Wonder
Local IDE and design canvas for running multiple Claude Code and Codex agents in parallel worktrees on your own subscriptions, with live app preview, element-select edits, cross-agent code review, and MCP connections to Cursor, Claude Desktop, Antigravity and Lovable.
Sources: wonder.design/dev, wonder.design/pricing JSON-LD Offers (Free price 0; Pro monthly 20; Pro yearly 192) checked 2026-11-06. Mac and Windows. Agent model usage billed by your Claude/ChatGPT (etc.) plan, not Wonder. Product Hunt Wonder Dev View launch 2026-11-06. Not yet tested or ranked.
| Website | https://wonder.design |
|---|---|
| Best for | Orchestrating Claude Code and Codex side by side on a real repo while designing on a canvas next to the running app |
| Pricing | Free $0 forever (one editor; own agent subscriptions; parallel agents; 3,000 AI-generated layers/week); Pro $20/editor/mo or $16/editor/mo billed yearly ($192/yr); Enterprise custom (SSO/SAML, SCIM, audit log) (last checked 2026-11-06, source) |
| Free tier | Yes |
| Open source | No |
| License | Not yet verified |
| Hands-on test | Test coming |
Machine-readable: Markdown · JSON dataset
For makers: add this badge
Wonder is listed on VibeTested. Show it on your site or README; the badge links back to this page. About badges
Wonder FAQ
How much does Wonder cost?
Free $0 forever (one editor; own agent subscriptions; parallel agents; 3,000 AI-generated layers/week); Pro $20/editor/mo or $16/editor/mo billed yearly ($192/yr); Enterprise custom (SSO/SAML, SCIM, audit log) (checked 2026-11-06 on https://wonder.design/pricing).
Does Wonder have a free tier?
Yes, Wonder has a free tier.
Is Wonder open source?
No, Wonder is not open source.
What are some alternatives to Wonder?
Other Agent workspace tools listed on VibeTested include Aether, Agent37, Alan, Beacon.
Alternatives to Wonder
Aether
Cloud Devboxes for Claude Code, Codex or OpenCode that run on the user’s existing AI subscription; each task gets a watchable Firecracker microVM with live terminal, file tree, port previews, optional agent-vs-agent PR review loop, and session video/screenshots on PRs.
Agent37
YC F26 prepaid cloud sandboxes for any agent harness (Hermes, OpenClaw, Claude Code, Codex, OpenCode, Grok, Pi or your Docker image): gVisor-isolated boxes with auto-sleep, persistent disk, networking, observability and optional managed models/integrations.
Alan
Orchestration workspace that runs Claude Code, Codex and Cursor in parallel on one task, with a shared codebase context layer, cloud sessions and sandboxes, tests and review inside the agent loop, approval checkpoints, and one task board for people and agents.
Beacon
Open-source memory layer from Asymptote Labs that records your sessions across Claude Code, Cursor, Codex, OpenCode, Cline and 20+ other agents, then turns reviewed fixes and workflows into skills any agent can pull in through MCP or Agent Skills.