一线界面：进展预设短语、任务附件，列表与看板卡片按方案 p7/p16/p17 补齐

售前演示优先级 p0。 方案 v7 p7「一线执行：完成一条工作，一次点击」承诺了三件现在没有的事，p16/p17 画的列表和看板比现在的丰富。全部是元数据，不写代码。

1. 进展预设短语（p7 ③）

duly_task.progress 单选：on_time（已按期完成）/ distributed（已下发各部门）/ awaiting_feedback（待反馈）/ in_hand（处理中）+ 保留 note 自由文本作「自己写…」。

列表里行内可编辑（Edit inline 已经是平台能力，确认这个字段进了行内编辑）。
task.hook.ts 的 last_update_at 只在 status/note/skip_reason 变化时打戳——把 progress 加进那份列表（改了进展就是「有人在管」），并补测试；test/seed-history.test.ts 必须保持绿。

2. 任务附件（p7「附件自主上传」、p6「全程留痕」）

duly_task.attachments：平台文件字段（多文件）。可传可不传、不校验、不拦截完成——这是产品不变量（无证据门槛），写进字段 description 和一条测试（done 不要求附件）。先量平台的文件字段类型叫什么、StorageServicePlugin 已加载能不能直接用；能用就用，不能就如实记录上报。

3. 列表与看板卡片（p16、p17）

「我的工作 / 本周待办」列表列：状态 · 工作事项 · 来源 · 到期 · 最新进展（=progress 或 note）· 附件标记。到期列显示相对日期（平台是否支持——量一下，不支持就用 late_after 列代替「逾期 N 天」）。
看板（board）卡片字段：来源·到期·进展；泳道或分组按来源（自建 / 上级交办 / 系统派发）——量 kanban.cardFields / swimlaneField 能否表达；能就配。
记录页 fieldGroups：基本 / 进展与附件 / 历史。

不做

不加 action（objectui#7234 未修，渲染不出来）。
不做移动端。

冲突提示

本卡改 task.object.ts、task.view.ts、task.hook.ts。同批派发的清单审定卡片改 duty.*，仪表盘卡片改 dashboards/、datasets/——互不重叠，但都会碰 zh-CN.ts，后落地的合 main 重跑门。

验收

浏览器截图：列表行内选一个预设进展、上传一个附件、看板卡片显示来源/到期/进展
四道门绿；#18 的 i18n 门对新字段/选项变绿（要写中文）

Claiming this issue.

Session: session_01SqkTcrxUFci7nqXdbBSe2p
Branch: claude/issue-108-frontline-ui
Worktree: ../duly-issue-108 off main @ dd3c619

Scope per the issue: src/objects/task.object.ts, src/views/task.view.ts, src/hooks/task.hook.ts, plus zh-CN entries and tests. Not touching duty.*, dashboards/, datasets/, pages/ (parallel cards #107 / #109 / #13).

