Repository navigation
feat: add a Home and side list to the Artifacts page on desktop - #34776
Conversation
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: the screenshots cover this revision, the new fields are additive, and Home gets its split from the sweep's own check. Notes
[DESIGN-REVIEWED] 68640b9 Description read[DESCRIPTION-READ] eced50f34db0ef577b6ba272fcede72ea3418be6b4af4f1870525c830b0e369a That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
UX Review (Opus 5.5) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing blocks: the blind reader read Home, All artifacts, the folder rows, the Doc tile, the ⋯ add menu, the star and "Recent output" the way the diff means them, and every new control appears in a screenshot. Notes
[UX-REVIEWED] 68640b9 Description read[DESCRIPTION-READ] 4deafbb827276d20358fef79a31bcb211a359c59064522887adb9ce75f1aadf0 That sha256 names everything this verdict read: the PR description, plus the 18 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
First Principles Review (Opus 5.5) — ✅ human override accepted@dodgeblaster overrode this lane for |
GPT 6.1 Review — ✅ no blocking findingsGPT 6.1 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
5ddfb81 to
8384038
Compare
|
|
|
|
|
|
Code change approver recommendation: A person should decideWhy: the decider thinks the change is probably fine but estimates a 20% chance that this change introduces a defect, above the 12% limit for approving on its own. The desktop Artifacts page gets a side list (Home / All artifacts / folder tree) and a Home view split into Recent and Recent output. The backend marks each record An automated recommendation, not a review: a maintainer still decides. It approves on its own only when the estimated chance of a defect is at most 12% and the change has at most 400 changed lines. Decider v3c (217fe607) · commit |
8384038 to
5835cbc
Compare
|
Both points were real, fixed in 5835cbc:
|
5835cbc to
19243d1
Compare
|
|
|
|
|
|
/ai-review override first-principles aeb5cb9 owner-decision: gjjnn accepted the product shape as a UI refactor of the existing Artifacts page with design and UX sign-off, no RFC needed: #34776 (comment) |
Human judgment recorded@dodgeblaster recorded a owner-decision release (class
This decision applies only to this commit. A new push requires a new judgment. |
|
self-added: yes
|
|
/ai-review override first-principles bad5aae owner-decision: gjjnn accepted the product shape as a UI refactor of the existing Artifacts page with design and UX sign-off, no RFC needed: #34776 (comment) |
Human judgment recorded@dodgeblaster recorded a owner-decision release (class
This decision applies only to this commit. A new push requires a new judgment. |
|
/ai-review override first-principles 6ef382b owner-decision: gjjnn accepted the product shape as a UI refactor of the existing Artifacts page with design and UX sign-off, no RFC needed: #34776 (comment) |
Human judgment recorded@dodgeblaster recorded a owner-decision release (class
This decision applies only to this commit. A new push requires a new judgment. |
|
Backend Tests (3.12, 6) on head 6ef382b fails in |
|
/ai-review override first-principles 9c5e9d0 owner-decision: gjjnn accepted the product shape as a UI refactor of the existing Artifacts page with design and UX sign-off, no RFC needed: #34776 (comment) |
Human judgment recorded@dodgeblaster recorded a owner-decision release (class
This decision applies only to this commit. A new push requires a new judgment. |
|
self-added: yes
|
|
self-added: yes
|
|
/ai-review override first-principles e6e5d6c owner-decision: gjjnn accepted the product shape as a UI refactor of the existing Artifacts page with design and UX sign-off, no RFC needed: #34776 (comment) |
Human judgment recorded@dodgeblaster recorded a owner-decision release (class
This decision applies only to this commit. A new push requires a new judgment. |
|
self-added: yes
|
|
|
/ai-review override first-principles f0da114 owner-decision: gjjnn accepted the product shape as a UI refactor of the existing Artifacts page with design and UX sign-off, no RFC needed: #34776 (comment) |
Human judgment recorded@dodgeblaster recorded a owner-decision release (class
This decision applies only to this commit. A new push requires a new judgment. |
|
/ai-review override first-principles 68640b9 owner-decision: gjjnn accepted the product shape as a UI refactor of the existing Artifacts page with design and UX sign-off, no RFC needed: #34776 (comment) |
Human judgment recorded@dodgeblaster recorded a owner-decision release (class
This decision applies only to this commit. A new push requires a new judgment. |
KiroCrewCaptain
left a comment
There was a problem hiding this comment.
Thanks for this. One thing blocks merge on this head: the UX evidence does not match what the code ships.
- Missing evidence for the shipped copy (
