Jinwoo Hong 3bb9a4e261 fix(mobile): keep a painted frame under the page until its first paint (#22264)
* fix(mobile): keep a painted frame under the page until its first paint

The shell tore its own frame down the moment a generation was on screen
(`MobileWebShellScreen.tsx`, the `ready` branch), and a mounted WebView draws
nothing until its document paints. What showed for the whole of the page's boot
was the surface behind it with nothing on it: 1.42 s on a cached generation,
against a one-frame budget.

The page is the only thing that knows when it has a frame, so it says so. It
declares `painted` in `ready.reports` and posts the notify after the browser has
painted its first commit; the shell holds the same neutral frame it was already
painting while it opened the generation, then fades it out. The wait is bounded
by the declaration and never by a timer: a generation served by an older desktop
declares nothing and is uncovered on `ready`, which is what every shell did
before this.

iOS painted white rather than nothing: a WKWebView is opaque by default, so the
shell's own surface never showed through. It is now transparent, as the Android
view already was.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* fix(mobile): hold the cover through the compositor handover

The page reports the paint its own renderer made; putting that on the app's
surface costs another frame or two. A linear fade from the report left two
frames of bare surface between the two on an emulator, which is the hole the
cover exists to close. Eased in over 220 ms, the cover keeps most of its opacity
across that handover: five reopens now show 0-21 ms of bare surface against
102-2043 ms on the build without it.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* fix(mobile): negotiate the paint report in both directions

The page posted `painted` whatever shell it met, and `notify` is a closed union:
every shell installed before this answered it with an error frame, once per
mount. The shell now advertises the name in `init.accepts` beside the param
clear and the client identity, and the page posts only when it was advertised.
The declaration in `ready.reports` stays unconditional, because it is an
optional field an older reader strips rather than a new opcode, and because the
first `ready` — the only one that matters for the first paint — is sent before
any `init` has arrived.

The accepts list moves into `bridge-init-frame.ts` beside the grants, which is
the module that builds the frame carrying it.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* test(mobile): read the cover's colour instead of asserting its shape

Two gate findings on the round-two head. The cover test reached the background
through a cast of the style prop; it now reads it through a checked narrowing,
so the test proves the shape it depends on rather than declaring it.

`use-mobile-web-shell-bridge.test.ts` stopped typechecking when the bridge args
gained `onPagePainted`: its harness is a literal, so a new required handler is a
missing property. The probe now counts paints and one case spends the counter,
which is what a handler wired only to satisfy a type would not do.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* refactor(mobile): move the cached-generation opening out of the reducer

`mobile-web-shell-session.ts` crossed `max-lines` after the merge: the refused-
update work and the paint handling both grew it. What comes out is one thing —
putting a generation already on disk on screen, and deciding whether this route
is one that bundle carries. It is the reducer's cache path and its refused-
update path both, and it was already three functions sitting together.

`step` goes into a module of its own because the two now share it; a copy in
each would be two spellings of one transition, and exporting it from either
would point the dependency the wrong way.

No behaviour moves: the reducer's table tests are unchanged and the page closure
is unchanged at 4,211, since neither new module is reachable from a page route.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* fix(mobile): drop the previous document's paint when a new one starts

A document that replaced a painted one inside the same mount inherited its
`pagePainted`, so the cover lifted before the replacement had drawn anything.
The native view already reports `loading`; the screen dropped it. It now
reaches the reducer as `document-started` and clears the page document state.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* test(mobile): make the declaration case call the frame policy

The case compared the name to itself and never called `shellPageFrame`, so it
passed for a policy that ignored the declaration entirely.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* fix(mobile): report the page's frame from the route screen, not the router

Every route screen is behind `import()`, so the wrapper above expo-router
commits with a suspense fallback while the chunk is still arriving. The paint
report hung there, which uncovered the shell's view over an empty body on a
cold chunk. It now hangs on the screen the manifest resolves, layouts excluded.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* fix(mobile): retire the readiness wait a replaced document armed

`document-started` cleared the page document state and left the flow alone, so
the previous document's readiness deadline passed the flow check, read
`pageReady` as false and failed a session whose replacement was still loading.
The flow moves with the document, for the reason `remounted` already moves it.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* fix(mobile): let a departing route screen take its paint report back

The report waits two frames, and nothing cancelled the second one, so a screen
unmounted in between still told the shell to uncover. The reporter now answers
with a take-back the wrapper returns as its cleanup, and the once-per-document
latch frees only when a report was cancelled before it landed.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* fix(mobile): let the screen that arrived take over a frame still owed

A screen committing inside the two frames an earlier one was owed found the
latch taken and reported nothing; the earlier screen then freed that latch on
its way out and nobody was left to lift the cover. The newest commit now
supersedes the pending report, and only a posted one spends the latch.

