stablyai/orca
Orca puts multiple CLI coding agents in isolated worktrees and adds shared monitoring, review, browser context, remote execution, and mobile steering.
Orca runs **Codex, Claude Code, OpenCode, and Pi** side by side in isolated git worktrees. It adds diff review, persistent terminals, Chromium element capture, remote hosts, and **mobile monitoring and follow-ups**.
Builders juggling parallel agents can use the worktree as the isolation boundary, compare multiple implementations, and keep review and steering in one workspace. Its CLI also exposes Orca workflows back to agents.
Orca runs **Codex, Claude Code, OpenCode, and Pi** side by side in isolated git worktrees. It adds diff review, persistent terminals, Chromium element capture, remote hosts, and **mobile monitoring and follow-ups**. Builders juggling parallel agents can use the worktree as the isolation boundary, compare multiple implementations, and keep review and steering in one workspace. Its CLI also exposes Orca workflows back to agents. The material is a feature overview, with no reliability, resource-use, or security evaluation. Mobile use pairs with the desktop app, and remote execution still places repository access and terminal authority on the configured host.
Orca consolidates several emerging coding-agent UX patterns into one desktop workspace: worktree isolation, side-by-side agents, persistent terminals, visual context capture, remote execution, and mobile steering. It goes beyond terminal-only supervision or single-agent handoff, but the candidates do not resolve its open questions about reliability, resource cost, or the authority granted to remote hosts.