Skip to content

feat(files): open a folder the tree cap still cuts in its folder tab - #21026

Merged
bolichen97 merged 1 commit into
mainfrom
fix/files-rail-search-this-folder-20632
Oct 16, 2026
Merged

bolichen97 merged 1 commit into
mainfrom
fix/files-rail-search-this-folder-20632

Conversation

@bolichen97

@bolichen97 bolichen97 commented Oct 16, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: From the Files rail, a folder the project tree's cap still cuts can be opened in one click in its folder tab, where it can be listed and searched on its own.

(Goal amended by the maintainer-side conductor after round 3: the first version added a folder-scoped search inside the rail. #20632's Task 1 is now answered "no" for that, see below.)

#20523 made every folder past the 10,000-row cap reachable from the rail. A folder that is still cut on its own listing shows a limit row, "Some items not shown". The rail's name search answers one ranked page of 50 for the whole project, so a match deep in that folder can stay out of reach. The folder tab (FolderPanel) already searches one folder by name with a page that grows, but nothing in the rail leads there.

Why it matters

Large asset or vendor folders are where the cap bites. Today the user has to know that the folder tab exists and find their way to it. One link on the limit row closes that gap without a second search surface.

Not a goal

  • A folder-scoped search inside the rail (dropped: the folder tab already does it).
  • Focusing the folder tab's search box on open. That needs a new openFolder option and tab field, so it is a possible follow-up.
  • A link for a folder that has no row in the tree at all; the rail's name search still reaches those.
  • Changing /api/file-search, /api/project/tree, FolderPanel, the cap or any path or access check.

What changed (motivation → approach → change)

#20632 Task 1 asked whether a folder-scoped name search belongs in the rail. Answer: no. FolderPanel already calls /api/file-search rooted at its folder and widens its page on demand, so a rail copy would be a second scope state for the same job.

So the rail only bridges to it. Where the host can open folder tabs, the limit row under a cut folder reads "Open in folder tab", styled as a link, and opens that folder's tab. The label promises only what the click does. The folder keeps its "some items not shown" badge, so the cut is still named. Hosts that pass no opener keep the old inert row.

Where: website/src/pierre/PierreWorkspaceTreeImpl.tsx (the row and its selection), treeStateRows.ts and config.ts (the action row's decoration and link style), tree.tsx (the onOpenFolder prop), website/src/pages/chat/FileBrowserRail.tsx (new optional onOpenFolder, given the absolute path), FilesHomePanel.tsx and SidePanel.tsx (wire it to the existing openFolder), one website/src/i18n/locales/*.json label, the capture harness frame, and the owning spec docs/system-specs/modules/learn-cron-dashboard.md (the one actionable state row, and the badge staying beside it). No new API: the rail gets one optional prop and calls the tab opener the side panel already has.

Backwards compatibility

Compatible: one new optional prop on the rail, the shared tree and FilesHomePanel; hosts that do not pass it render and behave as before. One new catalog key, row_truncated_open, in all 13 locales.

Removes nothing: the inert limit row stays for every host without an opener.

Tests

  • test/test_file_search.py::TestFolderScopedSearchReachesWhatTheProjectPageMisses: 60 louder matches push a deep file out of the project's 50-row page, and a search rooted at its folder (what the folder tab runs) finds it. This is why the row leads to the folder tab.
  • PierreWorkspaceTreeImpl.test.tsx "hands the cut folder to the host when its limit row is selected": the row reads "Open in folder tab" and selecting it calls onOpenFolder('big').
  • FileBrowserRail.test.tsx "opens a cut folder in a folder tab, as an absolute path, from its limit row" and "offers no folder-tab action when the host cannot open one".

Determinism: fixed fixtures and mocks, no timers or ordering; the rail and tree files ran 3 times with --sequence.shuffle, 217/217 each. Also ran vitest run src/i18n plus the SidePanel and FilesHomePanel tests (60 files, 1116 pass), pytest test/test_file_search.py, tsc -p tsconfig.app.json, eslint, lint:i18n, lint:theme-colors, lint:phantom-classes, lint:restyle-ratchet: all exit 0.

Manual verification

The real FileBrowserRail in a browser through website/scripts/capture-files-rail-past-tree-cap.mjs (stubbed APIs), light and dark. The folder tab itself is unchanged and not in that harness.

