tirth8205/code-review-graph ? reverse-engineered prompt
Reverse engineered prompt
Build me a Python command-line tool that makes my AI coding assistant, like Claude or Cursor, way more efficient and saves a ton on tokens. The core idea is to stop the AI from re-reading my entire project for every single task.
This tool should scan my codebase once to build a "knowledge graph" that understands how all the files, functions, and classes are connected. Then, when I change a file, it should be able to figure out the "blast radius"—all the other parts of the code that could be affected—and only feed that precise context to the AI.
It should have an installer that automatically detects and configures itself for my existing coding tools. After the first build, it needs to update this graph incrementally and automatically whenever I save a file or commit. Make sure it can parse a bunch of different languages, not just Python, and also handle Jupyter notebooks.
Are you gonna build this?
make sure you review the code using coderabbit