* fix(file-explorer): open symlink files when stat fails
* fix(file-explorer): grant symlink targets path access on activation
Following a symlink out of the workspace was denied by the main-process
path allow-list, so both the stat and the file read failed. Activating the
row is explicit intent, so authorize the target the way terminal links and
Quick Open already do.
---------
Co-authored-by: Brennan Benson <79079362+brennanb2025@users.noreply.github.com>
* fix: wrap unbroken auto-rename failure output
* test(sidebar): cover unbroken auto-rename failure containment
happy-dom does no intrinsic sizing, so assert the two declarations that
keep an unbroken token from widening DialogContent's grid column. The
test fails when either min-w-0 or overflow-wrap:anywhere is reverted.
---------
Co-authored-by: Brennan Benson <79079362+brennanb2025@users.noreply.github.com>
* refactor(runtime): extract pure path-candidate, review-branch, and folder-workspace helpers from orca-runtime.ts
Mechanical move of three closed, pure module-scope clusters out of
orca-runtime.ts (37,608 -> 37,207 lines) into domain-named siblings:
- terminal-output-path-candidates.ts: PTY output path harvesting and the
recent-candidate history bound (3 entry points + 15 private callees).
- selected-review-branch.ts: forge-agnostic selected-review predicates and
lookup hints (GitHub/GitLab/Bitbucket/Azure DevOps/Gitea).
- runtime-folder-workspace.ts: folder-workspace id math and the repo+meta
-> Worktree projection.
Bodies are token-identical to their previous form; the only production
changes are the moves, the new import statements, and `export` keywords.
The no-control-regex suppression travels with the path-candidate scanning
that needs it. No max-lines suppression was added and the ratchet is
unchanged.
Adds characterization tests for the two clusters that had no direct
coverage; the path-candidate cluster keeps its existing tests, repointed
at the new module.
* fix flaky timer on CI
* Add daily macOS dev build release channel
Publish once-daily signed macOS builds from main at a dedicated cadence,
separate from hourly (too noisy) and release branches (too infrequent).
Builds are notarized and installable via the updater, but unvetted —
published to stablyai/orca-daily rather than the main repo to avoid
evicting stable/RC entries from the releases feed.
* fix lint
* fix commit
* Add third token mint to daily macOS build workflow
The upload step's 2x45m retry budget can outlive the one-hour token, so a third
is minted after it for verify and cleanup operations. Release notes are moved to
a file to ensure consistency between draft creation and publish. Daily channel
description updated with specific UTC release time.
* Strip liveness gate from AI Vault session delete
Delete now requires only path validation + user confirmation — no process
roster, no liveness check, no quiescence, no ownership ledger.
Co-authored-by: Orca <help@stably.ai>
* Remove obsolete AI Vault liveness delete reliability gate
Session delete no longer checks process liveness, so drop the
manifest entry that still referenced the deleted test files.
* minor fix
---------
Co-authored-by: Orca <help@stably.ai>
* fix(ui): align toggle switch handle symmetrically in on state
The switch handle (translate-x-4 / 16px) left a 6px gap on the right
in the on state while the off state had only a 2px gap on the left.
Accounting for border-box sizing (1px border each side reduces content
width to 34px), the correct on-position is translate-x-4.5 (18px) so
both sides have a 2px inset.
* docs: add JSDoc to exported functions in changed files for docstring coverage
* fix(ui): align toggle switch handle symmetrically in on state
The `h-5 w-9` switch track is `border-box` with a 1px border, leaving 34px
of content for the 14px `size-3.5` handle. `translate-x-0.5` insets the off
state by 2px, so the on state needs 34 - 14 - 2 = 18px. It used
`translate-x-4` (16px), leaving 4px on the right against 2px on the left.
Adds a boundary test so the offset cannot drift again across the 26
hand-rolled switch call sites, and drops the docstrings the original patch
added to satisfy a coverage bot (the surrounding files carry none).
Original patch by @sei0.
Co-authored-by: Orca <help@stably.ai>
* test(ui): widen the switch-handle alignment guard to every h-5 w-9 track
Scan the renderer with the readdirSync walk and TypeScript AST already used by
no-top-level-translate.test.ts instead of shelling out to `git grep`, and find
each knob inside its own track rather than by scanning source text forward.
Why: gating on `role="switch"` silently skipped ClaudeUsageLoadingState.tsx (a
loading skeleton, so a div with no role) and HiddenExperimentalGroup.tsx, so
regressing the loading-state handle back to translate-x-4 left the guard green.
The guard now also fails when the premise behind 34 - 14 - 2 = 18px stops
holding: a track that loses its 1px border or gains padding, a conditional whose
two branches no longer pair off/on, and a track whose knob moved into a child
component where the scan would otherwise just stop seeing it.
---------
Co-authored-by: Neil <4138956+nwparker@users.noreply.github.com>
Co-authored-by: Orca <help@stably.ai>
Co-authored-by: Brennan Benson <79079362+brennanb2025@users.noreply.github.com>
* fix(composer): autosize Create Worktree note on PR prefill
The Note textarea only grew on onInput, so programmatic PR title prefills
left the box at one row with overflow hidden. Resize whenever the note
value changes and allow scroll under max-height (fixes#10575).
* review: size the composer note with field-sizing instead of a measure pass
The prefill bug is the failure mode of imperative sizing: the height is only
recomputed at the events someone remembered to hook, so a programmatic setNote
(and a pane resize, and a font reflow) leaves it stale. Let the layout engine
own the height, matching NativeChatComposerField and LinearIssueTextEditor.
Drops the extracted helper and its unit test — that test asserted the two
assignments it wrote and stayed green with the bug present. The composer test
now pins the class contract and goes red on the pre-fix markup.
---------
Co-authored-by: Brennan Benson <79079362+brennanb2025@users.noreply.github.com>
* fix(editor): map .jsp/.jspf to the html language id
* docs(editor): mark the jsp html mapping as a markup-only stopgap
---------
Co-authored-by: Brennan Benson <79079362+brennanb2025@users.noreply.github.com>
* fix(browser): stop impersonating the source browser's UA on cookie import (STA-3514)
Delete the import-time UA synthesis entirely instead of patching its version
detection. The session now always keeps the engine-derived UA the registry
sets at startup (clean or native), for imported and non-imported profiles
alike.
Why deletion is the fix:
- The synthesis read CFBundleShortVersionString, which on forks is a
marketing version — Arc imports presented Chrome/1.x and sites rejected
the browser as ancient (STA-3514).
- Its stated purpose (keep Google-bound cookies valid) is handled by
isGoogleSourceBoundCookie excluding those cookies since #736, and #12884
measured that Google rejects re-transplants regardless of UA identity.
- #12608 measured the synthesized Chrome-shaped UA reaching
/v3/signin/rejected while the engine UA reached account lookup.
Persisted userAgent/userAgentByPartition meta is no longer read; legacy keys
drop off on the next meta write. The #12811 gate existed only to catch bad
synthesized values, so it leaves with the synthesis.
* test(browser): catch source UA impersonation regression
Neil
Brennan Benson
Mark XianandBrennan Benson
Doyoon LeeandBrennan Benson
Jinjing
Jinwoo HongandJinwoo-H
OrcaWinandOrcaWin
Wooseong KimandBrennan Benson
TaeHwan JungandBrennan Benson