Claude records a prompt it absorbs mid-turn only as a queued_command
attachment whose images are inline base64 with no path. The decoder
dropped those images, so an image-only send had no transcript row to
match and its pending bubble stayed at the bottom of the chat.
Decode queued inline images as data URLs. Pathless image-only rows
cannot name their send, so each retires one delivered image-only send
after its boundary, oldest first, and pruning hands the claimed row to
the remaining sends so none reuses it. Failed or uncertain writes never
claim one. Keep inline image bytes out of handoff and worker text.
Co-authored-by: Xulin Wu <wuxulin@bytedance.com>
* feat(appearance): let app chrome follow the terminal theme
Add an "App Chrome Appearance" setting (Default / Match Terminal) next to
Left Sidebar Appearance. In match-terminal mode the terminal theme's
background/foreground are written as document-root CSS variables
(--bg-titlebar, --background/--card/--popover/..., both sidebar token
families), so the tab strip, status bar, side panels, full-page views and
popovers all follow the terminal instead of the app theme. Left-sidebar
modes still win because they scope their own variables below the root.
The terminal color pair and token derivation move into
terminal-surface-colors.ts and are shared with the sidebar resolver.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(status-bar): configurable usage format template
Add Settings → Appearance → Status Bar → "Usage format": a template that
replaces the built-in per-provider usage text when non-empty. Placeholders
cover {provider}, {plan}, the 5h/7d/fable/30d windows as percentages
(following Used/Remaining), {x.reset} countdowns, {x.resetAt} clock times
and Gemini {buckets}; "[...]" groups drop out when a placeholder inside is
empty so one template works across providers with different windows.
Whitespace is rendered as authored (whitespace-pre).
Persisted as PersistedUIState.statusBarUsageFormat with a byProvider
override map reserved for later; the RPC UI-update schema gains the
optional field.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore(appearance): document touched functions; surface usage format in search
Add one-line JSDoc to every function this branch adds or edits so the
review docstring gate passes. Include the usage-format entry in the
Status Bar section's search match so searching "template" or
"placeholder" no longer hides the parent row, and mention popovers in
the App Chrome Appearance descriptions since the setting styles them.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat(appearance): let file editors follow the terminal theme in match-terminal mode
With App Chrome Appearance = Match Terminal, the tab strip, status bar and
side panels took the terminal palette but every file pane kept the stock app
theme: Monaco (code, plain text, diffs, notebooks, markdown source) was
hard-wired to `vs`/`vs-dark`, and the markdown preview / rich editor painted
the fixed `--editor-surface` with GitHub-colored hljs tokens.
- `terminal-editor-palette.ts` derives an editor palette from the active
terminal theme (background/foreground/cursor/selection + syntax roles mapped
onto the ANSI slots) and builds a Monaco theme (`orca-terminal`) from it;
light/dark is rated by the terminal background, so a light terminal theme
under a dark OS setting still gets light styling.
- match-terminal chrome variables now include `--editor-surface` and
`--syntax-*`; the hljs stylesheets read those ahead of their defaults.
- `useEditorSurfaceAppearance` / `useMonacoEditorTheme` replace the inline
`isDark ? 'vs-dark' : 'vs'` computations; the diff-section chain passes the
Monaco theme name instead of a dark flag.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VbWbChctXeMvaR2zdbFY99
(cherry picked from commit dc85fadee0a2664adbf55a49dd3c97c9b10fc111)
* fix(status-bar): honor the usage template in compact usage mode
ProviderSegment only rendered a user template on the verbose branch, so
switching usage mode to compact silently fell back to the built-in
tightest-window label. A non-empty template now replaces the built-in
text in both modes; the mini bar stays verbose-only.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VbWbChctXeMvaR2zdbFY99
* docs(appearance): document remaining functions touched by the diff
Adds one-line JSDoc to the editor surfaces, diff viewers, palette helpers
and UI-slice entry points this branch modifies so the diff's docstring
coverage clears the pre-merge threshold.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VbWbChctXeMvaR2zdbFY99
* fix(editor): keep Monaco theme registration and preview clock out of render
The upgraded oxlint React rules flag reassigning a module variable during
render and calling Date.now() inside useMemo. Move the terminal Monaco
theme's signature cache into ensureTerminalMonacoTheme() and pin the
usage-format preview clock with a lazy useState initializer.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VbWbChctXeMvaR2zdbFY99
* localization: complete Brazilian Portuguese catalogs
* fix(i18n): localize remaining hard-coded UI copy and complete Simplified Chinese coverage
Route the English literals and shared English labels that ignored the UI
language through the existing catalog: host labels, plugin cards and consent,
CLI and SSH errors, the checks pill, browser profile labels, shortcut groups,
app and context menus, and the macOS computer-use permission window. Shared
labels are translated where they render, so src/shared stays English for
mobile, CLI and older paired hosts.
