krishnaKanta2008/Igris ? reverse-engineered prompt
Reverse engineered prompt
Build me a Linux based AI native OS environment that puts a safe permission layer between an AI agent and the machine.
I want a Rust core daemon and command line client that can talk over a local Unix socket, with default deny permissions, audit logging, and structured tools for system info, reading and listing files inside a confined folder, looking at processes, and sending a safe terminate signal only after confirmation. Please keep everything locked down so the AI never runs raw shell commands, only approved operations.
Also include the basic Python agent layer as a scaffold so it can call those tools later, plus the docs and helper scripts needed to run and test it. If anything depends on current Linux details or library behavior, look up the latest docs online if needed. Make it runnable on a normal Linux dev setup, with clear build and test commands, and make sure the security rules are actually enforced.
Are you gonna build this?
make sure you review the code using coderabbit