Is WonderAgentPlay an official Meta product?

No. It is an independent third-party playground and is not affiliated with Meta.

Do I need a Meta API key or local GPU?

No. The site says its browser workflow does not require a Meta API key or local GPU.

Does it support Meta Active-Memory?

No. Conversation history is kept in this browser through localStorage, and Meta native Active-Memory is not available.

What templates are available?

The playground provides Code fix, Unit tests, and JSON format templates. Custom prompts are also supported.

Does the JSON option provide a native schema guarantee?

No. It makes a soft request and performs a server-side parse check; it is not a native JSON Schema guarantee.

Does the playground run tests or access a terminal?

No. It does not provide a sandboxed test runner, arbitrary tool execution, terminal access, or filesystem access.

How are credits charged?

When usage is known, credits are charged from actual input and output tokens. Unknown usage is held conservatively for reconciliation.

Wonder Agent Play

Independent playground for Muse-Wonder-1.0 (coming soon)

AI Tools Developer Tools

Visit website

Launched: Nov 8, 2026
Website: wonderagentplay.com
Pricing: Free
Platforms: Web
Based in: United States

An independent, unofficial workbench for the Muse-Wonder-1.0 coding model with prompt templates for code fixes, unit tests and JSON output. The playground is locked until the model goes live; not affiliated with Meta.

What Wonder Agent Play does

Coding prompt templates: Start with Code fix, Unit tests, or JSON format templates, or write a custom prompt.
Editable system prompt: Adjust the system prompt; it counts toward input cost and can be reset to the default helper prompt.
Code-fix workflow: Use a template to diagnose runtime errors and propose a minimal fix.
Unit-test drafting: Generate focused test drafts for a pure function, including Vitest-style workflows.
JSON request and validation: Request JSON-shaped output and see a server-side parse check; this is not a native schema guarantee.
Browser-only conversation history: Keep in-session conversation history in localStorage on the current browser; persistent Active-Memory is unavailable.

Who it's for

Developers and other users who want to try Muse-Wonder coding prompts in a browser without managing a Meta API key or local GPU.

Why people use Wonder Agent Play

Trying a coding model can require API credentials, local hardware, or a separate development setup; users also need a simple way to test prompts for coding tasks.

Provides a browser playground with coding templates, editable prompts, output controls, conversation context, and JSON parse checks for Muse-Wonder responses.

It is an independent third-party playground specifically for meta/muse-wonder-1.0, with browser-based session history and no requirement for a Meta API key or local GPU.

Pricing

Wonder Agent Play is free to use.

Platforms

Wonder Agent Play runs on Web.

Wonder Agent Play FAQ

Is WonderAgentPlay an official Meta product?: No. It is an independent third-party playground and is not affiliated with Meta.
Do I need a Meta API key or local GPU?: No. The site says its browser workflow does not require a Meta API key or local GPU.
Does it support Meta Active-Memory?: No. Conversation history is kept in this browser through localStorage, and Meta native Active-Memory is not available.
What templates are available?: The playground provides Code fix, Unit tests, and JSON format templates. Custom prompts are also supported.
Does the JSON option provide a native schema guarantee?: No. It makes a soft request and performs a server-side parse check; it is not a native JSON Schema guarantee.
Does the playground run tests or access a terminal?: No. It does not provide a sandboxed test runner, arbitrary tool execution, terminal access, or filesystem access.
How are credits charged?: When usage is known, credits are charged from actual input and output tokens. Unknown usage is held conservatively for reconciliation.