Commit Graph
20612 Commits
Author SHA1 Message Date
Neil a93f73c032 fix(terminal): skip an escape tail an abandoned hidden-output restore would print (#33905)
* fix(terminal): skip an escape tail an abandoned hidden-output restore would print (#33766)

Main drops a hidden pane's renderer delivery after its model ingests the
bytes, so the first chunk sent on reveal can begin inside an escape whose
start was dropped. When the restore is abandoned, the renderer resets its
parser and writes the queued chunks, and xterm printed the tail (';242m',
the rest of a Codex OSC title).

Main tracks each PTY's escape state over every accepted payload and labels
the first delivery after a hidden or pending-cap drop with the escape it
resumes inside (gapEscapeState). Only the abandon path strips that
continuation; a successful restore keeps every byte so the snapshot's
pending escape tail can complete. No label means no change.

* fix(terminal): make the escape-continuation scan's switch exhaustive (#33766)
2026-10-26 06:33:23 -07:00
Daniele SolombrinoandNeil 0ea1ac609c fix(terminal): release replay guards when a pane's xterm is disposed (#31095)
A disposed xterm silently drops its queued write callbacks, so replay
guards engaged on it never saw a parse completion. The stall probe then
certified a false wedge, logged a wedged-release breadcrumb and fired the
undeliverable-write recovery against a pane that was already gone.

Track each pending guard release per terminal and settle them with a
'target-disposed' reason from disposePane, and skip new replays into a
disposed xterm the same way certified-dead pipelines are skipped.

Co-authored-by: Neil <neil@stably.ai>
2026-10-26 06:30:47 -07:00
d290035e6e fix(cli): reject empty --terminal/--from handles instead of falling back to defaults (#16745)
An explicit empty --terminal (or orchestration --from) was treated as absent,
so `orca terminal send --terminal ""` targeted the active pane and
orchestration check/inbox/worker-start fell back to the ambient
ORCA_TERMINAL_HANDLE mailbox. Hoist the existing getPresentStringFlag into
flags.ts and use it for these handles so a present-but-empty value fails with
invalid_argument.

Fixes #16694

Co-authored-by: Neil <neil@stably.ai>
Co-authored-by: wusiyuan05 <wusiyuan05@zuoyebang.com>
2026-10-26 06:30:34 -07:00
Neilandzerone0x 4fda57bb11 feat(open-in): open SSH workspaces in Cursor and Zed (#33877)
* feat(open-in): open SSH workspaces in Cursor and Zed

Cursor shares VS Code's Remote-SSH argv and now opens SSH and SSH-reached
server workspaces the same way, in a new window like its local launch.
Zed opens them through its ssh://[user@]host[:port]/path URL, so manual
custom-port hosts work without a local SSH config alias.

Co-authored-by: zerone0x <soberzml2333@gmail.com>

* fix(open-in): keep manual Zed targets on their saved route

Always send the saved port for manual targets so a matching ssh config Port
cannot redirect them, ask for an ssh config alias when a manual host needs a
key, agent or proxy the URL cannot carry, reject option-like hosts, and accept
the zeditor CLI name.

* refactor(open-in): move the remote SSH editor launch into its own module

Keeps external-editor-launch.ts under the max-lines limit after the main merge,
and moves main's new macOS CLI fallback test onto the remote launch API.

---------

Co-authored-by: zerone0x <soberzml2333@gmail.com>
2026-10-26 06:28:29 -07:00
NeilandClaude 0cf40b1687 fix(session): show a toast when Force Reload is vetoed by a failed shutdown checkpoint (#33941)
Browser-initiated beforeunload (Force Reload, reload) had no caller to surface
the checkpoint veto, so the reload silently did nothing. Show a reload-specific
failure toast naming the cause for trusted unload events; synthetic close and
restart dispatchers keep reporting their own verdict. Paired web skips the toast
since trusted unloads there also cover tab close/navigation and the browser's
leave prompt already reports the veto.

