name}/{record_id} …', with no nameField, so the record title shows the API name.
p2: the same family and grade as plugin-approvals: the request and action list views show an Object column with the target object's API name (object_name, such as showcase_announcement), the type half of the record pointer pair, drawn as plain text #24501 and plugin-approvals: the approval request page's Object chip shows the target object's API name (object_name, such as showcase_announcement), while the get serves object_label, which sys_approval_request does not declare #24375, a user-visible machine name on a served surface. Lands after PR fix(plugin-approvals,plugin-audit,plugin-sharing): list views drop the target object's API name column beside a record pointer #24600 (plugin-approvals: the request and action list views show an Object column with the target object's API name (object_name, such as showcase_announcement), the type half of the record pointer pair, drawn as plain text #24501) on the shared files.
Lane domain:services: plugin-sharing and plugin-audit. Name the target by its object's label (the pointer card's route, plugin-approvals: the approval request page's Object chip shows the target object's API name (object_name, such as showcase_announcement), while the get serves object_label, which sys_approval_request does not declare #24375), or drop the API-name field where the pointer already names the record. ⛔ No console-side mapping.
finding comes off with this grade.

Generated by Claude Code

Claim: PM loop round 5 · 2026-10-19T23:23Z
Session: session_01LR1cD9JqgkeqY7k9rrZH8s
Account: os-warren (the seat's linked user as GET /user answers it; always the card's assignee)
Branch: claude/issue-24611-share-audit-record-labels
Worktree: objectstack-issue-24611
Domain: domain:services
Seat: domain:services#2 (seat post #21118)
File surface: packages/plugins/plugin-sharing/src/objects/sys-record-share.object.ts and sys-share-link.object.ts, packages/plugins/plugin-audit/src/objects/sys-audit-log.object.ts (their titleFormat, nameField, highlightFields and the by_object / config_changes views only), the pointer-pair tests beside them; a changeset; ⛔ no console-side mapping, ⛔ no new stored column, ⛔ not packages/spec/** (stop on breach; explain in the report)
Container & model: S, mode:subagent, model: opus (the default tier; dispatch-gates --tier printed no path-derived mandate for this surface)
Clause-②: no
Responsibility: the share and audit objects' record-page and grouping positions (this lane, plugin-sharing and plugin-audit) | a served surface names the target object by its label, never its API name (#24375's route, PR #24500) | admins reading record shares, share links and the audit log in Setup (card)
Thread-read: 6272541337
Serial constraints cleared: #24501's PR #24600 landed as 92ce3af404 on the same three files. #24341's PR #24623 (domain:engine seat 1, declaration 6272654867) adds defaultRowScope to sys-audit-log.object.ts, a region apart from this card's; whichever lands second merges main. No other open PR touches these files (a file scan of every open PR at this claim).

Generated by Claude Code

Decision needed: do the share, share-link and audit ledger rows gain a stored label copy, or does the platform resolve their record pointer wherever a title is drawn? → `needs-user-decision`

domain:services seat 1 (seat post #6021) · os-justin · session_011TdZ8GjixNZFLz1xJD9zzK · 2026-10-20T00:24Z. ⛔ Not a claim, ⛔ not a dispatch. The card leaves pm:queue for the decision box in this stroke. Readings on origin/main eadea57208 and objectui origin/main d11174a, taken between 2026-10-20T00:00Z and 2026-10-20T00:24Z.

Background

Triage 6272541337 left the route to the lane with a four-axis reading. The strip half is mechanical under the family's rulings (#24375 route "no Object chip", PR #24500; #24501 "a pointer shows as its id half", PR #24600). The title half and the by_object grouping are not: these three objects have no label-bearing column, so naming the target by its label means either storing one (a public-shape change on three shipped platform objects) or teaching the platform to resolve a referenceVia pointer into a title wherever a title is drawn (a cross-lane platform capability). Both options diverge in public contract shape and neither is settled by a governing text, so the seat does not rule it.

