Wonder Dev View is a development environment for macOS that connects coding agents, Git worktrees, and terminal sessions directly beside a live application preview and design canvas.
What the product does and how it is positioned
Wonder Dev View is a specialized developer harness that positions coding agents and repository tools alongside an interactive design canvas. Operating on macOS, the software displays project files, command-line terminals, agent chats, and GitHub pull requests in a unified interface designed for engineers building user interfaces.
The tool operates on a bring-your-own-agent model, integrating with existing Claude Code and Codex subscriptions. By isolating each task into an individual git worktree, the environment enables developers to run parallel agent tasks simultaneously without merge conflicts or manual dependency installations.
Source-supported ways to use the product
Running multiple coding agents simultaneously on separate git worktrees to develop distinct features or bug fixes without working directory collisions.
Previewing live application components next to agent chat windows to inspect layout behavior and direct agents to adjust front-end code.
Triggering secondary agent code reviews on open branches and pull requests to identify issues before completing code merges.
The documented workflow, where available
Open a repository branch or teammate pull request, which creates a dedicated, isolated git worktree for the task.
Instruct a connected agent model to analyze the codebase and generate required modifications across project files.
Launch the local project using configured run commands to view the live interface inside the canvas preview.
Review staged file diffs and use built-in Git tools to commit, push, and submit a pull request without leaving the application.
Run an automated review using a second agent model to catch issues and refine code quality prior to merging.
Wonder Dev View coordinates multiple coding agents by giving each workspace its own independent git worktree. Because each worktree represents a distinct copy of a branch, parallel tasks avoid the file locks, dependency collisions, and terminal switching typically associated with concurrent local development.
The environment interfaces with user-supplied AI agent subscriptions, including Claude Code and Codex. Developers can register multiple user accounts and select the desired model and reasoning effort level on a per-chat basis, supporting appropriate agent allocation for each specific coding task.
Checks to run with your own material and workflow
What was checked and when
Answers based on the source-checked product record
No, cloud workspaces are not currently supported. Coding agents currently run locally on the user machine, though cloud workspaces are planned for future release.
Yes. Wonder functions as the interface harness, and users must connect their own existing subscriptions for agents such as Claude Code or Codex.
Wonder isolates each workspace into its own git worktree, providing a separate copy of the branch so that parallel agent tasks operate without colliding.
The canvas renders real front-end structure rather than static images, allowing agents to pull screens from code, apply visual edits back to code, and show live app previews.
Yes, users need at least basic knowledge of Git and GitHub, as workspaces and agent chats are structured around git branches, commits, pushes, and pull request tracking.
