there's a almost-year-old thread about spec driven development which has recently devolved into claude slop but CLEARLY demonstrates how hard it is to try and keep ALL your specs in sync with ALL your code - you basically have just manufactured a new problem that doesn't actually yield that much leverage

(you may save a little bit of time/tokens during research but if you codebase research is dialed in, just spend the tokens my friends - trying to compact the moving target of "what is the structure and intent of this codebase" has enough downsides that its probably not worth the upside)

https://github.com/github/spec-kit/discussions/152#discussioncomment-18080849