- en.json gains 314 keys bootstrapped from inline fallbacks; four duplicate
"Unassigned" keys collapse into translateUnassignedLabel()
- zh gains 337 entries and refreshes existing ones; the other nine locales get
the newly extracted keys
- joinSentences ends a CJK sentence that has no final stop with 。 before the
next one
- the desktop caller passes the computer-use helper its locale; orcad and the
CLI stay Electron-free
- three files the copy pushed past max-lines each shed a focused module
* test: expect labeled roles in the editable context menu
The edit menu now passes explicit Cut, Copy and Select All labels so the roles render in the UI language; the upstream template assertion still expected bare roles.
* fix(i18n): key translated memos on the UI language without bare void reads
Upstream's React Compiler ratchet rejects `void language;`, which the compiler drops, and treats a react-hooks lint suppression as a bailout. The two host-label hooks now pass the language through readUiLanguageMemoInput so it stays a real memo input. ContextualTourOverlay drops its language key: getContextualTour is a static lookup, and step copy is localized at render.
* fix(i18n): translate editor theme settings
* Fix reactive localization and catalog extraction
* Fix combined localization validation
* Localize nested import failures and refine catalogs
* Handle older nested import responses
* Fix localized merge prompts and quota windows
* Address localization and runtime review feedback
* Keep host search and configuration examples usable
* Preserve configured names in add-project host selector
* Preserve host label origins across views
* Use localized fallback for missing local host
* Preserve custom names and refresh localized host labels
* Refresh task host labels without repeating searches
* Preserve Portuguese mobile tab translations
* Address localization review feedback
* fix(status-bar): import window visibility interval from its new path (#40951)
---------
Co-authored-by: lisi <lisi.see@bytedance.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: marlico77 <marlonssoficial@gmail.com>
Co-authored-by: x-Ai <5061489+x-Ai@users.noreply.github.com>
Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com>
* test(ephemeral-vm): make VM runtime store suites pass on Windows
Drop the process.platform=linux stubs (they made secure writes fsync a
read-only handle, EPERM on Windows, and spawned recipes POSIX-style) and
remove temp dirs through one allSettled removeTree helper so Windows ACL
work cannot fail teardown with EBUSY.
* test: restore the rmSync import dropped by the merge from main
* fix(recipes): warn which credentials recipe scripts inherit, add opt-in passEnv allowlist
Recipe lifecycle scripts run locally with Orca's full environment. Keep that
default (provider CLIs and git auth rely on it) but name the credential-like
variables in Settings and let a recipe opt into passEnv so only listed
variables plus a safe base set reach its scripts.
* fix(recipes): fail closed on a malformed passEnv instead of dropping the recipe
Dropping the recipe let a stored snapshot or a same-id plugin recipe run with
the full environment. Keep it with an empty allowlist and a diagnostic, match
base variable names exactly outside Windows, and make the warning copy apply
to any recipe source.
* fix(recipes): import user-agent-platform from its new app-shell location
* fix(activity): keep stale monitoring panes visible
* fix(activity): prioritize stale state over old previews
* fix(activity): share the card's row state so any stale live-PTY pane stays listed
The monitoring-only fallback missed stale foreground working panes, which also
vanished from Activity while the card showed No recent update. Activity and the
card now resolve state through one resolveAgentRowState.
* fix(activity): keep stale blocked panes as attention and unverifiable rows out of running work
---------
Co-authored-by: QiongyinZhang <53255639+qioyaz@users.noreply.github.com>
New profiles default artifactSharingEnabled to true. Existing users are
unchanged: stored settings win over defaults on load, and a stored profile
saved before the key existed loads as false instead of inheriting the new
default. The gate still requires an explicit true, sign-in is still
required, and web clients still mirror the host (their local fallback no
longer inherits the desktop default). showArtifactsButton is unchanged.