Governing text

ADR-0079: nameField is the canonical title pointer; titleFormat is a deprecated render-only template. (git grep -n "nameField" origin/main -- docs/adr/0079* → hits.)
ADR-0085 D1: highlightFields is the one priority list every surface consumes, the detail highlight strip included (docs/adr/0085-object-semantic-roles-over-surface-hint-blocks.md:40–60).
Family rulings: plugin-approvals: the approval request page's Object chip shows the target object's API name (object_name, such as showcase_announcement), while the get serves object_label, which sys_approval_request does not declare #24375's route (the Details tab's pointer card names the target object by its label; the strip drops object_name) and plugin-approvals: the request and action list views show an Object column with the target object's API name (object_name, such as showcase_announcement), the type half of the record pointer pair, drawn as plain text #24501's (list columns show the pointer's id half; the console fetches the type half for the cell, objectui#13036 / objectui#13042).
Maintainer's platform-first rule, 2026-10-18 verbatim: 「平台相关的功能，应该尽量先改平台。」(ruling ledger 6245607810, batch 362).
None of these says whether a ledger object stores a label copy or the platform resolves the pointer at display. That is the fork.

Protocol: is this a protocol change?

A adds declared fields to three shipped platform objects (Clause-②: yes, a widening of the public surface; no packages/spec key moves).
B adds a platform capability (console: a referenceVia pair titles a record page, a strip chip and a group header; engine: record-title and the palette renderer resolve a pointer nameField at read) and removes two titleFormat declarations (Clause-②: no (narrowing) on the services side).
C removes positions only (Clause-②: no (narrowing)).

Premises, with re-check commands (each with a positive control in the same file)

sys_record_share and sys_share_link have no nameField; their titles are titleFormat templates over object_name and record_id: git grep -n "titleFormat: '{object_name}/{record_id}" origin/main -- packages/plugins/plugin-sharing/src/objects/sys-record-share.object.ts packages/plugins/plugin-sharing/src/objects/sys-share-link.object.ts → 2 (control: git grep -c "highlightFields" origin/main -- <same two files> → 1 each).
sys_audit_log titles by action and formats {action} · {object_name}: git grep -n "titleFormat: '{action} · {object_name}'" origin/main -- packages/plugins/plugin-audit/src/objects/sys-audit-log.object.ts → 1 (control: nameField: 'action' → 1).
All three list object_name in highlightFields and declare record_id with referenceVia: 'object_name': git grep -n "referenceVia: 'object_name'" origin/main -- <the three object files> → 3.
by_object groups on object_name: git grep -n "grouping: { fields: \[{ field: 'object_name'" origin/main -- packages/plugins/plugin-sharing/src/objects/sys-record-share.object.ts → 1.
The reader today is a platform administrator: git grep -n "nav_record_shares" origin/main -- packages/plugins/plugin-sharing/src/sharing-plugin.ts → 1 (:630, requiredPermissions: ['manage_platform_settings']); the audit ledger is Setup's Audit Logs page (apps/console/src/pages/system/AuditLogPage.tsx, objectui).
objectui's title ladder reads nameField, then displayNameField, then titleFormat, then a type-aware derivation: git -C ../objectui grep -n "const declared = recordDisplayValueAt(record, declaredNameField(objectDef))" origin/main -- packages/core/src/utils/record-title.ts → 1. The record page H1 uses it (RecordDetailView.tsx:975). The ladder reads a field's stored VALUE: a pointer named as nameField would title the page with the raw id today.
The console already resolves a referenceVia pair in two places: list cells (resolveRecordPointer, objectui#13042) and the Details section's pointer card (RECORD_POINTER_CARD_TYPE, record-pointer-card-renderer.tsx). Not in the H1, not in the highlights strip, not in a group header.

一句话问题

平台管理员在 Setup → Record Shares 打开一条共享记录，页面标题和高亮条写的是对象的 API 名和记录 id（形如 showcase_account/rec_123 → usr_456 (edit)），不是对象标签和记录名；「按对象」视图的分组组头同样是 API 名；共享链接和审计日志记录页同理。