Fixes #21017

Co-authored-by: Claude <noreply@anthropic.com>
2026-10-26 06:25:13 -07:00
Neil 9f02111538 fix(persistence): a paired server's workspaces keep their saved tabs across desktop restarts (#33926)
* fix(persistence): a paired server's workspaces keep their saved tabs across desktop restarts

Main's per-repo terminal topology revision fences window saves so a stale save
cannot undo main's own membership change. The window never sends revisions, so
on a runtime partition, which mirrors a paired server, one removal or main-side
close froze every workspace of that repo: from then on each window save had its
terminal rows rebased onto the prior rows, and every restart reopened those
workspaces on the server's per-device tab pick.

On runtime partitions the fence now lasts only through the race it guards: a
repo main changed is fenced for RUNTIME_PARTITION_TOPOLOGY_FENCE_MS, then the
window's saves win again. Local and SSH partitions keep the permanent fence.

* test(persistence): pin that a window save can drop a tab the server closed after a restart (#21236)

* fix(persistence): fence the retirements the sanitizer makes from legacy tombstones on runtime partitions
2026-10-26 06:11:15 -07:00
Neil 639d52952d feat(ssh): per-host Reconnect on launch toggle (#33935)
* feat(ssh): per-host Reconnect on launch toggle

A host with Reconnect on launch turned off starts each launch as if the user
had disconnected it: main records the session user-disconnect for it and the
managed servers reached through it, so neither the startup restore, pane
mounts nor managed-server calls dial it until the user connects it. The
renderer also skips it in the startup restore. The switch lives in Settings ->
SSH -> Advanced for hosts this desktop dials; undefined keeps today's behaviour.

* fix(ssh): hold opted-out hosts before server status probes start

Server status probes start inside registerCoreHandlers, before the SSH
handlers register, so hold launch opt-outs right before it (once per
process). Relay-era panes that hit a held host now wait for Connect instead
of reporting a failed connect.
2026-10-26 06:01:54 -07:00
NeilandTian-yi Liang c9eee6b54c fix(runtime-environments): every window sees a server saved from the CLI (#33930)
`orca environment add` writes the saved-server file directly. Main already
watched that file but only repaired the Active Server preference, so every
renderer surface listing saved servers (the Agent Session History host
picker, host menus, the status bar) kept the old list until something
else refreshed it. Main now tells every window when the file changes, and
the renderer refreshes through the existing coalesced catalog refresher.

Co-authored-by: Tian-yi Liang <tianyilt@qq.com>
2026-10-26 06:00:15 -07:00
ae8e3e1c37 fix(computer-use): send the plain chord for uppercase hotkey letters on Windows (#27642)
SendKeys presses Shift for an uppercase letter (VkKeyScan reports the
shift state), so `orca computer hotkey --key CmdOrCtrl+A` (the documented
example) sent Ctrl+Shift+A on Windows. Lower-case a single-character key
when the hotkey has modifiers; Shift is still sent when written explicitly.

Extract ConvertTo-OrcaSendKeysHotkey so runtime-render.test.ps1 can assert
the SendKeys strings case-sensitively.

Fixes #27589

Co-authored-by: jespo2021 <joeespo@gmail.com>
Co-authored-by: Neil <neil@stably.ai>
2026-10-26 05:59:08 -07:00
99631994bc fix(ssh): move through the remote folder browser with the arrow keys (#33914)
* fix(ssh): move through the remote folder browser with the arrow keys

The browser had no highlighted row and its filter box ignored Up/Down, so
the keys only scrolled the list. The first entry is now highlighted, the
arrows move it into view, and Enter opens it. A filter that leaves one
folder still starts on that folder, and a fully typed path starts on no
row so Enter still goes to that path. Also drops the duplicated
"No matches" line.

Fixes #31555

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Name the remote folder listbox and point at no row while a listing loads

While a listing loads only the spinner renders, so the highlight and
aria-activedescendant had been left on rows that were no longer there.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Keep Enter and the highlight off rows that are about to be replaced

While the committed listing loads, Enter fell back to the previous
listing's entries and could open a folder from it. While a path preview
resolves, its prior rows stay visible but are no longer targetable, so the
highlight and aria-activedescendant now clear. The filter input also gets
combobox semantics and a name.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(ssh): drop a moved highlight when a typed path previews the same listing

* fix(ssh): reset the folder highlight per keystroke and hold Enter until a typed path resolves

* fix(ssh): hold Enter on typed path input before its first preview

---------

Co-authored-by: Younghwan Chae <chyohw97@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-26 05:57:19 -07:00
NeilandMarkXian 94f562e634 fix(repos): prune each restored folder owner once its own host answers (#33863)
* fix(repos): prune a restored folder owner once its own host's catalog answers

One unreachable remote used to skip the restored-owner prune for every
workspace, so a local folder workspace stayed pinned to a dead runtime for
the whole session. Each host's applied catalog now counts on its own; a
missing workspace's owner is dropped only once that owner host hydrated.

