AIToday
Large Language ModelsAI Coding AssistantsZenn AI/MLPublished: Oct 15, 2026, 22:00 JST

Orca review: Sonicmove engineer nohe finds parallel agent work seamless

Orca review: Sonicmove engineer nohe finds parallel agent work seamless

3 Key Points

  1. What happened

    nohe, an engineer at 株式会社ソニックムーブ, switched to Orca v1.4.224 about a month ago and found the move from a VS Code-centric setup felt seamless, since worktree creation, setup, and agent launch happen in one operation.

  2. Why it matters

    Orca's automated setup via orca.yaml files makes git worktree practical for everyday parallel Claude Code sessions, which nohe had avoided before because installing dependencies for each worktree was too tedious.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

nohe's switch was driven less by a feature wishlist than by an accident that kept recurring: with implementation and review running in parallel, branches under active work would sometimes get switched out from under him. About a month before writing, he moved to Orca, a tool built on the premise that agents, not humans, do the writing.

The most-cited benefit in the review is not the parallel execution itself but the removal of setup friction. nohe had known about git worktree for a long time, yet avoided it because every new worktree needs its own .env, vendor/, and node_modules/, plus cleanup afterward. With .worktreeinclude and orca.yaml, that chore disappears, and he says worktree use finally became a daily habit.

Other details shape the picture of what daily use looks like. Orca's Design Mode lets you click an element in the built-in browser and attach a comment, and in the demo the agent turned gap-0.5 into gap-3 and pushed the fix without any verbal back-and-forth. Token consumption is managed by delegating work to subagents, clearing or compacting context, and watching usage in the status bar. The rough edges he flags — Laravel logins overwriting each other across workspaces because cookies ignore ports, and rising memory use — are noted as things to work around rather than dealbreakers.

FAQ
What is Orca?
Orca is a desktop app for running multiple AI coding agents in parallel, which its makers call an ADE (Agent Development Environment). It creates an independent git worktree per task and lets you use existing CLI agents like Claude Code, Codex, and Cursor CLI. It is open source under MIT, free, and runs on macOS, Windows, and Linux.
How did nohe automate Orca's workspace setup?
By placing two files in the repository root: .worktreeinclude, which lists gitignored files to copy into each new worktree from the primary folder, and orca.yaml, whose scripts.setup lists commands to run on creation. A setting called setupAgentStartupPolicy: wait-for-setup makes agents wait until setup finishes.
What are the downsides nohe noticed after one month?
Too many features to keep up with (CLI operations, remote SSH execution, and a smartphone companion app remain untried), frequent updates that can change behavior, and memory use of about 670MB for Orca plus one terminal, rising to about 1.7GB while working.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleEiliko toy leaks 13 million conversations on Chinese servers