Generated by Claude Code

{
  "issue": 108,
  "status": "done",
  "branch": "claude/issue-108-frontline-ui",
  "pr": "https://github.com/objectstack-ai/duly/pull/115",
  "premise_still_valid": true,
  "summary": "All three parts shipped as metadata. (1) `duly_task.progress` — four preset phrases beside `note`, no default, inline-editable on `my_week` and the board; `progress` added to `task.hook.ts`'s `last_update_at` field list, with `attachments` deliberately left OFF it so an upload cannot be the cheap way to look busy. (2) `duly_task.attachments` — the platform `file` field, `multiple: true`; `storage` is in PLATFORM_ALWAYS_ON_CAPABILITIES so StorageServicePlugin mounts with nothing configured, and a real upload landed through the record form. Never a completion gate: pinned in the metadata (required / requiredWhen / no rule names it) and against a booted engine. (3) List columns in the deck's p16 order, board card face with source·due·progress, record page fieldGroups 基本/进展与附件/历史. Two changes the card did not ask for, both named in the PR: the board is scoped to `owner = {current_user_id}` (dragging a card writes status, and unscoped the first 100 of 186 rows were all `done` so Open/In-progress rendered empty), and the demo's in-flight tasks carry a phrase so the new column is not empty on every row.",
  "tests": "Four gates through the shared verify lock on the MERGED head 3b717e0 (origin/main merged after #110/#112 landed; no conflicts): `pnpm validate` → '✓ Validation passed (499ms)  Data: 5 Objects  61 Fields'; `pnpm typecheck` → `tsc --noEmit` exit 0; `pnpm test` → 'Test Files 31 passed (31) · Tests 789 passed (789)'; `pnpm build` → '✓ Build complete (755ms)'. Lock line: 'os-verify-lock: VERDICT command-exit 0 · held the lock 34s · waited 0s'. RED-FIRST ABLATION on the hook: `perl -0pi` reverted the field list to ['status','note','skip_reason'], mutation confirmed on disk by grepping BOTH spellings before reading any result ('real spelling now 0 (want 0) · mutated spelling now 1 (want 1)') — vitest then exited 1 with exactly the two new assertions red and 62 others green ('advances when a progress phrase is picked', 'advances when the phrase CHANGES, and not when it is re-sent'). No rebuild leg was needed: the suite imports src through vitest and pins artifactPath at a nonexistent file. Restore ran from a trap on EXIT/INT/TERM and was verified in both directions on disk (mutated spelling 0, real spelling 1) plus a clean `git status`. BROWSER (the acceptance, live `pnpm demo` on port 4108, Chromium 1194, my PIDs only, all killed after): inline-selected a preset phrase on a `My week` row and pressed Save All — API before `progress: null, last_update_at: 2026-08-03T07:00:00.000Z`, after `progress: \"in_hand\", last_update_at: 2026-09-02T05:00:35.098Z`, i.e. the hook stamped through a real HTTP write; uploaded a file into `attachments` via the record form — stored as `[{id, name: 'audit-cover-sheet.txt', size: 73, mimeType: 'text/plain', url: '/api/v1/storage/files/…'}]` with `last_update_at` unchanged, which is the asymmetry pin confirmed in the running app; record page rendered Basics / Progress and attachments / History; board rendered OPEN 5 · IN PROGRESS 1 · DONE 20 · SKIPPED 0 with source and progress chips on the cards; relative dates rendered with no `format` key ('Overdue 7d' in red, 'In 3 days'); zh-CN console showed 最新进展 / 附件 / 待反馈 / 已下发各部门 / 处理中 / 逾期 7 天 and the 基本 / 进展与附件 / 历史 headings. 8 screenshots pushed to the throwaway branch `screenshots/issue-108` and embedded in the PR.",
  "open_questions": [
    {
      "question": "The card says of swimlanes '能就配' — configure them if they can be expressed. They CAN (view-level `grouping`, not `kanban.swimlaneField`), and I still left them off. Confirm that call, or tell me to turn them on.",
      "options": [
        "A — ship without lanes (what the PR does): the status column-header row renders at HEIGHT 0 in swimlane mode on console 17.2.0, so a laned board has no OPEN/IN PROGRESS/DONE labels at all. `source` sits on every card face instead, and the one-line re-enable is in the view comment. Filed as objectstack-ai/objectui#7303.",
        "B — turn lanes on now and accept an unlabelled board until #7303 lands: the deck's p17 layout is literally satisfied, and the demo screen loses the one thing a kanban is for."
      ],
      "recommendation": "A, because this is a p0 pre-sales screen and a board whose columns are unlabelled is not a board — a viewer cannot tell Open from Done except by dragging a card. Both halves are measured, not inferred: with `grouping` authored the lanes render (`▼ CATALOG (19)`) and the five titles sit in the DOM at y=233 inside a flex container whose measured height is 0, with a hit test at a title's own centre returning the lane's collapse button behind it; with the key absent the same board renders them as real <h3>s. Worth flagging that my first draft of this comment claimed swimlanes were IMPOSSIBLE and would have filed a bug that does not exist — that reading came from the console's other relay (ObjectView, which forwards no `grouping`), and these views do not go through it. The browser corrected it."
    }
  ],
  "out_of_scope_findings": [
    "filed as objectstack-ai/objectui#7303: kanban swimlanes render the status column-header row at height 0, so a laned board loses its column labels (measured both ways on console 17.2.0, with DOM coordinates and a hit test)",
    "not filed, recorded here: in a zh-CN console the attachments cell renders the count as the console's own untranslated string ('1 file' beside 附件). It is a platform i18n hole, not ours — cosmetic, and I did not want to file it without measuring which key is missing."
  ]
}

Generated by Claude Code

os-dev-report

(Re-posted as literal text: the HTML-comment marker on the identical report above was stripped by GitHub's body sanitizer after it landed, which makes that copy invisible to a marker scan. Same JSON, nothing changed.)