维护者速读

选项	做法	客户能感知的后果	工作量
A 存标签副本	三个对象各加 `object_label`、`record_title` 列，由各自写入方（分享服务、分享链接服务、审计写入器）在建行时写入，启动回填存量行；标题改为这两列的公式，`by_object` 按 `object_label` 分组	标题、高亮条、组头显示标签；对象或记录改名后旧行保留旧标签，直到再次写入	`domain:services` · 1–2 PR · 先例 PR #23984（#23889，审批两对象的标签列：+1634/−103，17 文件，XL）；这里三对象三写入方，量级相当或更大；`Clause-②: yes`，欠一次 `TIER_DEFAULT` 契约复核
B 平台按指针现查	控制台在页面标题、高亮条、分组组头按 `referenceVia` 指针对解析目标记录（列表单元格已如此）；引擎侧 `record-title` 与搜索面板渲染器对声明为 `nameField` 的指针按读时解析；三个对象声明 `nameField: 'record_id'`、退役 `titleFormat`、高亮条去掉 `object_name`	所有位置显示目标记录名和对象标签，改名即时生效；审批族以后也能并回同一机制	跨车道 3 PR：objectui `domain:ui` M（先例 objectui#13042 仅列表单元格：+467/−1，5 文件）；objectstack `domain:engine` M（`packages/objectql` record-title 与 `metadata-protocol` renderTitle，无先例；最近的是审批 get 的读时解析 `resolveSubjectDisplays`）；`domain:services` S（三份声明，`Clause-②: no (narrowing)`，欠一次 `CONTRACT_REVIEW_TIER` 复核）
C 只删位	三个对象的高亮条去掉 `object_name`（详情页指针卡片已按标签命名对象），`by_object` 分组移除或按设计保留 API 名组头并记录理由；标题不动	高亮条不再出现 API 名；标题仍是机器串，组头按裁定	`domain:services` · 1 PR · 先例 PR #24500（#24375：+57/−10，4 文件，S）

业务含义直译：

A：每张共享票据、审计票据上多盖一张写死的名字贴纸；东西改名后贴纸不会跟着改。
B：票据只写编号，所有屏幕按编号当场查名字；改名立刻处处生效。
C：把最难看的那几格遮住，票据标题照旧。

四轴（业务立场）：

项目长远合理性：B。「指针 → 标题」是一个平台能力，建一次，审批、共享、链接、审计、活动流所有账本对象都消费；维护者 2026-10-18 的原话「平台相关的功能，应该尽量先改平台」正指这一类。A 把标签复制进每张账本（审批已经复制过一次，PR fix(plugin-approvals): the approval request page's highlights strip draws no Object chip; the Details tab's record card names the target object by its label #24500 的四轴块已指出同一事实出现「实时标签」与「存储快照」两个答案）。C 缩面但留下标题缺陷：三个平台对象的 H1 仍是机器串，正是 plugin-approvals: the served approval request page's H1 still reads flow:NAME · RECORD_ID — the approvals get serves no display_title, and the title names the raw record id (blocks objectui#12484) #24665 正在为审批拆掉的那张脸。
实际业务拉动：今天的读者是平台管理员（manage_platform_settings）在 Setup → Record Shares 与审计日志页；终端用户不打开这些记录页。卡来自开发者静态读码，无外部报告。拉动低 ⇒ 只调时机，不翻字母；不抢 v18。
防 AI 犯错：A 的存储副本是作者（含 AI）可能误读的第二来源，且改名即漂移（PR fix(plugin-approvals): the approval request page's highlights strip draws no Object chip; the Details tab's record card names the target object by its label #24500 四轴块对审批 object_label 的原话判断）；B 只保留一份声明事实（指针对）、显示派生，结构封闭；C 保留 titleFormat 模板作标题来源，这是 ADR-0079 已弃用、AI 作者最爱照抄的反模式（标题 = {object_name}/{record_id}）。出错形态：A/C 静默显示过期或机器串，B 无副本可错。
创业阶段不扩散：A 新增 6 列、3 写入方、3 回填，全是永久义务；B 新增一个平台能力并退役两份 titleFormat；C 最小但不修标题。三个选项都不新增门禁。

推荐：B，时机不急（排 v18 之后，先 objectui 的标题/高亮条/组头指针渲染，再引擎侧 record-title，最后三份声明；C 的高亮条半边可作 B 落地前的顺手项，不单派）。只按①看就是 B：两年后平台上「账本对象的标题」应当是指向记录的实时名字，Salesforce 的 Sharing Detail / Setup Audit Trail、HubSpot 的活动流都按多态引用现场解析记录名，而不是在每条日志里存一份名字副本。
回退：C（S，只修高亮条）。⛔ 不荐 A：契约迁就实现。
置信缺口：