Rebased from #11285 onto current main.

Co-authored-by: Neil <4138956+nwparker@users.noreply.github.com>

* fix(repos): only the owner host's copy or catalog settles a restored folder owner

Folder ids can repeat across hosts, so a same-id copy on another host no
longer clears a runtime's restored owner before that runtime answers; an
owner the environment list no longer names counts as settled.

* fix(repos): a superseded all-host load's environment list settles no restored owner

---------

Co-authored-by: MarkXian <mark-xian@foxmail.com>
2026-10-26 05:51:42 -07:00
Neil 4a37a6b952 test(startup): transform the ready graph at collection, not inside the first test (#33925)
Each test resets modules and re-imports main-process-ready to reset the
browser session registry's one-shot latch. Without a static import, the first
tests paid the cold transform of that graph inside the 30s test timeout and
timed out under CI load; later tests re-evaluate cached transforms in ~200ms.
2026-10-26 05:49:58 -07:00
Neil 1a77543ee0 test(layout): drop the warm-up that calls a helper #33919 removed (#33934)
#33919 and #33923 both fixed this file's cold-import timeout and crossed: #33923's beforeAll calls switchedBuild, which #33919 replaced with static imports. Main fails typecheck (TS2304). The static imports already pay the transform once.
2026-10-26 05:36:41 -07:00
hwantageandNeil 1187abdbc2 fix(windows): mark PowerShell startup command boundaries with OSC 133 C/D (#18182)
A startup command embedded in the PowerShell -EncodedCommand payload runs
before the first prompt, and the prompt wrapper suppresses 133;D until it
has printed once. An agent launched this way exited with no
command-finished boundary, so its pane was never retired.

Emit 133;C before the embedded startup command and prime the prompt latch
so the first prompt reports 133;D with the exit code. Stdin-typed startup
commands keep readline's own marks. The encoded-command budget grows by
the mark's size so real commands keep the prior limit.

Co-authored-by: Neil <4138956+nwparker@users.noreply.github.com>
2026-10-26 05:14:19 -07:00
Neil 90493a9879 perf(orcad): keep SSH hosts on the 10-minute scanner idle, give WSL hosts 3 (#33897)
* test(layout): load the runtime graph once before the layout capability tests

The first test paid the cold transform of the whole runtime graph inside its
30 s budget and timed out whenever a shard placed it without warm neighbors.

* perf(orcad): keep SSH hosts on the 10-minute scanner idle, give WSL hosts 3

Retiring the idle session scanner after a minute cost the next scan of a
large history ~14 s. SSH hosts have no memory budget to justify that, so
they return to the scanner's default; WSL hosts, which do, wait 3 minutes.
2026-10-26 05:08:34 -07:00
Neil cd1fcc5168 refactor(mobile): colocate native chat composer and drafts (#33922) 2026-10-26 05:07:28 -07:00
Neil f3059ddd6c test(layout): load the runtime graph once before the layout capability tests (#33923)
The first test paid the cold transform of the whole runtime graph inside its
30 s budget and timed out whenever a shard placed it without warm neighbors.
2026-10-26 05:06:21 -07:00
Anddy Sahir AgudeloandNeil fd25eb0872 fix(settings): open the deep-linked project setup, not the host's first (#12176)
A repo-only Settings deep link (sidebar "Project Settings") recorded just the
entry and host, so when one entry holds two setups on the same host (e.g. a
Windows and a WSL clone of one remote, or a provisioned plus a user clone) the
pane resolved the host's first setup and showed the other clone's settings.
Carry the setup id like the explicit-host path already does.

Addresses #11689.

