We’re opening up early access to AgentSphere, a secure runtime platform for executing AI-generated code inside cloud sandboxes. If you’re building with LLMs (ChatGPT, Claude, etc.), and want them to not just write code but actually run it — this is for you.
--------------------------------------------
What is AgentSphere?
AgentSphere provides an easy way for you to launch lightweight, isolated environments in the cloud using our JavaScript SDK, Python SDK or APIs. Each sandbox behaves like a containerized runtime — Think of it as Execution-as-a-Service: no local setup, no config hell — just send a command and it runs securely.
--------------------------------------------
How AgentSphere Works with MCP Clients?
AgentSphere is integrated into the MCP ecosystem and works seamlessly with:
- Cursor
- Claude Code
- Raycast
- Gemini CLI
- and more MCP clients...
Reference Docs: https://www.agentsphere.run/docs/mcp-server
--------------------------------------------
Demo
The following video clip demonstrates what AgentSphere can help in action.
- Web Game Mario: https://www.youtube.com/watch?v=5mtqUzAwXoA
- Calculator: https://www.youtube.com/watch?v=ZXTPOddl_Jo
--------------------------------------------
Join the Beta
If you want to:
- Build LLM agents that actually execute code and handle data
- Use Cursor, Claude Code, Raycast, Gemini CLI, or similar MCP clients and want the output to “just run”
- Experiment with autonomous agent workflows
- Test AI-generated code in a safe, isolated environment
Just give AgentSphere a try.
Want to build with AgentSphere? Register now and grab your API key in just a few clicks!
We're looking for developers, builders, and AI toolmakers to test it out, and tell us what is great and what doesn’t. We'd love your feedback.
Got questions? Drop them below. Thank you!