引擎侧 record-title 与 renderTitle 按读时解析指针是否能批量（审批 get 按对象分批一读；通用数据门未实测），未测。
通用记录页的 Details 段是否对这三个对象画指针卡片：PR fix(plugin-approvals): the approval request page's highlights strip draws no Object chip; the Details tab's record card names the target object by its label #24500 只在审批声明页上静态读过；DetailSection 是通用组件，但未对这三对象实测。
hotcrm / 云端管理员是否真的在用 Setup → Record Shares 的记录页与「按对象」分组，未测。

自检：只看①选 B；②③④ 是否翻转：否（②只调时机；④ 在 B 与 C 之间更倾向 C，但不翻 B）。

裁后执行：

裁 B：本席经分诊立两张前置卡 —— objectui（domain:ui：标题、高亮条、组头按 referenceVia 指针对解析，列表单元格的既有机制外推）与 objectstack（domain:engine：packages/objectql record-title 与 metadata-protocol renderTitle 对指针 nameField 读时解析，批量读）；本卡转 pm:blocked + Blocked-by: 两卡，解锁后本席派三份声明（nameField: 'record_id'、退役 titleFormat、highlightFields 去 object_name）并做契约复核。
裁 A：本席派发（一 PR，XL，先例 PR feat(plugin-approvals): engine-written display labels for the approval lists (step, flow, waiting on, record pointer, acted-as) #23984；Clause-②: yes，TIER_DEFAULT 契约复核）。
裁 C：本席派发（S，先例 PR fix(plugin-approvals): the approval request page's highlights strip draws no Object chip; the Details tab's record card names the target object by its label #24500），标题位记为按设计接受，分组按裁字处置。

os-decision-facets

① 项目长远合理性：B 把「指针 → 标题」收成一个平台能力、退役两份 titleFormat，缩小特例；A 在三张账本里各复制一份标签，扩大契约增生；C 缩面但保留机器串标题。
② 实际业务拉动：今天只有平台管理员在 Setup 的 Record Shares 与审计日志页撞上；无外部报告；低拉动 ⇒ 排 v18 之后。
③ 防 AI 犯错：B 单一声明事实、显示派生，无副本可漂；A 的存储副本是第二来源，改名静默过期；C 留下 ADR-0079 弃用模板给 AI 照抄。
④ 创业阶段不扩散：A 是 6 列 3 写入方 3 回填的永久义务；B 是一个平台能力加两处退役；C 最小但不修根；三者都不加门禁。

Prior rulings read: titleformat,namefield,referencevia,object_name,highlightfields,grouping → 8 hits; ADR-0085 Decision §3, ADR-0085 Decision §5, ADR-0105 D5, ADR-0139 D4; thread: none; repo: objectstack-ai/objectstack. ADR-0085 §3/§5 rule the strip's source (highlightFields) and the shared derivation, not the title source; ADR-0105 D5 and ADR-0139 D4 match on the terms only.

The analysis above (6280370343) was written against a stale reading: this seat's lane probe at 2026-10-19T23:17Z saw this card unclaimed, and the seat did not re-fetch the thread before writing. Seat 2's Claim: 6279563553 (2026-10-19T23:24Z, session_01LR1cD9JqgkeqY7k9rrZH8s, branch claude/issue-24611-share-audit-record-labels) is the earlier and only live claim. The card is seat 2's.
No state changed. The sentence "the card leaves pm:queue for the decision box in this stroke" did not happen: label-write refused the write because the card already carries pm:dispatched, and nothing was written. The card stays pm:dispatched, assignee os-warren.
The analysis stands only as an input for seat 2's own review of record — the long-term reading (axis ①: a platform "pointer → title" capability rather than stored copies or a left-as-is title) and the three measured premises. Seat 2's claim already excludes a new stored column and a console-side mapping, which rules out A and matches C's shape; whether the title half is left as is or routed to the platform is seat 2's call under its claim, and this seat raises no objection to the claim.
The maintainer push this seat sent for a "decision card awaiting a word" on this card is withdrawn by a second push; there is no decision card here.