* fix(settings): list the host's models in agent settings
Settings read the built-in model list, so new Codex models never appeared.
It now reads agentSession.modelCatalog when an agent's row is open.
Fixes#38643
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(settings): mark the built-in model list and keep new-chat defaults out
Say when Settings shows the built-in list because the host gave none.
Read the host's answer through the snapshot, which drops a listed default
that a workspace's own config could replace.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(settings): show Codex's account models and don't trust an old list
A Codex account answer is scoped by workspace, so reading it through the
new-chat snapshot brought back the built-in list. Use the answer directly.
A failed read now shows the marked built-in list, not an older answer.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(settings): build the older model list without a type assertion
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(settings): read the model list again when the agent command or account changes
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(settings): build the changed settings without a type assertion
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(settings): assert the shown models instead of the read count
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(settings): restore the store settings after each model list test
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(settings): refresh the model list when a host drops its lists
A paired host can change account on reconnect with no setting changing here,
so an open row kept the old list until it was reopened.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(settings): follow the moved native-chat runtime modules
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(settings): show the built-in list note only when there is a built-in list
* fix(settings): keep the Settings model list current while a host is still listing
Wait again when a host listing outlasts one wait, re-read when a paired host's launch
settings change, and stop writing the account answer into the new-chat snapshot, where
it dropped newer workspace answers.
* fix(settings): ask paired hosts to wait only after they report a running listing
Older hosts refuse waitForListing, so the first read is plain, the same order the new-chat
picker uses. A paired host's settings change re-reads without hiding the list shown.
* refactor(settings): read the host catalog without a hooks-rule suppression
---------
Co-authored-by: temp_final_v5 <>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com>
* fix(droid): keep sub-droid sessions from replacing the pane resume target
Mission workers inherit the orchestrator's pane env, so their hooks
reported their own session ids on the orchestrator's pane and the last
worker became the tab's resume target after a restart.
Fixes#39278
* fix(droid): keep the orchestrator's session after the pane row is retained
Count a retained Droid row as the pane's owner, so worker hooks that arrive after the row is dismissed or reseeded still keep the orchestrator's session, and never hand a worker's own id to a pane with no Droid owner. Moves the test next to the other provider server tests.
* fix(droid): keep the pane's Droid session when a relay forwards a worker hook without one
---------
Co-authored-by: iskic-marko <142354657+iskic-marko@users.noreply.github.com>
* Show Source Control menu rows on one line with a short status
* Keep Push and Force Push clickable; force push shows remote-only commits it replaces
* Match Force Push description to remote-only commits; list review blockers exhaustively
* Show short statuses for every blocked review action
* fix(github): title GitHub issue links that were saved without one
A workspace linked to a GitHub issue from the CLI (`--issue N`), a phone, the paired web
client or a typed number kept only the number, so search could match its issue title only
while a client happened to hold that issue. The host that owns the workspace now reads each
newly linked untitled issue once, after the write returns, the way its card would, and saves
the title on the link. It keeps the current review selection and never resurrects a removed link.
* fix(github): bound and publish issue-title fills like other metadata writes
Skip writes that leave links untouched before normalizing, check a queued link is still wanted
before reading it, publish filled titles once per repo per 2 s window, refresh the resolved rows
remote clients read, never wake a stopped WSL distro for a background read, keep a selected
review's PR suppression, and keep a failing write observer from failing a saved write.
* fix(github): keep link-title fills off unrelated writes and out of per-title refreshes
Ignore linkedWorkItem in the unchanged-links check (it is rebuilt on every write and always
titled), publish filled titles at once on a quiet host and at most once per repo per 30 s
during a burst on a timer, and only lift a PR suppression when the selected PR changes, so
restating the current PR (as a title fill does) keeps a hidden branch PR hidden.
* perf(github): publish filled link titles as addressed workspace metadata
Instead of a whole-repo worktreesChanged refresh per burst window, the host publishes one
addressed workspaceMetadata set record {hostId, epoch, seq, id, row} for the workspace whose
link was titled: to its desktop window over IPC and to paired clients as a runtime client event.
Clients patch that one row (Cmd-J reads it) and drop replayed records; older clients ignore the
new event and see the title on their next listing.