Covers the redirect window with a render check against the pr route, whose
target chunk is held open while the document sits on the hub's fallback.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb

* test(mobile): make the take-over case turn on the take-over

The case cancelled the first screen's frame through the cleanup path, so it
passed with the take-over deleted. It now leaves that screen mounted and reads
the clock: the frame after the replacement commits is the replacement's first,
not the one the screen behind it was still owed.

Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
2026-09-22 11:43:26 -04:00
2026-09-22 12:36:58 +00:00
2026-05-04 20:42:03 -07:00
2026-03-16 22:27:51 -07:00
2026-03-28 10:19:14 -07:00

Orca Orca

GitHub stars Total downloads across all releases License: MIT Join the Orca Discord Follow Orca on X Supported platforms: macOS, Windows, and Linux

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

Orca desktop app running agents in parallel worktrees, with the Orca mobile companion app in the corner

Features

Mobile Companion

Monitor and steer your agents from your phone — get notified when an agent finishes and send follow-ups from anywhere.

iOS App Store · TestFlight · Android APK 0.0.50 · Docs →

Orca desktop with the mobile companion app

Parallel Worktrees

Fan one prompt across five agents, each in its own isolated git worktree — compare the results and merge the winner.

Docs →

Parallel worktree orchestration

Terminal Splits

Ghostty-class terminals with WebGL rendering, infinite splits, and scrollback that survives restarts.

Docs →

Terminal splits

Design Mode

Click any UI element in a real Chromium window to send its HTML, CSS, and a cropped screenshot straight into your agent's prompt.

Docs →

Embedded browser and Design Mode

GitHub & Linear, Native

Browse PRs, issues, and project boards in-app — open a worktree from any task and review without a context switch.

Docs →

GitHub and Linear task workflows in Orca

SSH Worktrees

Run agents on a beefy remote box with full file editing, git, and terminals — auto-reconnect and port forwarding included.

Docs →

Remote worktrees over SSH

Annotate AI Diffs

Drop comments on any diff line and ship them back to the agent — review, edit, and commit without leaving Orca.

Docs →

Annotate AI-generated diffs

Drag Files to Agents

VS Code's editor with autosave everywhere — drag files or images straight into an agent prompt.

Docs →

Drag files and images into an agent prompt

Orca CLI

Agents drive Orca too — script every workflow with orca worktree create, snapshot, click, and fill.

Docs →

Script Orca from the CLI

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 logo Claude Code   Codex logo Codex   Grok logo Grok   Cursor logo Cursor   GitHub Copilot logo GitHub Copilot   OpenCode logo OpenCode   MiMo Code logo MiMo Code   Amp logo Amp   OpenClaude logo OpenClaude   Antigravity logo Antigravity   Pi logo Pi   oh-my-pi logo oh-my-pi   Hermes Agent logo Hermes Agent   Devin logo Devin   Goose logo Goose   Auggie logo Auggie   Autohand Code logo Autohand Code   Charm logo Charm   Cline logo Cline   Codebuff logo Codebuff   Freebuff logo Freebuff   Command Code logo Command Code   Continue logo Continue   Droid logo Droid   Kilocode logo Kilocode   Kimi logo Kimi   Kiro logo Kiro   Mistral Vibe logo Mistral Vibe   Qwen Code logo Qwen Code   Rovo Dev logo Rovo Dev   + any CLI agent


Install

Desktop — macOS, Windows, Linux

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.


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.

    WeChat group 9 QR code for the Orca community
  • 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.

Orca contributors

GitHub star history chart for stablyai/orca

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.

S
Description
Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.
Readme MIT
1.5 GiB
Languages
TypeScript 95.1%
JavaScript 4.1%
Swift 0.2%
CSS 0.2%
HCL 0.1%