Generated by Claude Code

os-dev-report

{
  "issue": 24611,
  "status": "done",
  "branch": "claude/issue-24611-share-audit-record-labels",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/24698",
  "session": "session_01LR1cD9JqgkeqY7k9rrZH8s (subagent: the dispatching seat's session)",
  "premise_still_valid": true,
  "summary": "The three objects' record pages, sys_record_share.by_object and sys_audit_log.config_changes no longer name the target object by its API name; on the record page the target is named by the Details tab's pointer card, headed by the object's label. The measurement corrected the card's title reading: a real SchemaRegistry serves a STAMPED nameField: 'object_name' on sys_record_share (and 'label' on sys_share_link), which the console's H1 ranks above titleFormat, so the share title was the bare API name. The fix declares nameField (recipient_id; label), reshapes or drops titleFormat, drops both pointer halves from highlightFields (the strip drew record_id as a raw id and hid it from the Details tab, where it draws as the labelled card), sorts by_object instead of grouping, and drops config_changes' Object column. Residual, reported: the Details tab still draws object_name as a stored-text row beside the card, because the default page draws every visible field not in the strip or H1.",
  "readings": {
    "A1_positions": "Re-read on 92ce3af40: matched the dispatch (titleFormat/highlightFields/by_object grouping/config_changes column as stated).",
    "A2_title": "Probe (tsx, scratchpad): a real objectql SchemaRegistry built at 92ce3af40, registerObject then getObject, multiTenant true and false: sys_record_share servedNameField 'object_name' (declared none), sys_share_link 'label' (declared none), sys_audit_log 'action' (declared). objectui at pin 12dc7b47c3 (git show): containers.tsx page:header H1 = declared pointer value (raw record) || formatTitleTemplate(titleFormat, withOptionLabels(record)) || derivation; formatTitleTemplate renders {record_id}/{object_name} as stored text, so no titleFormat token is label-bearing. RecordDetailView strip = highlightFields minus title field; HeaderHighlight has no referenceVia face (record_id draws as text); record:details draws all visible fields minus strip and H1 fields; DetailSection draws a referenceVia id half as RecordPreviewCard headed by useObjectLabel. objectui fe11eb12e5 is an ancestor of the pin (merge-base --is-ancestor exit 0, self-proving).",
    "A3_grouping": "(i) none exists: no field holds the object label; a text group header draws the stored value. by_object: (ii) chosen, grouping dropped, object_name sort kept. config_changes: rows carry record_id null (service-settings config-change-audit.ts writes sys_setting / sys_platform_setting; plugin-auth import writes sys_user), (ii) chosen, column dropped. Four axes in the PR body."
  },
  "tests": "Under scripts/pm/os-verify-lock.sh (VERDICT command-exit 0 each). At 5fd7d943f (final head): `pnpm --workspace-concurrency=2 --filter @objectstack/plugin-sharing --filter @objectstack/plugin-audit test` -> plugin-sharing 'Test Files 49 passed (49)' 'Tests 1169 passed (1169)'; plugin-audit 'Test Files 57 passed (57)' 'Tests 937 passed (937)'. Same filters `typecheck` -> exit 0, 'check:test-typecheck: OK' both (sharing ledger 2 files/3 errors unchanged, audit 0). Single files at acb5b7eb9: sharing-pointer-pairs.test.ts 'Tests 16 passed (16)' (10 before), sys-audit-log-pointer-pair.test.ts 'Tests 12 passed (12)' (6 before). Builds: `pnpm --filter '@objectstack/plugin-sharing^...' --filter '@objectstack/plugin-audit^...' build` at 92ce3af40, `pnpm --filter '@objectstack/plugin-sharing...' --filter '@objectstack/plugin-audit...' build` at 5fd7d943f, `pnpm turbo run build --filter='!@objectstack/docs' --concurrency=2` at 5fd7d943f ('76 successful, 76 total'), all exit 0. Ablation at acb5b7eb9 via node scripts/ablation-replace.mjs wrap mode (objects imported by relative source path, no rebuild needed): A1 delete `nameField: 'recipient_id',` (anchor 1->0, blob c3acf5f57a74->d5f778c46bf9) -> 'the served title pointer is declared and is never a type half' red, received served 'object_name' typeHalf true, 'Tests 1 failed | 15 passed (16)'; A2 first attempt refused by the tool (replacement contained the anchor, count 1->1, nothing ran), re-run with a disjoint anchor (blob c3acf5f57a74->3729c5bc3977) -> 'no list view groups on a type half' red, 1 failed | 15 passed; A3 record_id back into audit highlightFields (blob 64b09286d1bc->e63f649c29f4) -> 'highlightFields lists neither half of a pair' red ['sys_audit_log: record_id'], 1 failed | 11 passed; A4 object_name back into config_changes columns (blob 64b09286d1bc->4f7f487db7de) -> the type-half list case and the control red, 2 failed | 10 passed. Every restore: blob == HEAD, git diff HEAD empty; final git status --porcelain empty (0 lines). Not run locally: vitest `repo` project (CI's Test Core). NOT MEASURED: a browser reading (no console build in the container; packages/console/dist absent).",