* fix(github): agreed workspaceMetadata record shape, per-id seqs and late-listing refill
Records name their kind and domain, clients keep the last applied seq per workspace for each
producer epoch (dropped when the epoch changes), and a listing that read metadata before the
title write cannot drop the title: recorded titles refill rows that still hold the same untitled
link for 60 s, never restoring a removed link. The IPC channel is marked as interim transport.
* fix(github): title-only, host-addressed application of workspaceMetadata records
Clients fill the record's issue titles into the named row's still-untitled links and never take
its links over the row's current ones, so a record that arrives after an unlink cannot restore
the issue. A paired host's records reach the row of their own execution host (an SSH twin is a
different row), and go through the bridge's connection-generation check like other host events.
* test(github): type the metadata-record test helper as AppState
* fix(github): keep recorded titles until expiry, refill only changed repos, retire with the pairing
A record that beats the listing adding its link keeps its titles for the full window; the refill
runs only for repos whose listing changed, indexes that repo's rows once and skips rows already
checked; per-workspace seqs live in the expiring entries; a re-pair or reconnect of the producing
environment retires its titles.
* fix(github): read the listing through the workspace record selector
* fix(github): keep link titles until their producer changes; fall back to a listing on eviction
Recorded issue titles no longer expire on a timer (a slow listing could outlast it): they live until
the producer's epoch, pairing or connection changes, refilling only present untitled links. When
more than the cap are in flight the oldest is evicted and its repo is listed afresh, which carries
the saved title. The refill runs in updater form so it never merges a stale store snapshot.
* fix(github): sequenced listings replace the retained-title refill
Per the re-plan ruling: the store keeps one workspaceMetadata cursor per process; a title record takes
the next seq in the same turn as its write, and each listing records the cursor when it reads metadata.
Clients note each record's seq per producer and repo; a listing at or past it ends the obligation, an
older one keeps titles the rows already show and lists the repo afresh (a new invocation, never joining
the stale one), a record whose row or link isn't listed yet asks for a fresh listing, and a seq gap
lists every repo of that producer. The retained-title table, its watcher, TTL and cap are gone.
* test(github): stamp capture, gap, restart and recovery coverage for workspace metadata listings
Also retire an environment's earlier coverage bookkeeping when it is re-paired or reconnects.
* fix(space): keep issue-title coverage per repo and host until a newer listing lands
Coverage now keeps each projection's newest record seq after its covering listing settles, so an older listing that finishes later still keeps shown titles. Projections are keyed by repo and host, paired producers keep their stream cursor, retired pairings and unlisted projections are dropped, every listing served by this host carries its metadata stamp, and a record with any unlisted titled link asks for a listing.
* fix(space): re-list after any stale listing and drop retired metadata producers
A stale listing now always asks for one fresh listing, so a link it dropped comes back and a failed scan can no longer end the wait. Removed environments drop their producer state, the producer's repo list is built only on a gap or a first-seen projection, and recovery listings skip the lineage refresh.
* perf(space): filter current rows only for listings that keep titles
* fix(space): forget removed repos on a metadata gap; keep titles for another producer run's listing
* fix(palette): search a linked issue by its saved title before its card renders
The issue's search entry took its title only from the issue cache, which fills when the card renders, and it shadowed the saved link's titled entry under the same id. It now falls back to the link's saved title.
* perf(palette): read a row's links once per palette document
* fix(palette): give each saved link its own search unit
Saved-link evidence was meant to carry an `attachment:` id but the override set an unused field, so a link shared the `issue:N` unit id with the legacy issue field and with same-number links from other repos, and only the first (often untitled) survived. The id now applies to the unit and its fields, so every saved link's title is searchable. Replaces the cache-title fallback.
* fix(space): keep the title queue draining past a failed link; encode environment ids in coverage keys
---------
The first-launch 'Trust <repo> for agents?' modal blocked agent launches
to ask a question that offered no real protection: undecided repos, and
every unattended launch, already pre-trusted. Drop the dialog, the launch
hold and the skipAgentWorkspaceTrust plumbing. Repos default to
pre-trust; the per-repo setting remains as an opt-out.
OMP's thinking menu belongs to the running model, but Orca copied it onto
every other OMP model in a new chat. Treat OMP like OpenCode: leave
non-running models' menus unknown, drop such copies saved by older
builds, and read a picked model's menu from a throwaway listing session
started on that model. Starting on it, rather than switching to it,
keeps OMP from recording the model as used in the user's account; if
OMP cannot start on it, the listing runs again without it.