Screenshots / video

A folder still cut on its own listing: its limit row now opens it in a folder tab; the badge still says items are not shown.

light: limit row reads Open in folder tab

Dark theme

dark: limit row reads Open in folder tab

Related Issues

Closes #20632

Task 1 of #20632 is answered "no" for a rail-scoped search; a cut folder is now one click from the folder tab that searches it.

Part of #16819 (Task 3). Follows #20523.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable): the tree's state-row paragraph in learn-cron-dashboard.md
  • No secrets, credentials, or internal references in the diff

@bolichen97
bolichen97 requested review from a team and KiroCrewCaptain as code owners October 16, 2026 00:43
@bolichen97
bolichen97 requested a review from CrysisDeu October 16, 2026 00:43
@bolichen97
bolichen97 enabled auto-merge (squash) October 16, 2026 00:43
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Oct 16, 2026
@github-actions

github-actions Bot commented Oct 16, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 627efd1239e210d91d9646bf3e0584d4f498cc3f — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Nothing to check.

[DESIGN-REVIEWED] 627efd1

Description read

[DESCRIPTION-READ] 1f26d87d8827566aa6329403f19ef2e6e17523e0ae17f4243e3bbc13fa176ae6

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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 16, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — 🟡 CONCERNS

UX-level review of 627efd1239e210d91d9646bf3e0584d4f498cc3f — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

UX-Verdict: CONCERNS

"Open in folder tab" names a place a first-time user has never seen, so the cold reader could only guess where the click goes and whether it shows the missing items.

Watch

  • "Folder tab" is unfamiliar to a first-time user. The new main control is row_truncated_open, "Open in folder tab" (shot-01/02). The reader understood it roughly ("opens this folder by itself somewhere else, maybe so I can see all of it") but rated it "a guess". They added: "I don't know what a 'folder tab' is, where it opens, or whether it will show the missing items." So the link's whole point (you can see and search everything here) is left for the user to infer. This hits every user of a large workspace the first time they see the row. Smallest fix: name what the user gets, e.g. "Show all in its own tab".
  • The warning reads as one message shown twice. The top note "limit of 10,000 items reached, some not shown" and the per-folder badge "some items not shown" look the same to the reader: "I can't tell if 'huge-assets' is the only folder with missing items." The PR keeps the badge on purpose, but because the two read as duplicates, the link under the folder isn't clearly tied to its own cut. Low impact: it shows only on large workspaces. Fix: make the badge folder-specific ("not all items listed here").

Suggestions

  • row_truncated_open: change "Open in folder tab" to "Show all items in a new tab" (or similar), so the label promises the result instead of naming the container.

[UX-REVIEWED] 627efd1

Description read

[DESCRIPTION-READ] 9eb01b58ff856197d732c0720a5ec0ec4ed6ac5d4c28f12577be4e679a28cca8

That sha256 names everything this verdict read: the PR description, plus the 5 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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 16, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 627efd1239e210d91d9646bf3e0584d4f498cc3f — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

First-Principles-Verdict: PASS

Check that the root-level folder tab opened from the limit row (openFolder on ${projectRoot}/<folder>) lists the folder by itself and not the project tree.

What this change ships

Inventory (7 items) — 7 justified

Intent: let a user reach files deep inside a folder the tree cap still cuts by opening that folder in the folder tab that already lists and searches it. This is an ADDITION.

  1. In the pinned Files tab and the file-tab rail, the limit row under a cut folder now reads "Open in folder tab", is styled as a link, and opens that folder's tab — justified
  2. The cut folder's "items not shown" badge now stays visible beside its expanded limit row — justified
  3. Hosts that pass no opener, including FolderPanel's own tree, keep the inert "Some items not shown" row — justified
  4. A new row_truncated_open label, added in all 13 locales — justified
  5. A backend test pins that a folder-rooted /api/file-search finds a match the project's 50-row page misses — justified
  6. The capture harness passes an opener, and its frame 05-still-cut is renamed 05-open-in-folder-tab — justified
  7. The state-row paragraph in the learn-cron-dashboard.md spec now describes the actionable row — justified

[FIRST-PRINCIPLES-REVIEWED] 627efd1

Description read

[DESCRIPTION-READ] 8c18cabd01a7716dc7b69bd90148067753eede4b09e27de0e8b4e8fbeb8e53ec

