thanuj916/Terminal-AI- — reverse-engineered prompt

Build me a Python terminal assistant that lets me type normal requests like “create a file”, “show my current folder”, “install this package”, or “set up a git repo”, then has an LLM turn that into a safe shell command.

I want it to show me the command first and ask for confirmation before running anything. When it runs a command, capture the output and show it back in the terminal so the AI can explain what happened. It should know the current directory, be able to create and write files, display results, and handle errors clearly.

Please make it a simple working version for Linux first, with a clean command loop where I can keep chatting instead of restarting every time. Add basic safety checks so it doesn’t blindly run dangerous commands, and make the timeout configurable instead of fixed. Keep the code easy to understand because this is more of a concept project than a polished product. Look up current docs online if you need to.