A read that asks for a picked model's menu now reports the listing in
progress, so an open picker updates when the menu lands instead of on
the next open.
Fixes#39015
* fix(native-chat): show the page a browser run drove, not another chat's
* fix(browser): run each queued agent command in its own RPC call
* fix(native-chat): once a chat claims its browser activity, ignore unclaimed activity
The settled-turn fold kept only the last prose row, so a completed Codex image generation folded behind the final answer. Mark generated-image rows and keep them out of the fold, like plans and turn notices.
* fix(codex): take Codex's idle report as its live composer so handoffs reach a blank status line
* fix(codex): drop the idle verdict when Codex reports blocked, error or clear
Prints the notes the user left on a browser tab and has not sent, the same text the tab's Copy button gives (--json returns them). Reads the saved session, so it works for closed-and-restored tabs; SSH hosts reach it through the desktop browser bridge. A page drawn by a paired desktop reports that its notes stay there.
* fix(mobile): serve the phone page from a Linux install updated in place
Package managers (pacman, apt, dnf) replace Orca's files while the old
process keeps running. The host cached the manifest for the life of the
process and read assets through Electron's launch-time asar header, so
every read of the replaced files failed with asset_changed until Orca
restarted. Read from app.asar.unpacked and re-read the manifest when
manifest.json changes on disk.
Fixes STA-14385
* fix(mobile): check the page manifest off the main thread, and retry a mismatch
Read and stat the manifest with fs/promises and keep the probed root, so
bundle RPCs add no sync file calls. Keep only passing hash verdicts: a
package install writes files one at a time, so a mismatch may clear.
* fix(mobile): key the page manifest cache on its bytes
Inode, size and mtime can all repeat when ext4 rewrites a same-length
manifest within one clock tick.
* fix(mobile): read a newer in-place install's page manifest
A release written in place may add schema-v1 manifest fields the running
host predates. Read it with a stripping schema so the host serves the
fields it supports instead of reporting no bundle until restart.
* refactor(mobile): derive the installed-manifest reader from the writer's shapes
* fix(browser): keep unsent page notes across restarts and ask before closing a tab that has them
Unsent browser notes lived only in the renderer store and hydration reset them, so a reload, restart or tab close lost them. Each saved browser page now carries its unsent notes, hydration restores them, and every close path (tab strip, keyboard, CLI, phone) asks before discarding them. A page script's window.close() can't discard them.
* fix(browser): keep tabs with unsent notes open on bulk close, like pinned tabs
* feat(orchestration): a paired Orca as a Run's coordinator on the server it placed work on (inert)
The server half of remote-owned workers. A paired Orca's call names one of its
Runs; the server admits it as a peer caller kind, writes that Run at frame
receipt, and holds it to the one access check. Its workers' mail and state
queue home under a sequence no reset reuses. Nothing sends the envelope until
the coordinator side ships, so this changes nothing today.
* fix(orchestration): reserve the peer address, keep conversation kinds apart, queue a correction once with its result
* fix(orchestration): hold back a corrected Task's change whatever its latest attempt; rotate pulls in a stable order per database
* chore(orchestration): typed field read, cast rationales, a wake test on fake timers
* fix(orchestration): a peer's receipts per conversation, its sends only into its Run, a page that fits every conversation, pending adopted Tasks keep their gate
* fix(orchestration): a peer's follow-up places where its worker ran
* fix(orchestration): a peer's current/new-child start gets the --on placement refusal
* fix(editor): keep Mermaid labels readable on pastel dark theme palettes
In dark mode the Mermaid config fills nodes, actors and boxes with the
theme's --chart-3/4/5 tokens while every label uses --foreground. Themes
with pastel chart tokens (e.g. Catppuccin Mocha) put light text on light
fills, so sequence actors, class boxes and similar shapes became
unreadable (contrast ~1.06:1). The flowchart contrast pass only covers
flowcharts.
Tint each palette fill toward the background just until the foreground
reaches WCAG 4.5:1. Palettes that already read (Orca's default dark
blues) are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(editor): tint only Mermaid fills under foreground labels
Darkening the palette seeds also darkened the git graph branch boxes and
venn colours, which Mermaid pairs with its own label colours, and the
active Gantt border. Keep the seeds and tint mainBkg, clusterBkg,
requirementBackground and the Gantt task fills instead.