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

github-actions Bot commented Oct 16, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed 627efd1239e210d91d9646bf3e0584d4f498cc3f — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 627efd1

Verdict parsed from the review's SHA-scoped output markers for commit 627efd1239e210d91d9646bf3e0584d4f498cc3f.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 627efd1239e210d91d9646bf3e0584d4f498cc3f: <one-sentence reason>

@github-actions

github-actions Bot commented Oct 16, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of 627efd1239e210d91d9646bf3e0584d4f498cc3f and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 627efd1

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 627efd1239e210d91d9646bf3e0584d4f498cc3f: <one-sentence reason>

@bolichen97
bolichen97 force-pushed the fix/files-rail-search-this-folder-20632 branch from 20b2742 to fa36adb Compare October 16, 2026 01:00
@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed in fa36adb

  • span=b7bc7b57dbdd

The scope line was 11px muted grey. It is now 12px at the normal text colour with an accent folder icon (rail-search-folder-scope in FileBrowserRail.tsx), so "Searching in huge-assets" reads as the meaning of the results, not a footnote. New capture light-05b-scoped-search.png in the body.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=9c7725717c64

The reader predicted the action correctly: the row makes the search box cover only this folder. The label is a verb phrase naming that action, the row takes the pointer cursor (STATE_ACTION_ROW in pierre/config.ts) and a dotted underline, and every other state row stays inert. No further affordance is needed for a correct read.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=feeb1140e607

The scope already sits directly under the field and, after fa36adb, at full text colour with an accent icon, so the reader is told where the search runs. Changing the placeholder would also change the field's aria-label (both read filter_placeholder), renaming the control screen-reader users and the capture harness locate, and add a fourth catalog key in 13 locales for a cue the scope line already gives.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed in fa36adb

  • span=e374ebcf936d

The scope line now renders with text-text (was text-muted) at 12px, with the folder icon in text-accent, as suggested.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed (description)

  • span=688eff7a56d7

The description now declares it: while scoped, the project-wide "searching beyond the tree" hint is hidden because the scope line replaces it and both would state where the search runs. No code change needed.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed (description)

  • span=c24e177c1800

Backwards compatibility now says three new catalog keys in all 13 locales: row_truncated_search, search_in_folder and search_whole_project. The diff was right; the line was wrong.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted (punchline: Windows mixed separators)

projectRoot is treeData.root with trailing separators stripped, and searchFolder comes from the server's own truncatedDirectories (forward-slash, project-relative). A root like C:\repo/huge is a valid Windows path: the server runs it through _resolve_search_root (realpath) before is_sensitive_path and the walk, so it answers with a normalized root. Hits are mapped back by relative(), which converts every backslash to / and tries both projectRoot and the server's data.root, the same mapping #20523's project-wide search already uses on Windows.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] Rebased: main moved in website/src/i18n/locales/*.json since this head went green (2a8c936 -> 5ddd790); CI re-running. Same head also carries the UX scope-line fix (span=e374ebcf936d).

@bolichen97
bolichen97 force-pushed the fix/files-rail-search-this-folder-20632 branch from fa36adb to 2cb12e5 Compare October 16, 2026 01:13
@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed in 2cb12e5

  • span=f3aa793cdcc2

NameSearchResults now takes scopeFolder as its only scope input. searchRoot is computed inside from projectRoot and scopeFolder, and the hint reads scopeFolder == null; the scoped and searchRoot props are gone from the component and its one caller.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed in 2cb12e5

  • span=22c42cc62505

The action row is now styled as a link: color:var(--trees-accent) with a solid underline (STATE_ACTION_ROW in pierre/config.ts), no longer the muted dotted status tone the inert state rows use. New capture dark-05-search-this-folder.png in the body.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed in 2cb12e5

  • span=089eaf996042

The scope now ends with the search: × and Escape run clearSearch (query and scope together), and switching Changed/Content resets searchFolder via a render-time mode key, so it never sits over a full tree or returns unasked. Pinned by FileBrowserRail.test.tsx "ends the folder scope when the field is cleared with Escape" and "... when the search switches to Content".

