Support for a wiki/codebase knowledge skill to optimize context token usage?

[ChrisPar19]

Hi,
kinda new in the world of coding agents and llms in general, so sorry in advance if this is a stupid question.
First of all, little-coder is a fantastic tool that made working with my local models such a bliss, I can't thank you enough for your work, I would say it's revolutionary!!
My question is about the usage of tools like graphify inside little-coder, from what I understood about graphify, it generates a graph knowledge of the codebase so that the llm can explore much easier and with less token usage than normally would. The same applies to other tools to build some kind of wikis, as llm-wiki and so on.
So if this is useful to reduce token usage of paid cloud models, couldn't it work also to consume less context on our limited humble workstations?