Co-authored-by: Neil <neil@stably.ai>
2026-10-26 05:03:56 -07:00
audichuangandNeil 015996d1b1 fix(persistence): clear canonical active workspace key on worktree owner removal (#6255)
Worktree session owners are removed by their raw id, but activeWorkspaceKey
stores the canonical worktree:<id> form, so deleting a worktree left the
persisted active key pointing at the removed workspace in every partition.
Match both forms in deleteOwnerKeyedSessionFields so every owner-removal
path (worktree delete, repo pruning, invalid-layout purge) clears it.

Co-authored-by: Neil <neil@stably.ai>
2026-10-26 05:03:44 -07:00
NeilandClaude f5b2cc5a79 fix(native-chat): say why Claude refused to start a chat, including Full access as root (#33812)
* fix(native-chat): say why Claude refused to start a chat, including Full access as root

Claude Code exits before initialize when Full access (--dangerously-skip-permissions)
meets a root uid, which is common on SSH hosts. Orca reported only that Claude
stopped before it finished starting, and retried four times. Classify that refusal
as fullAccessAsRoot with a next step, and show the CLI's own startup words behind
Details on the failure row and on the unsent message.

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(native-chat): keep message Details to start failures; cover the exit path

Route the exit-error module and delivery notices to the managed e2e, explain its
timeout, and drop the exec non-null assertion. Match the zh Full access label.

Co-Authored-By: Claude <noreply@anthropic.com>

* test(native-chat): type the refused fixture so the web typecheck passes

Co-Authored-By: Claude <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-10-26 05:00:45 -07:00
Neil 5652de8872 test(layout): load the runtime and RPC registry once instead of per test (#33919)
The layout-owner tests reset modules and re-imported orca-runtime, the
dispatcher and the full RPC registry inside each test. Under CI load the
cold transform of that graph exceeds the 30s test timeout. The capability
check reads the switch at call time, so it only needs a flipped getter; the
window-owned registration check uses the default (switch-off) registry.
2026-10-26 04:57:53 -07:00
eatmoreduckandNeil 1ac1d40cd7 fix(automations): advance schedules past a DST fall-back hour (#20179)
floorToMinute recomposed local wall-clock fields via setSeconds(0, 0), and
the FREQ=HOURLY branches used setMinutes. Inside a repeated fall-back hour
both resolve to the earlier offset, so nextAutomationOccurrenceAfter could
return `after` itself and latestAutomationOccurrenceAtOrBefore could return
the first pass's instant.

Run dedupe by scheduledFor prevents duplicate launches, but the second pass
was mishandled: a `30 1 * * *` cron's second run started 30 minutes late (or
was recorded as skipped_missed past its grace window), the 01:00 EST run of
`FREQ=HOURLY;BYMINUTE=0` was dropped, and the scheduler re-evaluated and
flushed state on roughly every tick for the hour because nextRunAt stayed
in the past.

Floor with epoch arithmetic and route cron and hourly schedules through one
absolute-minute walk, so each pass of a repeated minute fires once, on time.

Fixes #20154

Co-authored-by: Neil <neil@stably.ai>
2026-10-26 04:43:13 -07:00
Neil 34f55ca5e1 refactor(task-page): colocate cross-provider composition owners (#33913) 2026-10-26 04:30:52 -07:00
eatmoreduckandNeil 15fc7a6769 docs(settings): describe the GitHub integration as the local gh CLI login, not OAuth (#26609)
Orca has no OAuth app; it reads tokens from the gh CLI on the machine running
Orca. Document that, the per-host login for Remote Orca Servers, and the
repo/read:org/project scopes auth-diagnose expects.

Refs #26604

Co-authored-by: Neil <4138956+nwparker@users.noreply.github.com>
2026-10-26 04:22:49 -07:00
Neilandbrynnclaw e50fe1e10e fix(ai-vault): delete a server-hosted session on the server that owns it (#33906)
Agent Session History now routes Delete for a session on a paired or
SSH-managed Orca server to that server through a new aiVault.deleteSession
RPC. The host revalidates the path against its own roots, removes the
transcript (into Trash on a desktop host; permanently on a plain-Node host,
which has none), and invalidates its caches. Hosts advertise
aiVault.delete-session.v1; an older host keeps Delete disabled with
"Update <host> to delete its sessions from here."

