bhaskarjha-dev/kramak ? reverse-engineered prompt

Build me a small open source project called Kramak that helps AI coding agents work in a more reliable way.

I want the repo to clearly explain and support a simple Plan, Execute, Audit workflow, with a file based setup that can be copied into any project. Make it feel usable right away, with a .kramak folder, adapter files for common IDEs, example setup files, and clear docs for how someone would start a run by telling their agent to begin.

Please make the README easy to understand for a normal developer, show the core idea without too much jargon, and include the basic rules for how the process should move forward and how it should avoid going off scope. If anything needs current reference docs, look them up online. Keep it lightweight, model agnostic, and IDE agnostic, with no required install steps if possible.