{
  "issue": 108,
  "status": "done",
  "branch": "claude/issue-108-frontline-ui",
  "pr": "https://github.com/objectstack-ai/duly/pull/115",
  "premise_still_valid": true,
  "summary": "All three parts shipped as metadata. (1) `duly_task.progress` — four preset phrases beside `note`, no default, inline-editable on `my_week` and the board; `progress` added to `task.hook.ts`'s `last_update_at` field list, with `attachments` deliberately left OFF it so an upload cannot be the cheap way to look busy. (2) `duly_task.attachments` — the platform `file` field, `multiple: true`; `storage` is in PLATFORM_ALWAYS_ON_CAPABILITIES so StorageServicePlugin mounts with nothing configured, and a real upload landed through the record form. Never a completion gate: pinned in the metadata (required / requiredWhen / no rule names it) and against a booted engine. (3) List columns in the deck's p16 order, board card face with source·due·progress, record page fieldGroups 基本/进展与附件/历史. Two changes the card did not ask for, both named in the PR: the board is scoped to `owner = {current_user_id}` (dragging a card writes status, and unscoped the first 100 of 186 rows were all `done` so Open/In-progress rendered empty), and the demo's in-flight tasks carry a phrase so the new column is not empty on every row.",
  "tests": "Four gates through the shared verify lock on the MERGED head 3b717e0 (origin/main merged after #110/#112 landed; no conflicts): `pnpm validate` → '✓ Validation passed (499ms)  Data: 5 Objects  61 Fields'; `pnpm typecheck` → `tsc --noEmit` exit 0; `pnpm test` → 'Test Files 31 passed (31) · Tests 789 passed (789)'; `pnpm build` → '✓ Build complete (755ms)'. Lock line: 'os-verify-lock: VERDICT command-exit 0 · held the lock 34s · waited 0s'. RED-FIRST ABLATION on the hook: `perl -0pi` reverted the field list to ['status','note','skip_reason'], mutation confirmed on disk by grepping BOTH spellings before reading any result ('real spelling now 0 (want 0) · mutated spelling now 1 (want 1)') — vitest then exited 1 with exactly the two new assertions red and 62 others green ('advances when a progress phrase is picked', 'advances when the phrase CHANGES, and not when it is re-sent'). No rebuild leg was needed: the suite imports src through vitest and pins artifactPath at a nonexistent file. Restore ran from a trap on EXIT/INT/TERM and was verified in both directions on disk (mutated spelling 0, real spelling 1) plus a clean `git status`. BROWSER (the acceptance, live `pnpm demo` on port 4108, Chromium 1194, my PIDs only, all killed after): inline-selected a preset phrase on a `My week` row and pressed Save All — API before `progress: null, last_update_at: 2026-08-03T07:00:00.000Z`, after `progress: in_hand, last_update_at: 2026-09-02T05:00:35.098Z`, i.e. the hook stamped through a real HTTP write; uploaded a file into `attachments` via the record form — stored as `[{id, name: 'audit-cover-sheet.txt', size: 73, mimeType: 'text/plain', url: '/api/v1/storage/files/…'}]` with `last_update_at` unchanged, which is the asymmetry pin confirmed in the running app; record page rendered Basics / Progress and attachments / History; board rendered OPEN 5 · IN PROGRESS 1 · DONE 20 · SKIPPED 0 with source and progress chips on the cards; relative dates rendered with no `format` key ('Overdue 7d' in red, 'In 3 days'); zh-CN console showed 最新进展 / 附件 / 待反馈 / 已下发各部门 / 处理中 / 逾期 7 天 and the 基本 / 进展与附件 / 历史 headings. 8 screenshots pushed to the throwaway branch `screenshots/issue-108` and embedded in the PR.",
  "open_questions": [
    {
      "question": "The card says of swimlanes '能就配' — configure them if they can be expressed. They CAN (view-level `grouping`, not `kanban.swimlaneField`), and I still left them off. Confirm that call, or tell me to turn them on.",
      "options": [
        "A — ship without lanes (what the PR does): the status column-header row renders at HEIGHT 0 in swimlane mode on console 17.2.0, so a laned board has no OPEN/IN PROGRESS/DONE labels at all. `source` sits on every card face instead, and the one-line re-enable is in the view comment. Filed as objectstack-ai/objectui#7303.",
        "B — turn lanes on now and accept an unlabelled board until #7303 lands: the deck's p17 layout is literally satisfied, and the demo screen loses the one thing a kanban is for."
      ],
      "recommendation": "A, because this is a p0 pre-sales screen and a board whose columns are unlabelled is not a board — a viewer cannot tell Open from Done except by dragging a card. Both halves are measured, not inferred: with `grouping` authored the lanes render (`▼ CATALOG (19)`) and the five titles sit in the DOM at y=233 inside a flex container whose measured height is 0, with a hit test at a title's own centre returning the lane's collapse button behind it; with the key absent the same board renders them as real h3 elements. Worth flagging that my first draft of this comment claimed swimlanes were IMPOSSIBLE and would have filed a bug that does not exist — that reading came from the console's other relay (ObjectView, which forwards no `grouping`), and these views do not go through it. The browser corrected it."
    }
  ],
  "out_of_scope_findings": [
    "filed as objectstack-ai/objectui#7303: kanban swimlanes render the status column-header row at height 0, so a laned board loses its column labels (measured both ways on console 17.2.0, with DOM coordinates and a hit test)",
    "not filed, recorded here: in a zh-CN console the attachments cell renders the count as the console's own untranslated string ('1 file' beside 附件). It is a platform i18n hole, not ours — cosmetic, and I did not want to file it without measuring which key is missing."
  ]
}

Generated by Claude Code

Clear the 18 field-no-consumers warnings by grouping every field on i…

Navigation Menu

Repository navigation

一线界面：进展预设短语、任务附件，列表与看板卡片按方案 p7/p16/p17 补齐 #108

1. 进展预设短语（p7 ③）

2. 任务附件（p7「附件自主上传」、p6「全程留痕」）

3. 列表与看板卡片（p16、p17）

不做

冲突提示

验收

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

一线界面：进展预设短语、任务附件，列表与看板卡片按方案 p7/p16/p17 补齐 #108

Description

1. 进展预设短语（p7 ③）

2. 任务附件（p7「附件自主上传」、p6「全程留痕」）

3. 列表与看板卡片（p16、p17）

不做

冲突提示

验收

Activity

os-warren commented on Sep 2, 2026

os-warren commented on Sep 2, 2026

os-warren commented on Sep 2, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions