Redesign the workbench visual hierarchy #556

Merged
weichen merged 22 commits from feature/workbench-visual-redesign into main 2026-10-15 02:44:34 +00:00
Owner

Summary

Why: The workbench remained visually flat and difficult to scan even after its middle/right responsibilities were corrected. Navigation, conversation, and current work now use clearly different surfaces, while the conversation keeps a focused reading width and progressively folded process detail.

- white and gray surfaces with weak hierarchy across all three columns
+ dark navigation frame, distinct authored/result messages, focused work rail

Impact: Pin the merged Frontend and UI revisions together and make the new chrome tokens part of DESIGN.md. Fast/slow task behavior, schemas, permissions, and model calls do not change.

Refs: task record, Frontend #87, UI #56.

Evidence

  • The old 1920px conversation measured 1138px wide; the new visual regression caps the reading column at 840px and the authored message at 680px.
  • 19 isolated Chromium scenarios pass across light, dark, molv and 320–1920px, including process-history reload, middle/right deduplication, files and confirmations.
  • Frontend 273 tests, UI 163 tests, production build, and the current combination make test pass locally (SELF receipts in the task record).
  • UI 4cae1c4 passed exact-head GATE CI before merge. Frontend has no repository workflow; Dev combination CI is the final platform gate.

Sync and review

The candidate contains current Dev main cc40e10 and preserves the separately merged 8788 message-permission repair and active simplified merge policy. Independent review found one Important design-system gap: component-local colors/private motion and missing theme coverage. Frontend 9b9f24b, UI 4cae1c4, the expanded design guards, and three-theme browser checks close it; no Critical or Important visual finding remains.

The first Dev CI attempt exposed two unrelated timing flakes. A test-fixture-only follow-up disables Git auto-maintenance for every disposable task repository, closing the observed .git cleanup race without changing governance behavior. The original failing test passes 25/25, isolated make test-governance passes 344 tests, and normal make test passes on the updated branch.

Merge Danger

Door: two-way

Blast Radius: workbench presentation across all themes and responsive widths

The running 8788 RDM composition remains unchanged until this Dev combination passes exact-head CI and is adopted separately.

## Summary Why: The workbench remained visually flat and difficult to scan even after its middle/right responsibilities were corrected. Navigation, conversation, and current work now use clearly different surfaces, while the conversation keeps a focused reading width and progressively folded process detail. ```diff - white and gray surfaces with weak hierarchy across all three columns + dark navigation frame, distinct authored/result messages, focused work rail ``` Impact: Pin the merged Frontend and UI revisions together and make the new chrome tokens part of `DESIGN.md`. Fast/slow task behavior, schemas, permissions, and model calls do not change. Refs: [task record](https://git-home.u1680832.nyat.app:30319/weichen/Agent-System-Dev/src/branch/feature/workbench-visual-redesign/.task/feature-workbench-visual-redesign.md), [Frontend #87](https://git-home.u1680832.nyat.app:30319/weichen/Agent-System-Frontend/pulls/87), [UI #56](https://git-home.u1680832.nyat.app:30319/weichen/Agent-System-UI/pulls/56). ## Evidence - The old 1920px conversation measured 1138px wide; the new visual regression caps the reading column at 840px and the authored message at 680px. - 19 isolated Chromium scenarios pass across light, dark, molv and 320–1920px, including process-history reload, middle/right deduplication, files and confirmations. - Frontend 273 tests, UI 163 tests, production build, and the current combination `make test` pass locally (SELF receipts in the task record). - UI `4cae1c4` passed exact-head GATE CI before merge. Frontend has no repository workflow; Dev combination CI is the final platform gate. ## Sync and review The candidate contains current Dev main `cc40e10` and preserves the separately merged 8788 message-permission repair and active simplified merge policy. Independent review found one Important design-system gap: component-local colors/private motion and missing theme coverage. Frontend `9b9f24b`, UI `4cae1c4`, the expanded design guards, and three-theme browser checks close it; no Critical or Important visual finding remains. The first Dev CI attempt exposed two unrelated timing flakes. A test-fixture-only follow-up disables Git auto-maintenance for every disposable task repository, closing the observed `.git` cleanup race without changing governance behavior. The original failing test passes 25/25, isolated `make test-governance` passes 344 tests, and normal `make test` passes on the updated branch. ## Merge Danger **Door:** two-way **Blast Radius:** workbench presentation across all themes and responsive widths The running 8788 RDM composition remains unchanged until this Dev combination passes exact-head CI and is adopted separately.
weichen added 15 commits 2026-10-14 17:01:50 +00:00
Why: establish the authorized task workspace and its initial baseline.

Evidence: preserved precheck receipt in .task/feature-workbench-visual-redesign.md; R0=99119a733830bc120cd6747b8b61736ea6da6078.

