johannesjo/parallel-code

Parallel Code is a desktop application that runs multiple AI coding agents in parallel, each operating in its own git worktree.

The tool addresses the challenge of leveraging multiple AI coding assistants simultaneously without conflicts or sequential delays. It works by dispatching agents like Claude Code, Codex, and Gemini to work on separate tasks concurrently, with each agent's changes isolated in its own git branch and worktree. Users can then review the diffs side by side, merge successful changes, and discard unsuccessful ones. The approach eliminates the need to run agents sequentially and keeps all work organized and reviewable in a single interface.

Parallel Code suits developers who want to experiment with multiple AI coding tools on the same project without managing branches manually or paying for additional platform subscriptions. It works best for teams or individuals exploring which AI agent produces the best results for specific tasks, or for accelerating development by running multiple coding tasks in parallel. The tool supports Claude Code, Codex CLI, Gemini CLI, and Copilot CLI, letting users choose their preferred agents. It includes an AI Arena mode for head-to-head agent comparison and supports inline code Q&A with configurable LLM backends. The application runs on macOS and Linux and offers keyboard-first navigation with optional remote monitoring via QR code.

The project maintains active development with regular feature additions and refinements to its core workflows. The codebase demonstrates consistent attention to user experience through keyboard shortcuts, focused view modes, and diff review capabilities with inline comments. Development activity shows engagement with both core functionality and auxiliary features like remote monitoring and configurable LLM integrations.