website/src/i18n/locales/en.manual.json:5594, rendered inwebsite/src/pages/ArtifactsPageHome.tsx:414).
Trigger: the UX lane ran on 68640b9 and reported "cannot evaluate: missing the shipped Recent output warning". CI posted it as PASS only because the comment had no Trigger or Outcome line. Under the review rules, a verdict that cannot be reached for lack of evidence counts as a BLOCK.
Outcome: the PR body says the after shots come from this head, but the caption quotes "...10 of 200 before the oldest clear." That string is not in this head. This head ships "Unstarred output is deleted once there are more than {{cap}} ({{used}} now). Star one to keep it." No screenshot shows how this data-loss warning renders, and it is the main thing this PR adds.
Smallest fix: attach screenshots taken on 68640b9 of Home with the capped Recent output line, Home with an empty library (both empty-state lines), a card whose preview failed, and the side list with rename open, new folder open, "No folders yet", and the folder menu open. Then correct the caption and re-run the UX lane. No code change is needed for this item.
After that, this still needs a design sign-off from a maintainer, because the linked issue changes the desktop page that people already use.
Optional:
website/src/pages/chat/ActivityViewer.tsx:1045: "Browse all (N)" goes to bare/artifacts, which is now Home on desktop and not the full list. Should it point at/artifacts?scope=all?website/src/pages/ArtifactsPageHome.tsx:389: the deletion sentence also covers unsaved session documents, and the cleanup never deletes those. Consider giving them their own label without that sentence.website/src/test/ArtifactsPage.home.test.tsx:104: a synchronousgetByText('plan.md')reads the session-docs query after waiting only for the artifacts query.findByTextor awaitForwould make it deterministic, and the dedup checks at lines 387-388 have the same gap.
|
@KiroCrewCaptain thanks. Item 1 (missing evidence for the shipped copy) is addressed in the description, with no code change, on head 68640b9:
I've re-run the UX lane on this head. On design sign-off: @gjjnn will follow up on that. The optional notes (Browse all link target, the deletion sentence over unsaved session docs, findByText in the home test) are left out of this PR to keep its scope; happy to take them in a follow-up. |
|
@KiroCrewCaptain this has been signed off by the design team. |
buluoray
left a comment
There was a problem hiding this comment.
Approve on 68640b9. The desktop Artifacts Home splits kept work from output that cleanup can delete, using the server's own cleanup check, and reuses the existing sandboxed previews; targeted tests pass, the new screenshots match the shipped warning, and the design was accepted by a maintainer. Decisive lines: src/kiro_crew/dashboard/handlers/artifacts_parts/library_reads.py:285, src/kiro_crew/dashboard/handlers/artifacts_parts/library_reads.py:417, website/src/pages/ArtifactsPageHome.tsx:377-380, website/src/pages/ArtifactsPageHome.tsx:414, website/src/components/library/ArtifactThumbs.tsx:492-530, website/src/hooks/useArtifactFolders.ts:120
Problem / Motivation
Goal: On a desktop, the Artifacts page opens on a Home that separates what you made or kept from session output that cleanup will remove, with a side list for folders and cards that line up.
The page mixes deliberate work with what sessions produced as a by-product (auto-saved widgets, images, unsaved files), and nothing on screen tells them apart. A folder card previews its contents, so it reads like a document. The masonry gallery's uneven card heights make the library hard to scan, and nothing says that unstarred output is cleared once there are more than 200.
Why it matters
Most of a typical library is output nobody chose to keep. Without a visible split, a person cannot see what will survive, and the few things they made on purpose are buried among chat by-products.
Not a goal
What changed (motivation → approach → change)
A side list on desktop picks what the page shows: Home, All artifacts, or one folder from the tree. The choice rides the URL (
?scope=all,?folder=<id>, nothing for Home), and every entry accepts a dragged card.ArtifactsPageHome.tsx)ErrorNoticerather than drawing blank. The kind is a small sharedKindPill, also used in the library table.ArtifactsPageQuickCreate.tsx)useArtifactsPageCreate.ts).ArtifactsPageSideNav.tsx)library_reads.py,types/index.artifacts.ts)GET /api/artifactsmarks each recordsweepablewith the sweep's own_is_sweepable_auto_widget, and returnsoutput_cap. Home reads the split from the server, so a tag, comment, folder or edit keeps a record off Recent output exactly as it keeps it from the sweep.artifacts.newnames the Doc tile andartifacts.add-menuthe menu trigger, withartifacts.new-folderandartifacts.importunder it; a phone's New artifact button isartifacts.new-mobile, sharing the search terms, so the phone keeps a registered placement.artifacts.listmoves to one region around every view, each card carries its guide pick, and Home reports the picker available whenever the library holds a record, so "choose the artifact" guides find their target on desktop. A guide that picks an artifact opens All artifacts, and an artifact already open stays open (guide/guideActions.tsalreadyAtcounts a sub-route as already there). The index (ui-index.generated.json) and guide plans (guidePlans.gen.ts) are regenerated.website/src/i18n/locales/,en.context.json)A folder move refetches the artifact lists (