Co-authored-by: brynnclaw <261708852+brynnclaw@users.noreply.github.com>
2026-10-26 04:20:36 -07:00
67af384c4b fix(terminal): keep find-bar highlights when navigating with Cmd+G / Cmd+Shift+G (#18844)
The Cmd+G / Cmd+Shift+G path called findNext/findPrevious with only
{ caseSensitive, regex }. @xterm/addon-search only draws the active-match
decoration, fires onDidChangeResults, and keeps refreshing highlights on new
output when the call carries decorations, so the shortcut froze the active
highlight and the match counter.

Share one buildTerminalSearchOptions() between the find bar and the
shortcut path so both send the same decorated options.

Co-authored-by: Anton Kuligin <anton.kuligin@notraffic.tech>
Co-authored-by: Neil <4138956+nwparker@users.noreply.github.com>
2026-10-26 04:17:21 -07:00
Neil 2fc9ad54cf test(wsl): cover the bash bridge's argv, backpressure, fallbacks and cleanup (#33907)
Prove the relay reaches wsl.exe through --exec unchanged, carries every byte
value both ways, holds the desktop back under a stalled reader on either
side, falls back to Node for no bash, old bash, no /dev/tcp and refused
dials with unchanged sentinels, and leaves no bash or cat behind when
either side goes away.
2026-10-26 04:04:33 -07:00
Neil 5a973e29a7 fix(cli): show orphaned state in human-readable terminal list/show output (#33910)
Fixes #15313
2026-10-26 04:01:44 -07:00
Neil c5299b1169 fix(wsl-host): keep the bridge port across app restarts so the host isn't re-paired (#33109)
A WSL host's bridge listened on a fresh port each launch, which changed the saved
pairing endpoint. The renderer treats a changed WSL pairing as a different machine
and purges the host's workspaces, tabs and selection, then rebuilds them once the
host relists its projects.
2026-10-26 04:01:27 -07:00
OrcaWinandm4air bb5137969f fix(memory): show Activity Monitor's footprint in the resource panel on macOS (#33901)
The resource panel summed RSS. On macOS that counts the Electron framework's
shared pages once per helper process, so main and renderer read 2-8x too
high, while the GPU process read 3-5x too low because RSS excludes graphics
memory. The detached terminal daemon was also missing from the app total.

On macOS, sample per-process phys_footprint from top (the figure Activity
Monitor shows) at most every 10 s while the panel is open, and fall back to
RSS when top is unavailable. Count the local terminal daemon in the app's
'other' bucket; its PTY children stay attributed to their worktrees. Linux
and Windows keep RSS/working set.

Co-authored-by: m4air <m4air@Mac.localdomain>
2026-10-26 03:57:41 -07:00
17621d5368 fix(editor): match external file changes across Windows path spellings (#14812)
On drive and `\\` UNC roots, an open tab whose filePath used a different
separator or drive-letter spelling than the watcher event (for example
`c:\Repo\File.ts` vs `C:/Repo/File.ts`) never matched, so external edits
did not reload the tab or raise the changed-on-disk banner.

Both direct-match sites (the indexed batch lookup and the non-indexed
fallback in getOpenFilesForExternalFileChange) now compare through
externalChangePathKeyFor(root), which reuses editorBufferPathKey to fold
separators, the drive letter, and the UNC server only. Path-tail case stays
exact, and WSL, POSIX, SSH, and `//` roots keep byte-exact keys.

Fixes #14801

Co-authored-by: VXNCXNX <VXNCXNX@users.noreply.github.com>

Co-authored-by: Neil <neil@stably.ai>
2026-10-26 03:56:28 -07:00
Neilandoutof-place ac5c5ae4cf fix(process-table): pin macOS ps to en_US.UTF-8 so lstart and argv parse under any locale (#33898)
macOS ps formats lstart with the user's locale (%c) and vis-encodes
non-ASCII argv when no UTF-8 locale is set. Under de_DE/en_NZ/ja_JP the
process-table parsers dropped tty/startTime and folded them into command,
breaking agent recognition, pane fingerprints, agent-session start times
and daemon identity. Run those ps readers with LC_ALL=en_US.UTF-8 on
macOS; other platforms keep the inherited env.