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=31c7bf609b75

The two words name two different things already on main. "Filter files…" is the field, which filters the tree's own rows; on a cut tree #20523 already sends that same field to the server search under the hint search_beyond_tree. "Searching in {{folder}}" names that server search, which is not a filter of rows. "Filtering in" would be false, and renaming the field also renames its aria-label; both are outside this PR's goal.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=6ab9198502f2

Both texts are existing #20523 copy that this PR does not change: the workspace notice and the folder badge (workspace_directory_truncated). This PR only keeps the badge on where the limit row is the search action. Rewording that shared badge would change every truncated folder in every host, which is outside this PR's goal (the folder search).

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed in 2cb12e5 (punchline: scope survives a mode switch)

Switching to Changed or Content now resets searchFolder (render-time scopeModeKey in FileBrowserRail.tsx), so returning to Name starts project-wide. Pinned by "ends the folder scope when the search switches to Content".

@bolichen97 bolichen97 changed the title feat(files): search one folder the tree cap still cuts from the Files rail feat(files): open a folder the tree cap still cuts in its folder tab Oct 16, 2026
@bolichen97
bolichen97 force-pushed the fix/files-rail-search-this-folder-20632 branch from 2cb12e5 to a824782 Compare October 16, 2026 01:32
@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed

  • span=e77dbfcfb5e2

The rail-scoped search is removed. Per the maintainer-side decision on #20632 Task 1 (answer: no), the folder tab already does folder-scoped name search with a growable page, so the rail now only links to it: the limit row reads "Open in folder tab" and calls the existing openFolder. No scope state, no scope chip, no search_in_folder/search_whole_project keys remain (a824782).

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed

  • span=1863ad46cbf0

The row no longer promises a search. It reads "Open in folder tab", which names exactly what the click does; it sits directly under the folder it acts on (a824782).

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] Rebased: main moved in website/src/i18n/locales/*.json since this head went green (5d899b0 -> 8f63e25); CI re-running. The Design and UX BLOCKs on a824782 read zero images (the body was being re-edited at that moment); the two screenshots are attached in the description and this push re-runs both lanes on it.

@bolichen97
bolichen97 force-pushed the fix/files-rail-search-this-folder-20632 branch from a824782 to c24922e Compare October 16, 2026 01:48
@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: action required A blocking check or review needs attention labels Oct 16, 2026
@bolichen97
bolichen97 force-pushed the fix/files-rail-search-this-folder-20632 branch from c24922e to 9171391 Compare October 16, 2026 01:58
@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] fixed

  • span=c3b0cae4175d

The state-row paragraph of docs/system-specs/modules/learn-cron-dashboard.md now names the one exception to "the wrapper keeps the planned row inert": with onOpenFolder, a truncatedDirectories row reads Open in folder tab, is link-styled, and on selection hands its folder to the host; and the badge stays beside it instead of yielding.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=6fb341dacbe4

The reader predicted the result correctly ("opens this folder by itself"). The label names the destination on purpose: the maintainer-side decision on #20632 asked for a label that promises only what the click does, and the destination is that. The folder tab is the same tab every folder already opens in from the rail's context menu and from chat, so it is not a new concept.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=a1be89f0729c

They are two scopes of one cause and say so: the top notice is workspace-wide ("Large workspace: limit of 10,000 items reached") and the badge names this folder. With the limit row turned into a link, the badge is the only per-folder sign of the cut, which First Principles asked to keep; removing it would hide why the folder looks short.

@bolichen97

bolichen97 commented Oct 16, 2026 •

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=5d0046baddde

"Show all items" would over-promise. A folder tab below the project root lists that folder one level at a time (api.browseFiles(cwd) in FolderPanel.tsx), so the deep files behind the cut are not shown at once. What it adds is that listing plus a name search rooted at the folder with a page that grows. "Open in folder tab" is the label the maintainer-side decision chose because it claims only what the click does.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted (punchline: keyboard and screen reader)

The row is an ordinary Pierre tree item, so it is in the tree's roving focus: arrow keys reach it and Enter selects it, and selection is the one path that runs the action (focused.deselect() then onOpenFolderRef.current(folder) in PierreWorkspaceTreeImpl.tsx), the same as a click. A screen reader announces it as a tree item with its label, "Open in folder tab", which names the action; the link styling is visual only and adds no second role.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted (punchline: the limit text)

The cut is still named twice where it applies: the workspace notice at the top keeps "limit of 10,000 items reached", and the folder's "some items not shown" badge now stays on beside the link instead of yielding to it. Only the row's own wording changed, from a status to the action that answers it.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=54bb2648221e

