codekalakaars/vajra ? reverse-engineered prompt

Reverse engineered prompt

Build me a command line tool called Vajra that opens a shell inside the current project and keeps an AI coding agent boxed in so it can’t read my secrets, write outside the repo, or delete things it shouldn’t. I want vajra to start that confined shell, and vajra doctor to check whether this machine can actually enforce the protections and tell me how to fix it if not.

Inside the project, access should be controlled by a simple .vajra.json policy file, with projects read only by default if no policy exists. Outside the project, keep access very limited, and refuse to run if the machine can’t enforce the sandbox properly. It should also open a local control page in the browser so I can review file permissions, change them, save the policy, and restart the shell safely. If you need to look up current docs online for the sandbox pieces, go ahead.

Are you gonna build this?

make sure you review the code using coderabbit

Try freeSponsored — opens CodeRabbit in a new tab