depending on how large your codebase is, hopefully not. At this point use something like the IX plugin to ingest codebase and track context, rather than from the LLM itself.
tokensSaved = naiveTokens - actualTokens
- naiveTokens = 19.4M — what ix estimates it would have cost to answer your queries without graph intelligence (i.e., dumping full files/directories into context)
- actualTokens = 4.7M — what ix's targeted, graph-aware responses actually used
- tokensSaved = 14.7M — the difference