flamehaven01/ai-slop-detector — reverse-engineered prompt

Reverse engineered prompt

Build me a local code quality tool that catches the kind of AI generated code that looks polished but is actually hollow. I want it to scan a repo offline, give each file a simple risk score from 0 to 100, and flag things like fake looking functions that do nothing, imports that never connect to real usage, handlers or pipelines that are defined but not wired up, duplicated logic, and comments or docs that oversell what the code really does.

Make it feel practical for real teams. I want a clean command line tool with a few obvious commands for full scans, changed code review, overall repo health, and dead code cleanup. It should output both a human friendly summary and structured JSON for automation. Add config support, sensible defaults, and CI friendly pass fail modes.

Python should be the main focus, with optional support for JS, TS, and Go if that already fits. Also include the editor integration and basic tests so it feels production ready. Look up current docs online if you need to.

Want more depth? Deep Reverse