Fixes #27004

Co-authored-by: outof-place <contact@outofplace.space>
2026-10-26 03:56:09 -07:00
Wooseong Kim ab62ae7e96 fix(diff): keep the collapsed section header in view (#24466)
Collapsing the section the View all diff viewport was inside kept the
scroll anchor's within-section offset, which pointed into the removed
body, so the structural restore landed on later sections. Pin the anchor
to the section header before the collapse is published.

Fixes #20665
2026-10-26 03:50:58 -07:00
OrcaWinandOrcaWin 819d273e27 perf(lint): one ast-grep scan feeds both the lint gate and the flaky-test ratchet (#33903)
lint:ast-grep and check-flaky-test-pattern-ratchet each parsed the whole tree.
ast-grep-gates.mjs scans once; the lint gate re-runs ast-grep only on files with
warning-or-error findings (ast-grep's own report and exit code), and the flaky
ratchet counts test-* matches from the shared scan.

Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
2026-10-26 03:50:42 -07:00
5ee6e42a3b fix(drops): keep remote drops out of git and say why drops or file opens fail (#33831)
* fix(drops): keep files dropped onto a remote workspace out of git status

Uploads into `.orca/drops` now write `.orca/drops/.gitignore` (`*`) the first
time it is missing, so dropped files no longer block a plain workspace delete
or get committed by `git add -A`. Callers share one constant for the folder.

Fixes #23856

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Re-check the import session after the drop staging ignore step

The ignore step swallows its own errors, including a stale session. When
nothing was staged the loop never asserted again, so the caller got
skipped/failed results instead of the retry error.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(drops): say why a remote drop or file open failed

Co-authored-by: Jae Heo <83424878+jae-heo@users.noreply.github.com>
Co-authored-by: Bing.Z <zzb@gxsmjx.com>

* fix(drops): keep paths out of failure toasts and reuse the too-large classifier

* fix(drops): shorten whole quoted paths in failure toasts

* fix(drops): describe upload failures by category instead of host text

* fix(drops): match failure categories on unquoted text and name manual disconnects

---------

Co-authored-by: Younghwan Chae <chyohw97@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Jae Heo <83424878+jae-heo@users.noreply.github.com>
Co-authored-by: Bing.Z <zzb@gxsmjx.com>
2026-10-26 03:48:47 -07:00
Neil dab0cd1766 refactor: colocate task page PR presentation rules (#33890) 2026-10-26 03:45:47 -07:00
Neil 2c0c6ac0c4 docs: stop promising cross-worktree split panes and Shift-Enter split in Jump Palette (#33902)
Panes are scoped to a single worktree and the Jump Palette has no
Shift-Enter handler, so replace those promises with the real flow:
watch sidebar status indicators and jump between worktrees with
Cmd-J (macOS) / Ctrl-Shift-J (Windows / Linux).

Fixes #10477
2026-10-26 03:45:32 -07:00
Neilandoutof-place c36378c2b9 fix(build): keep index.js source map aligned with the bootstrap banner (#33899)
The main bootstrap banner was prepended to index.js in generateBundle,
after Rolldown had already emitted index.js.map, so every main-process
crash trace decoded ~200 lines off. Emit it via Rolldown's output.postBanner
instead, which shifts the source map natively.

Co-authored-by: outof-place <contact@outofplace.space>
2026-10-26 03:45:15 -07:00
BingZandNeil 7e573331ee fix(shell): resolve VS Code CLI from standard app installs when code is not on PATH (#14010)
GUI-launched Orca often inherits a minimal PATH, so bare `code` fails with
spawn ENOENT for Open in VS Code, including Remote-SSH worktrees. When PATH
and install-dir lookup leave `code`/`code-insiders` unresolved, fall back to
the CLI bundled in the standard macOS app bundle (/Applications or
~/Applications) or the Windows user/system install. Linux packages already
link the CLI onto PATH, so no Linux candidates are added.

Fixes #14008

Co-authored-by: Neil <neil@stably.ai>
2026-10-26 03:39:45 -07:00
Ethereal49andNeil 7cd63e1c66 fix(notifications): use the real macOS default sound for the System notification option (#26321)
On macOS the System sound option passed sound: "default" to Electron, which
the UNUserNotificationCenter backend resolves via soundNamed:"default" (a file
lookup) instead of UNNotificationSound.defaultSound, so banners were silent.
Leave sound unset so Electron plays the OS default, and drop the now-dead
sound field from the notification options type.