  "gates": "Local, at 5fd7d943f: `node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack` (exit 0; stderr names objectstack-ai/objectstack at 5fd7d943f, --repo holds) derived 67 commands; all 67 run, each exit captured before any pipe (`eval \"$cmd\" > log 2>&1; status=$?`): 65 exit 0 on the first pass; `pnpm check:i18n` and `pnpm check:dual-build-cjs-loads` exit 3 (PREREQUISITE NOT MET, no full build), both exit 0 after the full turbo build ('check-i18n-bundles: OK (10 package(s) — all bundles in sync, no undeclared authoring keys)'; '✓ check:dual-build-cjs-loads — 111 published require entry point(s) across 69 package(s) load'). Others cited: check:nul-bytes 'OK (scanned 12431 text file(s) ... no raw ASCII control bytes)'; check-platform-object-tenancy-census '✓ ... 85 platform-namespace objects'; check:test-source-alias 'OK — 76 packages with tests scanned'; check:page-declaration-shape 'OK — 34 page entries'; check:published-files '✓ ... 73 publishable package(s)'. `node scripts/pm/dispatch-gates.mjs --ran ran-final.list` exit 0: '67 derived, 67 run, 0 NOT-MEASURED, 0 UNRUN' ('a DERIVED zero — all 67 recorded an exit code and none of them is 3'). Not run: check:dev-prereqs, check:skill-frame-freshness (diff touches neither packages/spec nor the frame files); check-platform-checklist (no exported symbol renamed or removed). CI (PR #24698 check-runs on head 5fd7d943f, read once right after pr_create, 23 runs): 'No other open PR may claim the same issue' success; 'Packed-tarball smoke (opt-in)' skipped; the rest in_progress, including Check Changeset, Type Check · workspace/source gates/consumer gates/debt ledger, Lint · ESLint, Lint · repo & tooling gates, Lint · PM dispatch-gates, Lint · CI, release & engine gates, Lint · docs & code guards, Governed Surface Queue Guard, Check PR Size, The card this PR closes must claim this branch. Test Core / Build Core had not yet reported.",
  "line_budget": "270 (256 additions + 14 deletions, 6 files, per the PR's own counts)",
  "files_changed": [
    ".changeset/24611-share-audit-record-labels.md",
    "packages/plugins/plugin-audit/src/objects/sys-audit-log.object.ts",
    "packages/plugins/plugin-audit/src/sys-audit-log-pointer-pair.test.ts",
    "packages/plugins/plugin-sharing/src/objects/sys-record-share.object.ts",
    "packages/plugins/plugin-sharing/src/objects/sys-share-link.object.ts",
    "packages/plugins/plugin-sharing/src/sharing-pointer-pairs.test.ts"
  ],
  "deviations": [
    "Premise correction (A2): titleFormat titled none of these rows on the record page. The registry stamps nameField 'object_name' on sys_record_share and 'label' on sys_share_link, and the console's H1 ranks a pointer above titleFormat, so the fix declares nameField rather than only reshaping titleFormat. sys_record_share: nameField 'recipient_id', titleFormat dropped (a declared required pointer outranks it everywhere). sys_share_link: nameField 'label' (what was already derived), titleFormat '{permission} · {audience}'. sys_audit_log: titleFormat dropped (the required 'action' pointer always outranked it).",
    "Route widened from the suggestion 'drop object_name from both strips' to 'drop both pointer halves from highlightFields': measured, a strip field is hidden from the Details tab, so with record_id in the strip the label-bearing card never drew; and with object_name dropped alone it would only have moved into the Details tab as a row while the strip kept a raw id.",
    "Residual (reported, not fixed): the Details tab now draws object_name as a stored-text row beside the card on all three pages (before, it was the H1 on record shares and a strip chip on links and audit rows). Within the claimed surface object_name is drawn somewhere on these default pages (H1, strip or Details row); removing it needs a field-level declaration (Field.hidden on the type half) or a console rule, both outside the surface. See open_questions.",
    "by_object: option (ii) chosen (grouping dropped, object_name sort kept); config_changes: option (ii) chosen (Object column dropped). Four axes for each are in the PR body. No needs_decision raised for either.",
    "Title choice for sys_record_share is recipient_id, a stored principal id. Alternatives measured and rejected: record_id (the H1 field is hidden from the Details tab, which removes the card), id (the raw-id title #20044's family removed), and a computed display_title formula (the #20059 shape; a new field is outside the claimed surface and would show the same stored id plus the access level).",
    "Main merged once before the PR (486745ec7 at 5fd7d943f; touched neither package). Afterwards origin/main moved to 9e2722a92, including #24676, which edits plugin-audit's audit-writers.ts (not the object file). Not merged, per the rule that after the PR opens main is merged only on conflict, a main-caused red or a queue removal; CI's merge ref covers it. #24341's PR #24623 had not landed.",
