mirror of
https://github.com/stablyai/orca.git
synced 2026-09-22 08:02:28 +00:00
Jinwoo Hong
86b93e02a7
* feat(mobile): give the microphone its own screen lock (OTA phase C, ruling 36) An open microphone holds the screen; a closed one gives it back. The lock lives in the device-side capture on both hosts — the shell's `native.audio.start|stop` handler and the native seam — so the page never decides anything about the screen. One tag per capture, minted by the module that owns the mic. Both captures give it back on every close path: a stop, a page session ending with the capture open, a device that would not begin, and an engine that throws after the capture is open, which now ends the capture rather than leaving it live. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * feat(mobile): carry the capture's tail on the stop reply (OTA phase C, ruling 36) `native.audio.stop` drains what the ring still holds into its own reply, so the page's `end()` is one verb: stop, hand the bytes on, done. The drain, await and read-once-more ordering goes with it, and so do `ending`, `reading` and `released` — three variables that existed only to order a last read against the stop and to stop a refused read re-entering `end`. The tail fields default rather than being required: the page updates over the air and the shell does not, so a page this new can meet a shell that answers `stopped` alone. That dictation loses its tail where a required field would have lost it the stop. The heap case from PR D's bot round cannot recur: `end` issues no read, and a stop reply carries no interruption, so the lane that re-entered is gone. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * refactor(mobile): drop the dictation finish id, which ordered nothing `finishingIdRef` tracked the dictation a stop was finishing, and every state it could name was already named: `cancel`, a disable, an unmount and a newer start each bump the generation or clear the active id, so the finish guard answered the same either way. Its one distinguishing arm released pending audio bytes for a dictation whose budget `closeDictationAudio` had just reset, and could subtract those bytes from a newer dictation's reserve. `acceptingChunksRef` stays: it is what stops a late microphone event being sent after the capture handed over its tail and before the finish goes out. `pendingChunksRef` stays: `stop` awaits it so the finish cannot overtake the last chunk send. The finish guard is pinned by a case that cancels while the finish is in flight; neutered, it reds. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * feat(mobile): delete the page's wake-lock seam (OTA phase C, ruling 36) The page never names the screen now. `native.wakelock.set` is gone with its schemas, its shell server, its grant rows and its harness entry; so are the page's keep-awake owner, the Android foreground re-acquire, and the `DictationKeepAwakeDevice` the capture contract carried. One module holds the screen — the device calls the microphone's capture makes — and both device-side captures share its one tag, because there is one microphone. Deleted: native-wakelock.ts (120), native-wakelock.test.ts (140), mobile-dictation-keep-awake.ts (248), mobile-dictation-keep-awake.test.ts (440), mobile-dictation-foreground-keep-awake.ts (78). With the tag pools gone, the desktop-start flow has one stale check instead of two, no startup budget to wait out and nothing to release. A source-scanning census pins it: no module under mobile/src or mobile/app but the one owner imports expo-keep-awake, and nothing anywhere names the retired verb. It reports the file and line, and checks the owner does import the package so the absence is the rule holding and not the match missing. KNOWN RED, reported and not recorded over: 25 golden cases in the speech.* families fail. The recorder adapter had to drop its keep-awake owner, which moves `adapterSha256` for every golden that mounts it, and the deleted owner's id minting shifts the deterministic random sequence, so the recorded `dictationId` values move too. No speech.dictation.* param, reply or operation changed. Awaiting the lead's call on a scoped re-record. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * test(mobile): re-record the corpus after the wake-lock deletion Baseline bumped to491eb35b4dand the corpus recorded against it. Every changed line in `goldens/` falls into six classes and nothing else: baseline 1574 lines 787 goldens lockfileSha256 1574 lines 787 goldens adapterSha256 38 lines 19 goldens scenarioSha256 8 lines 4 goldens dictationId shift 728 lines 4 goldens keep-awake effects + renumbering 63 lines 2 goldens ---- 3985 lines, which is the whole diff `recorderSha256` is untouched: no recorder module outside `adapters/` moved. The id shift is attributable arithmetic, not a behaviour change. The recording scheduler seeds `Math.random` with an LCG from seed 1; replaying it gives draw 1 `8ig2henseon` and draw 2 `dakoxjr8wun`. The deleted keep-awake owner minted its id from draw 1 during the hook's mount, so the dictation id took draw 2. With the owner gone the dictation id takes draw 1, which is why four scenario steps that pinned the literal value move with it. `adapterSha256` covers `speech.setup-sheet` as well as the three dictation families, because one adapter module hosts them all. The two goldens with vanished effects also renumber the ordinals after them, which is what the removal of an entry from a sequential counter does. `lockfileSha256` is provenance that `compareGolden` copies from the actual and never fails on. It moves in all 787 files because main's own `mobile/pnpm-lock.yaml` has moved since the corpus was last recorded; this branch does not touch it. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * fix(mobile): drop the retired wake-lock grant C7.7's route row carried The merge brought in the session route's manifest entry, which names all four dictation grants including `native.wakelock.set`. This branch deleted that verb, so the row granted a page something the shell no longer serves. Ruling 32 item 6 said C7.7 takes the dictation grant from PR D's census in this merge; this is that. Nothing caught it automatically: the shell-side grant list is derived from the verb tuple and is already three, and the closure census holds a route to the grants it needs rather than refusing ones it does not. