On Windows, an alternate-screen TUI pane (Claude Code) could lose scroll-up
after being hidden (tab/worktree switch or app backgrounded) and shown again.
Root cause: on hidden->visible restore, applyMainBufferSnapshot
unconditionally wrote a destructive clear (\x1b[2J\x1b[3J\x1b[H) before
replaying the snapshot, wiping xterm's scrollback. The attach/reattach path
already guards this same clear behind !isAlternateScreen (pty-transport.ts);
the snapshot path did not.
Thread an alternateScreen flag from the backend headless serialization
through the pty:getMainBufferSnapshot IPC + preload types into
applyMainBufferSnapshot, and skip the destructive clear when restoring an
alternate-screen buffer. Add a regression test asserting the clear is NOT
emitted for an alt-screen restore (and the existing normal-buffer test still
asserts it IS).
Fixes#5723
* fix(automations): track exact run terminal targets
Record terminal pane keys and PTY ids on automation dispatches, persist them across run updates, and propagate them through runtime launch results.
Use the recorded pane and PTY for completion detection, session reuse, prompt submission, and View run navigation so split-pane tabs reopen the actual run terminal.
Keep legacy runs compatible by falling back to workspace resume when exact terminal identity is absent.
* chore: address automation targeting review cleanup
Co-authored-by: Orca <help@stably.ai>
---------
Co-authored-by: Jinwoo-H <jinwoo0825@gmail.com>
Co-authored-by: Orca <help@stably.ai>
* Adopt pending terminal tab reveals
* Materialize pending remote terminal tabs
Wait for mobile terminal creates to produce a ready PTY handle before resolving. If the renderer only publishes a pending tab shell, materialize a headless PTY into the same tab/leaf identity so later renderer focus adopts it instead of duplicating the tab.
Add a pane-key spawn reservation shared by runtime materialization and renderer pty:spawn so focus racing fallback reuses the winning PTY, including when SSH env stripping removes ORCA_PANE_KEY. Cover the pending-surface fallback and both overlapping spawn race directions with regression tests.
* Settle pane spawn reservation on any post-spawn failure
The per-paneKey spawn reservation was only settled on spawn error, persist
error, or final resolve. A throw in the post-spawn region (seedHeadlessTerminal,
registerPty, rememberPaneKeyForPty, track) rejected the IPC promise but left the
reservation in paneSpawnReservationsByPaneKey forever, so every later spawn for
that pane awaited a promise that never settled and hung.
Wrap each post-reservation body (both spawn paths) in a catch-all that rejects
the reservation with the real error and rethrows; reject is a no-op once
resolved. Remove the now-redundant inner reject calls. Add regression tests for
both paths.
* Tear down runtime-owned headless tabs on mobile session close
The renderer-attached branch of closeMobileSessionTab closed the renderer's
pane but never pruned mobileSessionTabsByWorktree or dropped the persisted
binding. For runtime-owned headless terminals — serve/SSH materialized, or a
pending shell the renderer never adopted, all introduced by the pending-surface
materialize fallback — syncMobileSessionTabs then re-hydrates the closed tab
from the persisted session, resurrecting it with a live PTY. "Close others" on
a paired client showed closed remote tabs reappearing with their sessions still
running.
Add isRuntimeOwnedHeadlessMobileTab (serve/SSH-owned, or no live PTY and absent
from the renderer graph). When the whole parent of such a tab is closed, route
through closeHeadlessMobileTerminalTab to kill the PTY, remove the persisted
binding, and prune+emit, then best-effort notify the renderer so no adopted
pane is left dead. Genuinely renderer-owned tabs and exact split-leaf closes
keep the existing path. Cover the serve parent close, renderer-published vs
runtime-unadopted pending tabs, and the serve split-leaf case with regressions.
* Tear down only headless tabs the renderer never published on mobile close
The close-others teardown discriminator returned false for any resolved
non-serve/ssh PTY, so headless tabs carrying a daemon session id
(<worktreeId>@@<shortUuid>) that the host materialized but the renderer never
published leaked into the session snapshot and the client mirrored them
(confirmed live: host published 5 tabs for a worktree the renderer showed 1 for).
That id shape is ALSO minted for ordinary renderer-owned daemon-backed local
terminals, so it can't classify ownership by shape — the renderer graph
(this.tabs) does. Classify a tab as runtime-owned when it is serve/ssh (always,
they're preserved + re-hydrated) or when the renderer graph never published it
(a leaked/unadopted shell); delegate everything the renderer graph lists,
including ordinary @@ local terminals and pending tabs.
Regression tests: renderer-owned @@ tab is delegated (not de-persisted); leaked
@@ tab the renderer never published is torn down + de-persisted.
* feat(agents): native Xiaomi MiMo Code support
Add mimo-code TUI agent (detect mimo, --prompt, --session resume).
Inject MIMOCODE_HOME overlay and /hook/mimo-code status plugin on mimo
launch when agent status hooks are enabled; restore via shell-ready
wrappers. Reuse OpenCode-family hook normalization in the listener.
SSH remote MiMo hook overlays are not included (local/daemon first).
Closes#6220
* fix(mimo): remirror overlay config idempotently
rmSync overlay config before mirror so a second mimo launch does not
hit EEXIST in mirrorEntry and fall back to the user MIMOCODE_HOME.
* review: harden mimo code support
Co-authored-by: Orca <help@stably.ai>
---------
Co-authored-by: Doan Bac Tam <24356000+doanbactam@users.noreply.github.com>
Co-authored-by: Jinwoo-H <jinwoo0825@gmail.com>
Co-authored-by: Orca <help@stably.ai>
- Decouple heartbeat and worker_done logic from the main coordinator
to allow re-use in the orchestration.check RPC method.