* fix(editor): pin Mermaid fills only when the theme colour needs tinting
An always-set mainBkg overrode a diagram's own primaryColor in every dark
theme. Pin the tinted fills only when the chart colour fails, and cover
git tag pills too.
* refactor(editor): reuse readableOn for Mermaid fill contrast
Drop the hand-rolled hex mix loop and the extra luminance exports; move
hexToRgb next to rgbaToHex so terminal appearance and Mermaid share it.
* fix(editor): draw clickable Mermaid Gantt labels in the foreground colour
Mermaid's default dark-blue clickable label reads only on light bars; on
tinted or dark bars, and beside a bar on a dark page, it disappears.
* fix(editor): tint Mermaid railroad terminal and comment fills too
---------
Co-authored-by: Pro100v <2681328+Pro100v@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(agent-status): accept OpenCode 2 hooks after OSC reports opencode
OpenCode 2's OSC app id is opencode, so the status plugin's opencode2 prompt was treated as a nested CLI and dropped. Branch rename never saw the first prompt. A different OpenCode process still does not own the pane.
* fix(agent-status): keep OpenCode 2 on rows its OSC 7501 calls opencode
OSC 7501 names every OpenCode `opencode`, so after an OpenCode 2 turn ended
its OSC idle relabeled the row and dropped the session it resumes from.
Read the OSC claim as the OpenCode 2 row's generation, and replace the
identity unit tests with one server test of the real report order.
---------
Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com>
* feat(workspaces): run setup and archive scripts for folder projects
Folder projects get the setup and archive scripts git projects have:
creating a workspace runs setup in a Setup tab, and removing one runs
archive first. The script sees the workspace's name and id, so
`orca worktree create` inside it attaches each new worktree to that
workspace, however the workspace was created.
- Folder creates (desktop IPC and runtime) prepare setup through the
shared setup prep; runtime creates start it with the worktree terminal
startup, so Setup-before-agent ordering matches.
- Runner files live in <folder>/.orca, one fresh file per launch, removed
with the workspace.
- Only the host whose catalog holds the folder reads its orca.yaml or runs
its scripts; a setup problem becomes a warning, never a failed create.
- Shared orca.yaml setup with no approval from an older client is skipped;
ask/inherit never refuses a folder create.
- No orca.yaml default tabs for folders, since every workspace shares the
directory.
- New capability worktree.folder-setup.v1 gates the Settings section on
runtime hosts; the composer, Settings and trust prompt reuse the
existing setup surfaces.
* i18n: add folder workspace hooks strings
* chore: drop the runner's sync-fs baseline and an array cast
* fix(workspaces): close review gaps in folder project setup
- Whole-project trust sends an explicit run for folder creates the
policy would run, since it carries no content hash.
- Folder setup skips file access for a skipped setup and never starts a
stopped WSL distro for a create nobody started; removal does the same.
- Runtime folder creates return the trust mismatch, so the window re-asks.
- The composer hides folder setup controls on hosts without
worktree.folder-setup.v1 and keeps their old explicit skip.
- Hook checks read a folder only on the host that owns it.
- Git archive hooks get the removed workspace's id; WSL hooks blank
inherited pane identity too.
- Runner files are written through the drive path WSL needs for its
drive mounts, and kept out of git status with .orca/.gitignore.
- The CLI's older-host warning covers --agent creates and --json output.
* fix(workspaces): wait for the folder setup capability and read orca.yaml once
Create now waits for a runtime host's folder-setup answer instead of
sending Skip while the probe is pending, and runtime setup reuses the
parsed orca.yaml when merging effective hooks.
* fix(workspaces): keep a failed folder setup probe unknown and retry it on create
* fix(workspaces): share folder setup capability answers with the composer
* fix(workspaces): re-ask the host for folder setup support on every create
* fix(settings): read orca.yaml for folder projects and use workspace copy in hook rows
QA found Settings reporting no orca.yaml for folder projects while the
composer ran that script, and the inner hook rows still said worktree.