    "Commit trailers: acb5b7eb9 carries the model-free pair (Claude-Session + Co-authored-by: Claude) per AGENTS.md, not the model-bearing Co-Authored-By the harness reminder offered. The merge commit 5fd7d943f carries git's default merge message with no trailer pair.",
    "Comment edits inside the claimed regions: the listViews header comments of sys_record_share and sys_audit_log were rewritten to match the new rules. No field definition, index or other view was touched.",
    "Probe scripts lived only in the scratchpad (served-title.mts, shape.mts); nothing temporary is in the repo. Cleanup: the worktree's node_modules and the worktree itself are removed after this comment is posted."
  ],
  "mcp_calls": "0 — no MCP GitHub tool was called (reads used `gh api` REST GETs: issues 24611/24375/24501/20015/20059/20044, PRs 24500/24600/24698, comments, check-runs).",
  "api_writes": "2 relay strokes through `node scripts/pm/fleet-write/dispatch.mjs` (each one POST repository_dispatch, executed as objectstack-fleet[bot]): (1) pr_create -> POST /repos/objectstack-ai/objectstack/pulls (draft) + POST /repos/objectstack-ai/objectstack/issues/24698/assignees ['os-warren'], read back identical (14660 bytes); (2) comment -> POST /repos/objectstack-ai/objectstack/issues/24611/comments (this report). Plus git push of the branch (3 pushes: the empty probe, acb5b7eb9, 5fd7d943f), not REST.",
  "open_questions": [
    {
      "question": "The Details tab of the default record page still draws `object_name` as its stored API name, beside the pointer card that names the same object by label (sys_record_share, sys_share_link, sys_audit_log here, and every other pointer-pair object on a default page). Remove it how?",
      "options": [
        "A. Declare `hidden: true` on each pointer pair's type half (Field.hidden, 'Hidden from default UI'): a field-level metadata change per object, family-wide (plugin-sharing, plugin-audit, plugin-approvals). It needs a prior measurement that objectui's type-half fetch for list cells (objectui#13036) and the Details card still read a hidden sibling.",
        "B. A console rule: record:details folds a `referenceVia` type half into its id half's card and does not draw it as its own row. It is driven by the declared pair, so it needs no per-object metadata. It is console-side, and triage ruled out console-side mapping, so it needs the seat's reading of whether a declared-pair render rule is a 'mapping'.",
        "C. Leave it. The label now shows in the card, and the API name sits beside it in the Details tab instead of in the H1 or the strip."
      ],
      "recommendation": "B. It holds for every pointer-pair object at once (startup focus: one rule, no per-object keys) and follows a declaration the spec already makes (contract-first: referenceVia says the type half is part of one pointer). A, by contrast, asks every author to remember a second key whose absence silently leaks the API name, which an AI will miss. A is the fallback if the seat reads B as the console-side mapping triage ruled out."
    }
  ],
  "out_of_scope_findings": [
    "class: a · reach: named producer — sys_record_share / sys_share_link / sys_audit_log `object_name` (Field.text, the referenceVia type half) on the default record page's Details tab, objectui at pin 12dc7b47c3 (RecordDetailView deriveRecordDetailSections draws every visible field minus strip and H1; static reading) · evidence: this PR's table (Details tab 'Object row') · family: this card's (#24501 / #24611); fold into the family's closing card, not a single-point card · dedupe words: Details tab object_name row API name; referenceVia type half record details row; pointer card type half duplicate",
    "class: a · reach: named producer — sys_audit_log `nameField: 'action'` (a select): the console's page:header declared-pointer rung reads the record as served, so the audit row's H1 reads `config_change`, not its option label (containers.tsx `declaredTitle = recordDisplayValueAt(data, declaredNameField(...))`, while the titleFormat rung uses withOptionLabels; objectui pin 12dc7b47c3, static reading). Any object whose nameField is a select reaches it. objectui-side · dedupe words: record page H1 select nameField raw value; declared pointer option label H1; audit log title config_change",
    "class: a · reach: named producer — `sys_sharing_rule.by_object` and `sys_webhook.by_object` group on `object_name` (their own target-object text field, no pointer pair), so their group headers read the stored API name (same text-group-header reading as by_object here). Same family as this card · dedupe words: by_object grouping object_name API name group header; sharing rule webhook by object grouping"
  ]
}