- Reconcile unread lifecycle messages during orchestration.check so that
manual coordinators can authoritatively process completions and
heartbeats even if they bypass the main coordinator loop.
- Add unit tests validating task completion, heartbeat recording, and
safety checks against stale or unauthorized worker_done payloads.
* Fix floating workspace terminal selector_not_found on a remote runtime
resolveWorktreeSelector threw selector_not_found for the floating-terminal
sentinel id (global-floating-terminal) because it is a repo-less synthetic
session with no entry in the worktree catalog. A remote client paired to this
serve sends that sentinel, so the normal id: lookup failed and the floating
pane rendered black. Resolve the sentinel (bare and id:-prefixed) to a virtual
ResolvedWorktree rooted at the serve user's home so the PTY spawns in a real
existing dir; unknown id selectors still throw selector_not_found. Adds two
regression tests.
Rebased onto current main (was 75 commits stale); net delta unchanged.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Scope floating terminal selector to terminal launches
Co-authored-by: Orca <help@stably.ai>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: Jinwoo-H <jinwoo0825@gmail.com>
Co-authored-by: Orca <help@stably.ai>
Harden mobile session-tab snapshot reconciliation and terminal creation. Rejects stale mobile snapshots, tombstones locally closed tabs until the publisher catches up, rolls back half-created terminal tabs when their surface never appears, and scopes terminal-create idempotency by worktree.
* Resolve push targets for linked hosted reviews
Automatically resolve and persist push targets for linked GitHub PRs
and GitLab MRs to align source control actions with the review branch.
- Add `ensureHostedReviewPushTarget` to hydrate missing push targets
- Implement GitLab MR push target resolution alongside GitHub PRs
- Block actions from fallback to unrelated upstreams while resolving
- Prevent duplicate concurrent API lookups with in-flight tracking
* Clear stale review push targets when unlinking or replacing reviews
- Ensure unlinking a hosted review or replacing it with another provider
properly clears the previously associated push target, preventing pushes
from being steered to outdated review heads.
- Encode explicit push target clears as null over JSON-RPC to bypass the
automatic omission of undefined properties during transit, then map them
back to undefined in the main runtime database.
- Add stricter validation to ensure pull request and merge request IDs are
positive integers, ignoring invalid or dummy numbers like zero.
* Avoid auto-reloading single-file diff tabs if their status row has
moved to a different area (e.g. from unstaged to staged).
* Prevent auto-reloading combined uncommitted diffs when they are
backed by an uncommitted entries snapshot.
* This preserves the open tab's original diff context as staging and
commit status changes, preventing useful context from being lost.
Ensure that when a visible fallback PR has been merged (e.g., outside
Orca with a deleted head branch), it is still accepted and refreshed by
branch lookup instead of being discarded as an implicit merged PR.
* Add `acceptMergedFallbackPR` option to GitHub branch lookups
* Enable this option during manual and background refreshes of fallback PRs
* Plumb the new option through preload APIs, IPC handlers, and RPC protocols
Introduce optional `task-title` and `display-name` parameters for orchestration tasks, persisting them in the database and propagating them through the RPC and Orca runtime.
This allows the CLI, dashboard, activity page, and sidebar to display concise, user-friendly labels for dispatched worker agents instead of verbose, raw system preambles.
* Add test file for workspace delete bug
Co-authored-by: Orca <help@stably.ai>
* Fix Windows workspace deletion runtime resolution
Resolve project-created workspace deletion through the selected project runtime so Windows paths are listed strictly without falsely tripping the unregistered worktree guard.
Design doc: docs/delete-workspace-windows-unregistered.md
---------
Co-authored-by: Orca <help@stably.ai>
Reveals the standard workspace delete action from sidebar rows on hover or keyboard focus, while keeping primary/main worktrees protected.
Maintainer follow-up:
- prevent hidden delete buttons from receiving pointer events
- merge latest main and resolve localization conflict
Validated with CI PR Checks plus local typecheck, lint, Vitest, and Electron hover/focus/delete-confirmation checks.
createLaunchEnv adds a team (with nested panes map) per agent-team leader,
keyed by a fresh teamId. The only eviction (removeTeamForLeaderHandle) ran
solely from closeTerminal (explicit user-close). The natural-exit paths
onPtyExit and dropDisconnectedPtyRecord tore down every other per-pty map
but never the team, so a leader exiting on its own leaked the team + panes
map permanently.
Evict via the leader handle (handleByPtyId.get(ptyId)) in both onPtyExit
and dropDisconnectedPtyRecord, mirroring the closeTerminal call. Add
getActiveTeamCount for assertability.
Regression test drives the real onPtyExit/dropDisconnectedPtyRecord: fails
before the fix (teams leak) and passes after.
Co-authored-by: Orca <help@stably.ai>
* Allow deleting downloaded speech models on desktop and mobile
Implement safe deletion of local speech models across desktop and
mobile platforms.
- Add a shared helper to coordinate STT service teardown, file deletion,
and voice settings cleanup.
- Prevent deleting speech models that are currently in use or starting.
- Update desktop and mobile settings UIs to expose the delete action.
* Track multiple pending model deletions and update action disabling
- Replace the single pending delete ID string state with a Set of IDs on desktop settings to track multiple concurrently deleting models.
- Switch the desktop delete action to use the standard themed Button component.
- Disable all row interactions on the mobile voice model list when any model action is busy.
- Add unit test coverage verifying that individual row delete button disabled states behave correctly.