Fixes #26043

Co-authored-by: Neil <neil@stably.ai>
2026-10-26 03:39:26 -07:00
faf173235b fix(telemetry): send an errno code, not the raw message, in agent_hook_install_failed (#25287)
agent_hook_install_failed sent up to 200 chars of the raw installer error,
which embeds home-directory paths (and so the OS user name) and can echo
config contents. Send only the error's errno code or 'unknown', and enforce
that in the event schema. The anchored, capped errno pattern now lives in
src/shared/errno-code.ts and is shared with the renderer launch probe and the
file-write precondition instead of their own copies.

Fixes #21492

Co-authored-by: brynnclaw <261708852+brynnclaw@users.noreply.github.com>
Co-authored-by: Neil <neil@stably.ai>
2026-10-26 03:39:17 -07:00
2sumtechandNeil 007dfff3a9 fix(browser): accept an empty --value in fill and select commands (#16030)
`orca fill --value ""` and `orca select --value ""` were rejected client-side
with "Missing required --value" even though the browser.fill/browser.select RPC
schemas already accept an empty string. Read --value with the existing
getRequiredStringFlagAllowingEmpty helper so an empty value clears a field or
picks the blank option, while an absent --value is still rejected.

Co-authored-by: Neil <4138956+nwparker@users.noreply.github.com>
2026-10-26 03:27:29 -07:00
03653ea856 fix(orchestration): keep the transcript handle open when a forward page fills (#32894)
Returning early from the for-await over the forward-page ReadStream
destroyed the stream, which closed the shared FileHandle before the
boundary checkpoint read, so any forward page that hit its limit failed
with transcript_parse_failed. Scan with positional handle.read calls
instead so the handle stays open and read errors reject inside the
caller's try/finally.

Fixes #32892

Co-authored-by: Kim Kaznowski <kim@doubledashstudios.com>
Co-authored-by: Neil <neil@stably.ai>
2026-10-26 03:27:10 -07:00
39d4c73b07 fix(automations): step schedule scans by calendar day so DST changes don't skip dates (#19086)
The RRULE day scan stepped local midnight by a fixed 24h. Walking back
over a 23-hour spring-forward day landed at 23:00 two dates earlier, so
the latest-occurrence lookup used for catch-up runs skipped a date. It
returned a run a day too old for daily schedules, and a week too old for
weekly ones.

Step with setDate via a shared addLocalDays helper, in both the RRULE
scan and the cron feasibility scan. Add a regression test for
America/New_York and Australia/Lord_Howe. The test is skipped on Windows,
where unstubbing TZ doesn't restore the OS zone.

Co-authored-by: liutian <liutian.6000@bytedance.com>
Co-authored-by: Neil <4138956+nwparker@users.noreply.github.com>
2026-10-26 03:27:00 -07:00
Samuel VilaandNeil bd3a6c875c fix(open-in): strip the AppImage runtime env from editors launched on Linux (#31675)
Open in started editors with Orca's full environment, so an AppImage
launch leaked ARGV0/APPDIR/APPIMAGE/OWD/APPIMAGE_LIBRARY_PATH and the
mount-path PATH and LD_LIBRARY_PATH entries into the editor's integrated
shells (zsh then runs host commands like cargo under the AppImage name).

Add envWithoutAppImageRuntime() next to removeAppImageRuntimeEnv and pass
its result to every editor spawn path on Linux; macOS and Windows keep
inheriting the env unchanged. The clipboard publisher and both AppImage
daemon command builders reuse the same helper instead of their own
copy-and-scrub loops.

Fixes #31455

Co-authored-by: Neil <neil@stably.ai>
2026-10-26 03:26:47 -07:00
Wooseong KimandNeil 6ad0a42425 fix(github): send issue bodies to gh over stdin instead of argv (#25215)
Issue create (including the empty-body fallback and follow-up PATCH),
issue body edits, and issue comments passed the body as
`--raw-field body=...` on argv. Windows caps the command line at
32,767 chars, so bodies with pasted inline images failed with
`spawn ENAMETOOLONG` before reaching GitHub.