hooks/useArtifactFolders.ts), since the server re-derivessweepablefrom the folder.ArtifactMiniThumbmoves tocomponents/library/ArtifactThumbs.tsxso the folder cards and Home share it.docs/system-specs/modules/artifacts.mddocuments the layout and the two list fields, and the Artifact library row indocs/feature-map/README.mdnames the new owners and fields.Backwards compatibility
Compatible:
sweepableandoutput_capare added fields; an older client ignores them and an older gateway's list leaves Home counting with no cap.Reader: website/src/pages/ArtifactsPage.tsx:desktop split button and New folder button -- entry: crew page -- the Doc tile, the more-ways-to-add menu (New folder, Import) and the side list's + run the same handlers, and
artifacts.add-menu/artifacts.importkeep their ids;ArtifactsPage.narrowHeader.test.tsxandArtifactsPageCoverage.test.tsxcover them.Reader: website/src/pages/ArtifactsPageGallery.tsx:LibraryMasonry
artifacts.listmarker -- entry: crew page -- the marker moves to the page's list region, which encloses the masonry on a phone and Home or the card grid on desktop;ArtifactsPage.home.test.tsxpins the region around Home.Reader: website/src/pages/ArtifactsPage.tsx:desktop root folder cards and masonry -- entry: crew page -- the side list lists the folders and the card grid lists the artifacts; the phone keeps both, and the suites that pin them run on the phone layout.
Tests
ArtifactsPage.home.test.tsx(new): Home splits onsweepable; output shows "N of 200" and falls back without a cap; starring keeps an output without opening it; Table puts the groups side by side; no Starred filter on Home; the side list opens a folder and All artifacts lists every record flat; Home sits inside theartifacts.listregion and offers a pick when every record is filed (red with the old root-scoped count); a preview that cannot be read shows its error; a card folds pop-out and remove into its ⋯ menu and keeps the publication mark, description and tags; a side-list rename stays open while the folder also shows in the table (red with one shared rename state, green with the fix).ArtifactsPage.liveLink.test.tsxruns once on the phone layout and once on desktop, where it covers the card grid and Home's table rows; the desktop half fails with the live link removed from the new card.test_artifacts_list_sweepable.py(new): the list flags only unclaimed auto-registered records with the sweep's own verdict, and reports the cap.Proof: not provable -- frontend test (prove.py runs Python only)
Local:
tsc, eslint, jscpd and the restyle ratchet clean; 1,934 frontend tests (Artifacts page, UI locations, i18n, library components) and 465 find_ui tests pass.Manual verification
Ran the branch in a pod with seeded artifacts and folders: Home in Gallery and Table, a folder, All artifacts, starring an output, and New doc filing into the open folder.
Screenshots / video
Same pod and data; before is
main's frontend, after is this branch.The table above was taken on an earlier head,
eb143a694a, before the Recent output warning was reworded. Those shots still show the gallery's create menu (New folder, Import from a file) and a card's menu (Pop out, Remove from library), but their warning line is the old wording. The shots below show what this head ships.On this head (
68640b920d), real pod gateway, 1440×900 at 2×:Empty library: both Home empty states, "No folders yet", and the warning "Unstarred output is deleted once there are more than 200 (0 now). Star one to keep it."

Side list, New folder open (name field and colours).

Home with items in both groups. The warning reads "Unstarred output is deleted once there are more than 200 (4 now). Star one to keep it."

Side list, a folder's ⋯ menu open (Rename, Move to folder, Color, Delete).

Side list, folder rename open.

A card whose preview failed: "Couldn't load this preview" with "Ask the agent".

How the data was made: the sample artifacts were created through the pod's
POST /api/artifacts; four of them were then marked as session output in their stored record so Recent output has items. For shot 06, the browser fulfilled that one artifact's detail request with a 500 so the error tile renders; every other request went to the pod.Related Issues
Closes #34774
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)