The label is the maintainer-side ruling on #20632 ("Label the action 'Open in folder tab' rather than 'Search', so it promises only what it does"), and the reader's own guess, "opens just that folder ... a separate tab", is exactly what happens. A folder tab is the existing panel tab every folder opens in from chat and from the rail's tree, not an OS window. Relabelling is a product-copy decision outside this PR's frozen goal; it is answered here, not changed.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=edafb640b3fd

The reason is still on screen twice: the folder's own "some items not shown" badge now stays on beside the link (it used to yield to the inert row), and the workspace notice above names the 10,000-item limit. The harness folder shows no children only because the fixture gives it none; a real cut folder lists the rows its own listing loaded above this row. Putting the reason into the link text is the relabel ruled out in span 54bb2648221e.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=485d44aa6cd7

The purple outline in shot-01 is the tree's ordinary selection ring: the capture clicks huge-assets to expand it, so it is the selected row. This PR does not touch selection styling; the link uses the tree's own --trees-accent token, as the selection ring does in this theme.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted (punchline: the opened tab lists the cut folder)

The rail hands openFolder the folder's ABSOLUTE path (${projectRoot}/${folder} in FileBrowserRail.tsx, pinned by "opens a cut folder in a folder tab, as an absolute path, from its limit row"), so FolderPanel mounts with path = that folder, not the project root; below the root it lists it with api.browseFiles(cwd), and its name search walks /api/file-search rooted at that folder, which is how deep files are reached (test_file_search.py::TestFolderScopedSearchReachesWhatTheProjectPageMisses).

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] Rebased: main moved in docs/system-specs/modules/learn-cron-dashboard.md and website/src/i18n/locales/*.json since this head went green (333061b -> e9306a9); CI re-running. Same head fixes the red Frontend Tests (3): hiStyle.test.ts '[changed-values] addresses the reader as तुम' flagged my hi value for row_truncated_open (खोलें, the आप form); it is now खोलो. All style suites pass locally with I18N_BASE_REF=origin/main.

@bolichen97
bolichen97 force-pushed the fix/files-rail-search-this-folder-20632 branch from 9171391 to 627efd1 Compare October 16, 2026 02:23
@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=3b4670c792b9

Same item as spans 6fb341dacbe4 and 54bb2648221e on earlier heads. The label is the maintainer-side ruling on #20632 ("Label the action 'Open in folder tab' ... so it promises only what it does"), and the reader's guess, "opens this folder by itself somewhere else", is what happens. Relabelling is product copy outside this PR's frozen goal; a maintainer can change it in a follow-up.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=5cab256e3495

Same item as spans a1be89f0729c and edafb640b3fd. The top notice is workspace-wide and names the 10,000 limit; the badge sits on the one folder it describes, so a folder without it is not cut. Both strings are #20523's and unchanged here; this PR only keeps the badge on beside the link, which First Principles asked for.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted

  • span=52a29c76e73d

Same suggestion as span 5d0046baddde. "Show all items" over-promises: below the project root the folder tab lists one level at a time (api.browseFiles(cwd) in FolderPanel.tsx), so deep files are not all shown at once. The label stays as the maintainer-side decision chose.

@bolichen97

Copy link
Copy Markdown
Contributor Author

[operator: bolichen97] rebutted (punchline: the opened tab lists the folder itself)

FolderPanel uses the project tree only when atProjectRoot (samePath(cwd, projectDir)); the rail passes ${projectRoot}/<folder>, which is never the root, so the tab lists that folder by itself with api.browseFiles(cwd) and its search is rooted there. Pinned on the rail side by "opens a cut folder in a folder tab, as an absolute path, from its limit row".

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: checking Automated validation is still running labels Oct 16, 2026

@KiroCrewCaptain KiroCrewCaptain left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approved by a maintainer who read the UX Review finding on this head and accepted it as follow-up work; the finding stays on the record.

@bolichen97
bolichen97 merged commit 9b99f5f into main Oct 16, 2026
103 checks passed
@bolichen97
bolichen97 deleted the fix/files-rail-search-this-folder-20632 branch October 16, 2026 07:16
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Oct 16, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Files rail: folder-scoped "Search this folder" for a folder past the tree cap

2 participants