These calls now use `gh api -X <METHOD> <endpoint> --input -` with a
JSON payload on stdin. ghExecFileWithScopeAsync also now forwards
`options.stdin` to the child process; it was previously dropped.

Fixes #15832

Co-authored-by: Neil <4138956+nwparker@users.noreply.github.com>
2026-10-26 03:26:35 -07:00
OrcaWinandOrcaWin a26006aa6a perf(ci): overlap preflight's serial tail; parallel ratchets; allow 2-10 unit shards (#33887)
- Plan unit selection starts after setup (overlaps lint) instead of last.
- Localization extraction starts with the typecheck; test-file-runners is background.
- run-ratchets.mjs runs ratchets in a pool (19s serial -> 8s locally).
- pnpm lint uses CI's single-parse check-import-boundaries.mjs.
- ORCA_UNIT_FULL_SHARD_COUNT accepts 2-10 (default stays 5).

Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
2026-10-26 03:17:34 -07:00
NeilandClaude 15a1ff75fc fix(ssh): refuse a second Orca server from a host cloned with the same machine-id (#33775)
* fix(ssh): keep machines cloned with one machine-id off each other's Orca server data (#32918)

Machines cloned from one image share /etc/machine-id. On a shared home both
got the same machine key, so both managed servers ran in ~/.orca, and each
judged the other's PID locally and reclaimed its live instance lock.

- Record the claiming hostname in the shared-home claim and machine homes;
  a Linux machine whose machine-id matches but hostname differs gets its own
  machine home. Older claims fall back to data-root owner records.
- orcad's instance lock no longer treats a holder with this machine-id but
  another hostname as local, and refuses to start when an older clone's
  profile-state owner record shows it is using the data root.

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(ssh): keep a clone's established machine home, and let a host renamed within its boot restart

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(ssh): call a machine a clone only with positive evidence

A renamed host keeps its boot and a local disk has one machine, so only
another host and boot with this machine-id on a shared filesystem counts as
a clone, on Linux only. When both clones already write the login home, stay
and let the server lock refuse the second writer. Drop the machine-home
owner markers.

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(ssh): refuse a second server from a cloned host instead of relocating data

Drop the inferred clone relocation. orcad refuses to start when a holder or
profile-state owner on a shared filesystem reports this machine-id from another
hostname and boot, naming that host and the fixes. Add ORCA_MACHINE_NAME, an
explicit host-env name that gives a managed SSH host its own machine home.

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(ssh): judge cloned machines by boot and freshness, not hostname

A peer with this machine-id is another live machine only when its record is
from another boot, was written after this boot began, and is fresh. The orcad
lock now heartbeats its mtime. That drops the hostname test, which renames
broke and same-hostname clones evaded. It also drops the participant-owner
precheck that stale records could block forever, and keeps profile-state
owner cleanup from deleting a live clone's owner. stale-lock reclaim uses the
same rule. ORCA_MACHINE_NAME now rejects invalid names and keeps a host on
the home its running server or own machine-id already uses. Hints are
role-specific and aimed only at the clone.

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(orcad): stale-lock reclaim waits out the lock freshness window

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(orcad): log a failed lock heartbeat and say a crashed clone's lock frees itself

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(orcad): align reclaim help, age evidence and clone hint with the freshness window

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(ssh): ORCA_MACHINE_NAME always selects its own home

Inferring which home to keep from running owners flipped after shutdown or
reboot and couldn't see macOS boots. A name now deterministically selects
its home, and a multi-line name is refused.

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(ssh): refuse a whitespace-only ORCA_MACHINE_NAME, and pin reclaim test clocks

Co-Authored-By: Claude <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-10-26 03:08:10 -07:00
eatmoreduck 8d18f318d7 fix(build): pin the native SwiftPM build system for the computer-use universal binary (#21146)
SwiftPM 6.4 (Xcode 27) defaults to the swiftbuild build system, which writes
products to an xcodebuild-style layout instead of .build/<triple>/release, so
the lipo step in build-computer-macos.mjs finds no inputs. Pass
--build-system native to both per-triple builds.

Fixes #21145
2026-10-26 03:05:37 -07:00
Neil 41c7167b03 refactor: colocate task page list presentation rules (#33830) 2026-10-26 03:03:56 -07:00