Impact: task metadata only; the feature workspace starts at 99119a7338.

Refs: .task/feature-workbench-visual-redesign.md
Why: The deployed workbench needs a clear visual hierarchy across context, conversation, and work without adding new product concepts.
Evidence: Frontend b04f4de passes 272 unit tests, 16 browser scenarios across 320-1920px, and a production build locally.
Impact: Advances only the Frontend pin and records the approved visual increment; runtime adoption remains separately verified.
Refs: feature/workbench-visual-redesign
Why: The first spacing-focused candidate still resembled the flat gray workspace and did not meet the user's visual target.
Evidence: Frontend ddf3135 passes 272 unit tests, 16 browser scenarios across 320-1920px, and a production build locally.
Impact: Advances the Frontend pin to the visibly differentiated navigation, conversation, and work surfaces.
Refs: feature/workbench-visual-redesign
Why: The review candidate needs reproducible evidence bound to the exact Dev and Frontend commits.
Evidence: SELF receipts record 272 unit tests, the production build, and 16 browser scenarios on Dev 49ef24b.
Impact: Updates only the task record; product behavior is unchanged.
Refs: feature/workbench-visual-redesign
Why: The inverse create control needed its divider and hover state adapted to the new dark navigation surface.
Evidence: Frontend 8dc68fd passes the focused browser layout test and production build locally.
Impact: Advances the Frontend pin and records the final local candidate.
Refs: feature/workbench-visual-redesign
Why: The review candidate needs reproducible evidence bound to the exact Dev and Frontend commits.
Evidence: SELF receipts record 272 unit tests, the production build, and 16 browser scenarios on Dev 49ef24b.
Impact: Updates only the task record; product behavior is unchanged.
Refs: feature/workbench-visual-redesign
Why: include the current integration baseline before verification.

Evidence: fetched origin/main=33261925a10c4eee40d1b6175b3895e78a47ef02.

Impact: integrate upstream changes without rewriting task history.

Refs: task-sync
Why: bind the task validation checkpoint to the integration baseline.

Evidence: SELF recursive component checks; sync_base=33261925a10c4eee40d1b6175b3895e78a47ef02; before_record=a09b4ae35ff36189797b14ee386cbb063eb6bc51.

Impact: task metadata only.

Refs: .task/feature-workbench-visual-redesign.md
Why: Independent review found that the visual frame bypassed token and motion ownership and lacked dark/molv proof.
Evidence: UI 4cae1c4 passes 163 tests; Frontend 9b9f24b passes its design guard and 19 browser scenarios across all themes.
Impact: Pairs the new UI tokens with their Frontend consumer and documents the governing design rule.
Refs: Frontend #87; UI #56
Why: The review candidate needs reproducible evidence bound to the exact Dev and Frontend commits.
Evidence: SELF receipts record 272 unit tests, the production build, and 16 browser scenarios on Dev 49ef24b.
Impact: Updates only the task record; product behavior is unchanged.
Refs: feature/workbench-visual-redesign
Why: The Dev composition must consume the reviewed Frontend and UI changes from their authoritative main branches.
Evidence: Frontend and UI merge trees equal the locally verified candidate trees; UI exact-head CI passed.
Impact: Advances only the Frontend and UI pins and records the component merge results.
Refs: Frontend #87; UI #56
Why: The review candidate needs reproducible evidence bound to the exact Dev and Frontend commits.
Evidence: SELF receipts record 272 unit tests, the production build, and 16 browser scenarios on Dev 49ef24b.
Impact: Updates only the task record; product behavior is unchanged.
Refs: feature/workbench-visual-redesign
Why: include the current integration baseline before verification.

Evidence: fetched origin/main=b8edf13dc7b818d6e61b829f6e88202ec553f2fc.

Impact: integrate upstream changes without rewriting task history.

Refs: task-sync
Why: bind the task validation checkpoint to the integration baseline.

Evidence: SELF recursive component checks; sync_base=b8edf13dc7b818d6e61b829f6e88202ec553f2fc; before_record=9bbd10dce38bdf7fd3bbfc94c90e37cdf19fdae3.

Impact: task metadata only.

