HiTa475/daytona — reverse-engineered prompt

Build me a platform like Daytona for safely running AI generated code in isolated sandboxes.

I want developers to be able to create a sandbox with an API key, choose a language like Python or TypeScript, run code inside it, get the output and error messages back, then keep the sandbox around or delete it. The main point is that untrusted code should never touch the user’s real system or infrastructure.

Please include SDK examples for Python, TypeScript, and Go, with a simple flow that creates a sandbox, runs a small piece of code, prints the result, and cleans up. It should also support basic programmatic control for files, git, language tools, and command execution. If possible, let users create sandboxes from Docker compatible images.

Make the README clear enough that someone can install the SDK, add their API key, and run the first example quickly. Look up current docs online if you need to.