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * test(mobile): repin the session-route closure at the measured 4,331 Measured on this merge rather than summed: 4,331 modules, 989 local. Main is red here on its own pin. `3cfb070294` measures 4,333 / 991 against a committed 4,330, three modules this branch never touched — measured in a throwaway worktree detached at that commit, with the same generators run. This merge measures 4,331 / 989, and diffing the two local lists gives the difference exactly: `mobile-dictation-keep-awake.ts` and `mobile-dictation-foreground-keep-awake.ts` leave, and nothing joins. So the branch's own effect is the -2 ruling 36 implies, and repinning to the measurement is also what takes main's closure test green again. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * test(mobile): repin the session-route closure at 4,331 on the merge with #22067 Measured on this merge: 4,331 modules, 989 local, against main's freshly repinned 4,333 / 991 at `3cfb070294`. Both provenances kept. #22067 names the three `src/shared` modules #21924 pulled into the page closure, which is what made main's earlier 4,330 stale; this branch's own -2 is the page's wake-tag owner and its Android foreground re-acquire, deleted by ruling 36. Diffing the two local lists gives exactly those two leaving and nothing joining, so the number is a reading rather than 4,333 minus an argument. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * fix(mobile): close the capture when the desktop start fails The hook opens the microphone before it asks the desktop for a session, so a refused session left the mic open — and, since the screen rides the mic, the display held until the user cancelled, retried, or the screen unmounted. The failure arm now runs the same `rollbackRecordingStart` the commit failure does, because "undo the capture this start opened" is one thing and the hook owns it; guarded like that arm, so a seam that throws on the way down cannot take the desktop cancel with it. Red-first on both hosts. Natively, a new test drives the real seam under the engine and keep-awake mocks: the refusal used to leave `initialize` with no `toggleRecording(false)` and a held screen. On the page, the mic control's own test over the port pair saw `native.audio.start` with no `native.audio.stop`. Two unit cases pin the call itself, including for a start nobody will report. Ruling 36's own words: mic closed means released. This closes the mic rather than adding a release beside it. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * test(mobile): re-record after the failed-start capture close Baseline bumped toa1bdee9af9and the corpus recorded against it. The whole diff is two classes: baseline 1574 lines 787 goldens content 74 lines 1 golden `lockfileSha256`, `adapterSha256`, `scenarioSha256` and `recorderSha256` do not move: no lockfile, adapter, scenario or recorder module changed. The one content golden is `matrix-speech.dictation-start-speech.dictation.start-1`. Its failure partitions now carry a `rollback-recording` effect at ordinal 3, which is the capture being closed, and the `speech.dictation.cancel#1` entries after it renumber from 3,4 to 3,4,4,5 — the pool holds one entry per distinct content, so a partition whose ordinal moved stops sharing an entry with the one it used to match. Removing the new effect and ignoring ordinals makes the two recordings identical, checked by dereferencing every hash rather than by reading the diff. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * fix(mobile): only the owning start rolls its capture back There is one capture seam and it carries no start identity, so round 1's rollback let a stale start's rejection end a live newer dictation: A opens the capture and waits on the desktop, the user cancels, B starts and is recording, A's request finally rejects and ends B's microphone and hands back B's screen. The rollback now runs only while this start is still the current one, which is what `wasCurrent` on the line above already reads; a stale failure still cancels its own desktop session and touches nothing else. Past the generation the capture was either already ended by whatever superseded this start, or belongs to the one that did. Red-first on both hosts, driving that exact sequence rather than a spy: the native test over the real seam saw the screen go `+ - + -`, and the page's mic-control test over the port pair saw a fourth `native.audio.` verb after B was recording. Both now end with B still holding what it took. The mirror image is covered and now pinned at host level too: A's request resolving late does not commit A over B, because the stale check after the desktop start returns through `cancelStaleStart`, which cancels A's session without touching the capture. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb * test(mobile): re-record after the stale start stopped rolling back Baseline bumped toe17b2cf603. Two classes, the whole diff: baseline 1574 lines 787 goldens content 74 lines 1 golden `lockfileSha256`, `adapterSha256`, `scenarioSha256` and `recorderSha256` do not move. The one content golden is `matrix-speech.dictation-start-speech.dictation.start-1`, whose scenario is the superseded start, so every partition in it is a stale one. The `rollback-recording` effect round 1 put there is gone, and the `speech.dictation.cancel#1` entries fold back from 4 to 2 as the ordinals after it renumber — the pool holds one entry per distinct content, so partitions whose ordinals agree again share an entry again. Dropping that effect from the superseded partitions and ignoring ordinals makes the two recordings identical, checked by dereferencing every hash. `speech-desktop-start-recording-failed` is untouched: that start still owns its capture at failure time, so it still rolls back. Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
Orca
中文 · 日本語 · 한국어 · Español · Français · Português
The AI Orchestrator for 100x builders.