* fix(folder-workspaces): open one Setup tab for an instant folder create
* fix(workspaces): run setup when a created workspace has only editor tabs
* perf(folder-workspaces): per-workspace runner dir, bounded setup answers, one archive config read
* fix(folder-workspaces): retain setup capability for active consumers
* fix(folder-workspaces): match hook search to workspace descriptions
* fix(i18n): drop obsolete hook search descriptions
* test(folder-workspaces): follow relocated setup trust modules
* refactor(renderer): group file and document feature owners
* refactor(renderer): remove orphaned editor recovery registration suite
* refactor(renderer): colocate file explorer tree result contract
* feat(plugins): add editor.readTypography and editor.typography.changed
Plugins that render code next to Orca's editor had to open the private
profile database and reimplement Orca's font rules to match it. Host API 1.6
adds editor.readTypography (no capability; worker, panel and editor frames):
the font family and CSS stack, size with editor zoom, weight, line height in
px, ligatures, word wrap and minimap that the file editor resolves now.
editor.typography.changed carries the same values after a settings write or
an editor zoom changes them; it is the first event that needs no capability.
The pure resolution (editor typography, editor zoom, monospace font stack)
moves to src/shared so main and the renderer use one implementation.
* fix(plugins): keep editor typography resolution from throwing in main
A hand-edited profile with a non-string font family made the shared resolver
throw, which the main-process watcher would surface at startup and on every
settings write. Font values are now type-checked, the watcher catches and
logs, and it reads live settings so a write without listeners cannot leave the
event payload stale. Over-long font families are cut to the schema bound
instead of failing every read. Regenerates the testing runtime.
* feat(dictation): offer the speech model update in the dictation pill
Moves the speech model update out of the launch notification stack and into
the dictation pill. Pressing the dictation shortcut now checks the selected
model before any mic capture; when the speech engine needs a download, the
pill offers it (notice, downloading, done, failed) instead of recording.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(dictation): address review on the setup pill check
- Drop a stale press before opening the setup pill, not only before recording.
- Swallow cancelDownload rejections.
- Don't cache a failed catalog load, and don't mark a model up to date from an empty catalog.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(dictation): cancel a pill download closed before its first progress
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(dictation): make the pre-record model check a cancellable pending start
- A toggle, Escape, stop or controller teardown during the model check now
drops the pending start, and the check rechecks current voice settings
before opening the mic.
- A press mid-download keeps the setup pill with its progress; only a model
seen ready is cached as ready.
- A cancelled or failed update no longer deletes the old model files, and the
renderer keeps offering the setup target it showed until the model is ready.
- The check is prefetched off the startup path and in-flight checks are shared,
so the first press rarely waits on main.
---------
Co-authored-by: Claude <noreply@anthropic.com>
#38730 added the reportFocusLost request inside v55, so v55 daemons from
1.4.251 reject it and an app that adopts such a daemon never replaces it.
Bump to v56 so updated apps start a daemon that is known to accept the
request, never ask pre-v55 daemons, and ask v55 daemons once per daemon
incarnation instead of once per adapter, so a replacement daemon at the
same endpoint is not treated as unsupported.
Co-authored-by: Claude <noreply@anthropic.com>
A remote chat view could sit on a stream that looked healthy but carried no
frames, so replies never rendered and sends stayed on "Sending…" until the tab
was reopened. Two paths left a stream open after delivery had stopped:
- The host dropped a chat subscriber whose frame failed to send, but never
ended its stream or logged why. It now logs the failure and ends the stream
once, so the client resubscribes from its cursor.
- The desktop discards frames from a retired transport generation, but a
rebuilt control connection advanced the generation without closing the
streams opened under the old one. Retiring a generation now closes those
streams, tells their windows (which resubscribe) and logs how many.
The client's reconnect wait now resets only after a stream has stayed up for
10 s, so a host that ends a stream right after its first frames cannot drive a
resubscribe every 750 ms.
Refs #39926
* feat(settings): group Voice settings like the other panes and fold the setup card into its rows
The setup card is gone. Its switch is the first row (Dictation), and its
messages move to the rows they are about: download progress and updates in
the model list, a missing microphone under Input. The model list is a table
with a one-line description per model, a plain quality word and equal-width
buttons aligned to the card edge.
