johannesjo/parallel-code

What it solves

Parallel Code solves the inefficiency of running AI coding agents sequentially. Instead of waiting for one agent to finish a task before starting the next, it allows developers to dispatch multiple agents in parallel, each working on a separate feature or bug without causing git conflicts or blocking the workflow.

How it works

When a user creates a task, the application automatically creates a new git branch and sets up a git worktree, providing the agent with its own isolated directory. It symlinks necessary directories like node_modules to maintain environment consistency and then spawns the chosen AI coding CLI (such as Claude Code, Codex, or Gemini) within that worktree. Once the agent completes the work, the user can review the diffs and merge the branch back into the main codebase.

Who it’s for

It is designed for developers who use AI coding CLIs and want to manage multiple concurrent AI-driven tasks through a single graphical interface rather than managing multiple terminal windows and manual git worktrees.

Highlights

• Multi-Agent Orchestration: Run multiple agents on different features simultaneously in isolated environments.
• CLI Integration: Works with existing tools like Claude Code, Codex CLI, Gemini CLI, and Copilot CLI.
• Git Isolation: Automatically handles branch and worktree creation for every task.
• Review Tools: Includes a built-in diff viewer with inline comments and a progress tracking panel.
• Sandboxing: Supports project-specific Dockerfiles to run agents in isolated containers.
• Remote Monitoring: Ability to monitor agent progress via a QR code over Wi-Fi or Tailscale.