fe789c5d52
* fix(renderer): contain corrupt lazy chunks when the recovery reload never lands
9 react-error-boundary crash reports across v1.4.171-1.4.175 (macOS, Linux,
Windows) all end the same way: a corrupt lazy chunk fails to import, recovery
requests a reload, the reload never lands, and loadLazyWithRetry re-throws the
raw SyntaxError/TypeError. RecoverableRenderErrorBoundary only suppresses
LazyChunkLoadError, so the raw error files a user-facing crash report.
LazyChunkLoadError was unreachable in production. Its precondition is a guard
written by a *different* document ('reload-landed'), but the finally block
clears that guard before the throw, so the only path that could construct it
never ran. Confirmed by the shipped bundles: 16/16 lazy_chunk_reload_vetoed
breadcrumbs carry outcome=never-landed, zero carry any other outcome, and no
bundle contains a boundary-degraded breadcrumb.
Route every exhausted-recovery path through exhaustedRecoveryFailure() so an
attempted-and-failed recovery yields a LazyChunkLoadError the boundary can
contain, and record a lazy_chunk_recovery_exhausted breadcrumb carrying the
call site, the real chunk error, and the outcome.
Deliberately unchanged: when recovery is never *attempted* (no window/SSR,
blocked sessionStorage, guard write failure) the raw error is still thrown so
normal crash reporting is unaffected. Only isKnownDynamicImportFailure matches
are contained, so module logic bugs keep reporting.
* perf(renderer): trim redundant work on the lazy-chunk failure path
Hoist the dynamic-import message patterns to module scope so classification
stops allocating seven RegExp objects per call, thread the already-computed
classification into exhaustedRecoveryFailure so the guard-not-landed path does
not re-run it, and bound recordedExhaustionKeys the way the breadcrumb and
renderer-error key stores are bounded, since error.name is library-controlled.
Failure path only; the success path is unchanged.
* refactor(renderer): remove a transposition trap on the lazy-chunk failure path
exhaustedRecoveryFailure ended in two adjacent booleans with opposite
consequences: transposing them would have returned the raw SyntaxError and
silently restored the crash this branch fixes, with no test able to catch it
(the only call site passed true for both). The isChunkFailure parameter saved
one regex scan on a path that only runs after a 10s reload wait, so drop it.
Also evict recordedExhaustionKeys oldest-first instead of clearing wholesale,
matching the breadcrumb and renderer-error key stores the comment cites, so an
overflow cannot re-open the entire set to a repeat burst.
* test(renderer): cover the exhaustion dedupe bound
The bound had no coverage, unlike the crash-breadcrumb store it mirrors, so a
refactor could drop it or invert the comparison with every test still green.
Drive 200 distinct error names through the contained path and assert the set
stays capped. Also move MAX_RECORDED_EXHAUSTION_KEYS above the comment that
describes the set, not between them.
* test(renderer): pin the exhaustion eviction policy, not just the cap
The bound test asserted only the size cap, so it stayed green under the old
wholesale clear(): after 200 distinct keys a clear-on-overflow leaves 72, which
still satisfies the cap. Replay a key that oldest-first eviction retains and
assert it emits no second breadcrumb — that fails under clear(), which would
otherwise silently re-open the whole set to a repeat burst and flush the
30-entry ring the dedupe exists to protect.
* refactor(renderer): cut the breadcrumb machinery down to the actual fix
The lazy_chunk_recovery_exhausted breadcrumb was an optional addition that paid
for itself in complexity and nothing else: it needed a dedupe set to avoid
flushing the 30-entry ring, the set needed a bound because error.name is
library-controlled, the bound needed an oldest-first eviction policy, and that
needed two more tests plus a boolean parameter that review flagged as a
transposition trap. On the dominant never-landed path it did not even fire,
because lazy_chunk_reload_vetoed already records the same reloadKey, message and
outcome.
Drop it. Observability on every path returns to the main baseline, and the fix
is what it always was: name an exhausted recovery so the boundary can contain
it. Also revert the unrelated regex hoist -- its only caller is the failure
path, so the saved allocations are noise.
* Verify ordinary errors bypass lazy chunk containment
Add test ensuring module evaluation bugs still surface despite
never-landed reload attempts. Clarify containment scope: recovery
only applies to known dynamic-import failures, not ordinary errors.
|
||
|---|---|---|
| .github | ||
| .husky | ||
| Casks | ||
| config | ||
| docs | ||
| examples/plugins | ||
| mobile | ||
| native | ||
| resources | ||
| skill-guides | ||
| skill-stubs | ||
| skills | ||
| src | ||
| tests | ||
| .gitattributes | ||
| .gitignore | ||
| .npmrc | ||
| .oxfmtrc.json | ||
| .oxlintrc.json | ||
| AGENTS.md | ||
| CLAUDE.md | ||
| LICENSE | ||
| README.md | ||
| components.json | ||
| electron.vite.config.ts | ||
| orca.yaml | ||
| package.json | ||
| pnpm-lock.yaml | ||
| pnpm-workspace.yaml | ||
| tsconfig.json | ||
| vite.web.config.ts | ||
README.md
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 0.0.37
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 7.
-
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.
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.