* fix(settings): keep the delete button's hover reveal off the Button itself
* fix(settings): line up the Voice model table and make the language and repo-hint rows read plainly
* fix(settings): wait for the catalog before the Dictation status, and open the tab that shows the recommended model
* refactor(settings): share the Voice download check, order keep-loaded choices at the source, count model languages from the catalog
* perf(mobile): let newer phone shells fetch the language catalog from the desktop instead of downloading all of them
* fix(mobile): hash only the buildId fields so phone readers can recompute it
* fix(mobile): use the catalog chunk when the generation still has it
* fix(desktop): give catalog reads their own read-slot budget
* fix(review): keep the active review linked through a rebase
A detached HEAD (rebase, bisect, checkout of a commit) is no longer treated
as a branch switch. Only an attached observation can clear the active
review and push target, judged against the branch that last proved
ownership: the in-session branch before the detach, else the owning host's
rebase verdict from the listing, else unknown (never cleared). linkedItems
are never cleared.
- shared rule (hosted-review-branch-ownership) used by the renderer writer
and the runtime WorkspaceGitIdentity
- readRebaseOperation: spawn-free, fail-closed owner-side verdict; git am is
not a rebase
- listings annotate detached rows with the verdict
- explicit push-target pushes refuse while a rebase is active or unverifiable
- opt-in verifyRebaseOperation status verdict gates review AI launches at
the action boundary
* fix(review): gate mobile review actions and bound the rebase listing probe
- mobile fix-checks, resolve-comments and hosted conflict launches require a
fresh owner-side rebase verdict after agent resolution; local conflict
recovery and generic launches are unchanged
- the listing's detached-row rebase probe has a 2s budget; unread rows stay
unknown, and a checkout whose read outlived its budget gets no new one
until it settles
- an empty or malformed HEAD is unverifiable, not a proven detach
- the apply-backend real-Git test is gated on Git 2.26
* fix(mobile): let folder repo review actions launch without a Git verdict
Repo-backed folder workspaces have repo::path ids, so the mobile review-action
gate sent git.status for them and refused the launch. The gate now reuses the
repo the launch context already resolved and, like desktop's isFolderRepo
check, skips the verdict for folder repos; Git workspaces still need a fresh
'none'.
* fix(review): reconcile detached ownership per host and selection
* Fix review action rebase checks for the selected workspace
- Read fresh rebase state from the owning host without a full status scan.
- Rotate detached worktree probes so later worktrees receive checks.
- Avoid reopening the composer when a rebase check refuses launch.
* refactor(review): scope rebase fix to branch ownership
Preserve review associations through detached HEAD using the last confirmed attached branch. Keep unknown ownership conservative and retain host isolation and relink race protection.
Remove the background rebase scanner, rebase-verdict transport, and launch/push gating from this change. Keep the deferred gating work separately and replace the real-Git pipeline tests with focused metadata regressions.
Commands render before skills in the slash menu, so typing an exact skill
name could leave a fuzzy command highlighted and Enter would pick the wrong
item. After each edit or caret move the picker now highlights the row whose
name ranks best against the query. Arrow-key navigation still owns the
highlight between edits, and a catalog refresh keeps the row the user
arrowed to (by id) instead of jumping.
Co-authored-by: Neil <neil@stably.ai>
* fix(chat): collapse orchestration notices and keep the rail human-only
* test(chat): assert human-only rail previews
* test(mobile): inspect compact orchestration mail through its disclosure
* test(mobile): keep orchestration coverage in a focused suite
* test: retain the agent source type used by message cases
* fix(chat): keep rail reply previews while ticks skip agent-written inputs
Restores the hover card's reply line (the issue asks to keep the existing
preview), drops the shared test fixture and folds the authorship cases
into the rail and outline tests.
* fix(chat): stop rereading an older host's outline as the history window moves
---------
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com>
On the phone, Codex's Standard/Fast speed was a Speed row opening a
list while Claude's Fast mode was a switch. A Fast-only speed now uses
the same switch row; more speeds, or a lone Flex, keep the list. The
lone-Fast rule moves to shared code so desktop and mobile follow one
rule.
* Keep saved agent models searchable while catalogs refresh
- Load saved catalogs independently of pending refreshes.
- Preserve edited presets until a completed catalog confirms removal.
* wip
* wip
* Preserve model discovery status when a read times out
Keep catalog responses provisional while discovery continues, seed account
catalogs when late results arrive, and retire queued discovery on shutdown.
* Simplify model catalog deadlines and picker state