Run Codex, ClaudeCode, OpenCode or Pi side-by-side — each in its own worktree, tracked in one place.
Download Orca
Features
Mobile CompanionMonitor and steer your agents from your phone — get notified when an agent finishes and send follow-ups from anywhere. |
 |
Parallel WorktreesFan one prompt across five agents, each in its own isolated git worktree — compare the results and merge the winner. |
 |
Terminal SplitsGhostty-class terminals with WebGL rendering, infinite splits, and scrollback that survives restarts. |
 |
Design ModeClick any UI element in a real Chromium window to send its HTML, CSS, and a cropped screenshot straight into your agent's prompt. |
 |
GitHub & Linear, NativeBrowse PRs, issues, and project boards in-app — open a worktree from any task and review without a context switch. |
 |
SSH WorktreesRun agents on a beefy remote box with full file editing, git, and terminals — auto-reconnect and port forwarding included. |
 |
Annotate AI DiffsDrop comments on any diff line and ship them back to the agent — review, edit, and commit without leaving Orca. |
 |
Drag Files to AgentsVS Code's editor with autosave everywhere — drag files or images straight into an agent prompt. |
 |
Orca CLIAgents drive Orca too — script every workflow with |
 |
Also in the box:
- Quick open — Search across worktrees, files, agents, commands, and repo context without leaving your flow.
- Account switcher & usage tracking — See Claude and Codex usage and rate-limit resets, and hot-swap accounts without re-logging in.
- Rich repo previews — Preview Markdown, images, PDFs, and repo docs in the workspace.
- Computer Use — Let agents operate desktop apps and visible UI when a workflow needs real interaction.
- Notifications and unread state — Know when an agent finishes or needs attention, then mark threads unread to come back later.
- And many, many more — we ship daily, so this list is perpetually behind. The changelog is the real feature list.
Supported Agents
Works with any CLI agent — if it runs in a terminal, it runs in Orca.
Claude Code
Codex
Grok
Cursor
GitHub Copilot
OpenCode
MiMo Code
Amp
OpenClaude
Antigravity
Pi
oh-my-pi
Hermes Agent
Devin
Goose
Auggie
Autohand Code
Charm
Cline
Codebuff
Command Code
Continue
Droid
Kilocode
Kimi
Kiro
Mistral Vibe
Qwen Code
Rovo Dev
+ any CLI agent
Install
Desktop — macOS, Windows, Linux
- Download from onOrca.dev
- Or grab a build directly: macOS Apple Silicon · macOS Intel · Windows (.exe) · Linux AppImage · All builds
- Running
orca serveon a headless Linux server? See the headless Linux server guide.
Or via a package manager:
# macOS (Homebrew)
brew install --cask stablyai/orca/orca
# Arch Linux (AUR) — or stably-orca-git to build from source
yay -S stably-orca-bin
Mobile Companion — iOS, Android
Pair with your desktop app to monitor and steer your agents from your phone.
- iOS: Download on the App Store or join TestFlight
- Android: Download APK 0.0.50 · Install guide
Community & Support
-
Discord: Join the community on Discord.
-
Twitter / X: Follow @orca_build for updates and announcements.
-
WeChat: Scan to join the Orca community WeChat group 9.
-
Feedback & Ideas: We ship fast. Missing something? Request a new feature.
-
Privacy: See the privacy & telemetry docs for what anonymous usage data Orca collects and how to opt out.
-
Show Support: Star this repo to follow along with our daily ships.
Developing
Want to contribute or run locally? See our CONTRIBUTING.md guide.
The relay that pairs the mobile app with a desktop host is also in this repository under
cloud/, with a separate pnpm workspace and setup guide.
Signed Builds
Windows code signing sponored/provided by SignPath.io, certificate by SignPath Foundation.
License
Orca is free and open source under the MIT License.