Refs: .task/feature-workbench-visual-redesign.md
Record visual redesign verification evidence
ci / svc-realcombo (pull_request) Successful in 2m40s
ci / test-pg (pull_request) Failing after 3m3s
ci / test (pull_request) Successful in 6m47s
4f290b4018
Why: The review candidate needs reproducible evidence bound to the exact Dev and Frontend commits.
Evidence: SELF receipts record 272 unit tests, the production build, and 16 browser scenarios on Dev 49ef24b.
Impact: Updates only the task record; product behavior is unchanged.
Refs: feature/workbench-visual-redesign
weichen added 1 commit 2026-10-14 17:02:16 +00:00
Link the visual redesign delivery PR
ci / test-pg (pull_request) Successful in 9m59s
ci / svc-realcombo (pull_request) Successful in 2m58s
ci / test (pull_request) Failing after 3m30s
3e70537cab
Why: The task record must point to the concrete Dev combination under review.
Evidence: Dev PR #556 targets current main and pins the merged Frontend and UI trees.
Impact: Updates only delivery metadata.
Refs: Dev #556
Author
Owner

[release blocked] workbench-visual-20261015-01 / Dev PR #556

  • fixed head: 3e70537cab1cfa112d03f4b2314c5d38d2bb84c1; run: 3237
  • passed: test-pg job 4952, svc-realcombo job 4953
  • first test failure (job 4951): unrelated Backend TestRewindFilesHTTPHappy exceeded its 2s terminal wait; fixed candidate has no Backend diff and the focused test passed locally 10/10
  • one permitted failed-job rerun was used (job 4965); Backend passed, but governance cleanup failed in test_long_merge_history_keeps_merged_card_visible_and_closable with OSError: Directory not empty: '.git'
  • decision: no second rerun, no metadata push, no merge, no 8788 deployment; window released under the manual queue retry budget
  • candidate remains review-complete with no remaining Critical/Important visual finding; further CI retry or an unrelated governance-test fix requires a new delivery decision
[release blocked] `workbench-visual-20261015-01` / Dev PR #556 - fixed head: `3e70537cab1cfa112d03f4b2314c5d38d2bb84c1`; run: 3237 - passed: `test-pg` job 4952, `svc-realcombo` job 4953 - first `test` failure (job 4951): unrelated Backend `TestRewindFilesHTTPHappy` exceeded its 2s terminal wait; fixed candidate has no Backend diff and the focused test passed locally 10/10 - one permitted failed-job rerun was used (job 4965); Backend passed, but governance cleanup failed in `test_long_merge_history_keeps_merged_card_visible_and_closable` with `OSError: Directory not empty: '.git'` - decision: no second rerun, no metadata push, no merge, no 8788 deployment; window released under the manual queue retry budget - candidate remains review-complete with no remaining Critical/Important visual finding; further CI retry or an unrelated governance-test fix requires a new delivery decision
weichen added 6 commits 2026-10-15 02:34:05 +00:00
Why: Long-history task tests could finish while Git background maintenance still rewrote a disposable clone, racing TemporaryDirectory cleanup and failing CI with a non-empty .git directory.
Evidence: The failing test passes 25/25 and all 344 governance tests pass locally.
Impact: Test fixture only; product, governance commands, and runtime behavior are unchanged.
Refs: Dev run 3237 jobs 4951 and 4965
Why: The review candidate needs reproducible evidence bound to the exact Dev and Frontend commits.
Evidence: SELF receipts record 272 unit tests, the production build, and 16 browser scenarios on Dev 49ef24b.
Impact: Updates only the task record; product behavior is unchanged.
Refs: feature/workbench-visual-redesign
Why: include the current integration baseline before verification.

Evidence: fetched origin/main=cc40e10f3b4b6e41cb3a7aa0826e08dd3b8b7270.

Impact: integrate upstream changes without rewriting task history.

Refs: task-sync
Why: bind the task validation checkpoint to the integration baseline.

Evidence: SELF recursive component checks; sync_base=cc40e10f3b4b6e41cb3a7aa0826e08dd3b8b7270; before_record=6c0c122bd6941bf1942be027f051565eebe7930f.

Impact: task metadata only.

Refs: .task/feature-workbench-visual-redesign.md
Why: The review candidate needs reproducible evidence bound to the exact Dev and Frontend commits.
Evidence: SELF receipts record 272 unit tests, the production build, and 16 browser scenarios on Dev 49ef24b.
Impact: Updates only the task record; product behavior is unchanged.
Refs: feature/workbench-visual-redesign
Record visual redesign verification evidence
ci / svc-realcombo (pull_request) Successful in 2m10s
ci / test (pull_request) Successful in 6m10s
ci / test-pg (pull_request) Successful in 8m28s
0fd95a1779
Why: The review candidate needs reproducible evidence bound to the exact Dev and Frontend commits.
Evidence: SELF receipts record 272 unit tests, the production build, and 16 browser scenarios on Dev 49ef24b.
Impact: Updates only the task record; product behavior is unchanged.
Refs: feature/workbench-visual-redesign
weichen merged commit 59cbf11096 into main 2026-10-15 02:44:34 +00:00
Sign in to join this conversation.
No Reviewers
No labels
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: weichen/Agent-System-Dev#556