Parallel Code

Parallel Code is a free, MIT-licensed desktop application built by Johannes Millan that lets developers run multiple AI coding agents simultaneously, each in its own isolated git branch and worktree. It works with Claude Code, Codex CLI, Gemini CLI, Copilot CLI, and Antigravity CLI out of the box, and supports any CLI-based coding agent that runs in a terminal. The app is available for macOS and Linux and is built with Electron, SolidJS, and TypeScript.

What It Is

Parallel Code sits between your AI coding agents and your git repository, acting as a coordination layer that eliminates the shared working-copy collisions that occur when multiple agents modify the same files. Instead of running agents one at a time or manually managing worktrees in a terminal, Parallel Code automates branch creation, worktree setup, and the review-and-merge workflow through a native desktop GUI. It is editor-agnostic — you keep VS Code, Cursor, JetBrains, or Sublime; Parallel Code manages the agents and worktrees while your IDE handles the code.

How the Isolation Model Works

When you create a task, Parallel Code automatically:

• Creates a new git branch from your main branch
• Sets up a git worktree so the agent works in a separate directory
• Symlinks node_modules and other gitignored directories into the worktree
• Spawns the chosen AI agent inside that worktree

Each task is fully isolated from every other running task. Branches can still conflict if they touch the same code, but those integration conflicts are surfaced at review time rather than silently overwriting each other mid-run. When satisfied, you merge back to main with one click from the sidebar.

Workflow Patterns It Enables

The parallel isolation model unlocks workflows that are impractical with sequential agents:

• Divide the backlog — assign six tickets to six agents and let them all run in one afternoon, each in its own worktree
• Background runner — send one agent after a flaky test while another builds the next feature; slow tasks finish off the critical path
• A/B the approach — run the same prompt with different framings, compare diffs side by side, and pick the design you actually want
• AI Arena mode — race agents head-to-head on the same problem and review the results competitively

Key Features

Beyond the core parallel dispatch model, the GitHub README lists a range of additional capabilities:

• Built-in diff viewer with inline review comments and per-commit navigation
• Steps tracking panel — an engineering-manager-style timeline of agent progress
• Notes panel per task with direct prompt injection
• PR CI status watcher with desktop notifications when GitHub checks settle
• Docker sandboxing via project-specific Dockerfiles (.parallel-code/Dockerfile)
• Coverage radar — per-file test-coverage badges in the Changed Files panel
• Phone monitoring via QR code over Wi-Fi or Tailscale
• 10 built-in themes and configurable keyboard shortcuts
• Direct mode for working on the main branch without isolation
• Existing worktree import

Update: v1.12.0

The latest release is v1.12.0, published on July 4, 2026. The repository was created in February 2026 and has accumulated 876 stars and 117 forks on GitHub as of mid-July 2026, with active issue tracking and ongoing pushes. The project is developed in public with a visible release history and issue tracker at the GitHub repository.

Tradeoffs to Know

Parallel Code currently supports macOS and Linux only — Windows is not listed as a supported platform. The Antigravity CLI agent cannot run inside Docker-isolated tasks because its keyring-based authentication cannot reach a secret-service daemon from inside a Linux container; it must be run as a native (non-Docker) task. The app requires at least one supported AI coding CLI to be installed separately before use, and API keys are managed by the user — nothing is proxied or collected by the app.