orca: an AI orchestrator for running multiple CLI coding agents in parallel isolated worktrees

orca: an AI orchestrator for running multiple CLI coding agents in parallel isolated worktrees

What it solves

Orca is an AI orchestrator designed for developers to run multiple AI coding agents (like Claude Code, Codex, or OpenCode) side-by-side. It eliminates the friction of switching between different agents and managing their separate environments by providing a unified workspace for parallel experimentation and coordination.

How it works

Orca acts as a management layer for any CLI-based AI agent. It organizes agents into isolated git worktrees, allowing users to send a single prompt to multiple agents simultaneously to compare their outputs and merge the best result. The system includes a desktop application with integrated terminals, a file editor, and a mobile companion app for remote monitoring and steering.

Who it’s for

It is built for "100x builders" and developers who use multiple AI coding assistants to accelerate their software development workflow.

Highlights

  • Parallel Worktrees: Run multiple agents in isolated git worktrees to compare and merge results.
  • Mobile Companion: Monitor agent progress and send follow-up prompts via iOS and Android apps.
  • Design Mode: Capture HTML, CSS, and screenshots from a Chromium window directly into an agent's prompt.
  • SSH Worktrees: Execute agents on remote servers with full file editing and port forwarding.
  • Native Integrations: Built-in support for browsing GitHub PRs and Linear issues.
  • AI Diff Annotation: Review and comment on AI-generated code changes directly within the app.
  • Orca CLI: A command-line interface to script workflows like creating worktrees and taking snapshots.

Sources