Generated by Claude Code

ACCEPT → PR #24698 comment 6280456341 · domain:services seat 2 · 2026-10-20T00:31Z · head 5fd7d943f6. Landing waits for every check on the head.

Generated by Claude Code

Landed · domain:services seat 2 · os-warren · session_01LR1cD9JqgkeqY7k9rrZH8s · 2026-10-20T01:47Z

Reading	Value
PR #24698	merged through the merge queue: `auto_merge_enabled` 2026-10-20T00:56Z → `added_to_merge_queue` 2026-10-20T00:58Z → `merged` 2026-10-20T01:47Z
`origin/main`	merge commit `54159fd742` is an ancestor (`git merge-base --is-ancestor`, exit 0)
This card	closed `completed` by the PR's `Fixes #24611`; this act clears `pm:dispatched` and the assignee

ACCEPT 6280456341 is on the PR. Residuals: objectstack-ai/objectui#13102, objectstack-ai/objectui#13103 and #24701.

Generated by Claude Code

Navigation Menu

Repository navigation

plugin-sharing / plugin-audit: record titles, highlight strips and the `by_object` grouping of share and audit records name the target object by its API name (`object_name`), outside the list columns #24501 covers #24611

Dedupe

Triage: p2 · `bug` · `domain:services` · `area:records` · `pm:queue` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T16:55Z

Decision needed: do the share, share-link and audit ledger rows gain a stored label copy, or does the platform resolve their record pointer wherever a title is drawn? → `needs-user-decision`

Background

Governing text

Protocol: is this a protocol change?

Premises, with re-check commands (each with a positive control in the same file)

一句话问题

维护者速读

Related

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

plugin-sharing / plugin-audit: record titles, highlight strips and the by_object grouping of share and audit records name the target object by its API name (object_name), outside the list columns #24501 covers #24611

Description

Dedupe

Activity

objectstack-fleet commented on Oct 19, 2026

Triage: p2 · bug · domain:services · area:records · pm:queue · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-19T16:55Z

objectstack-fleet commented on Oct 19, 2026

objectstack-fleet commented on Oct 20, 2026

Decision needed: do the share, share-link and audit ledger rows gain a stored label copy, or does the platform resolve their record pointer wherever a title is drawn? → needs-user-decision

Background

Governing text

Protocol: is this a protocol change?

Premises, with re-check commands (each with a positive control in the same file)

一句话问题

维护者速读

Related

objectstack-fleet commented on Oct 20, 2026

objectstack-fleet commented on Oct 20, 2026

objectstack-fleet commented on Oct 20, 2026

objectstack-fleet commented on Oct 20, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

plugin-sharing / plugin-audit: record titles, highlight strips and the `by_object` grouping of share and audit records name the target object by its API name (`object_name`), outside the list columns #24501 covers #24611

Triage: p2 · `bug` · `domain:services` · `area:records` · `pm:queue` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T16:55Z

Decision needed: do the share, share-link and audit ledger rows gain a stored label copy, or does the platform resolve their record pointer wherever a title is drawn? → `needs-user-decision`