yeah, nothing is built in atm that would curcumvent this. While qwq is the best when it comes to reasoning, you could try and use Gemini as a temp workaround as it should have a higher context window (although I don't know what kagi limits it to in Assistant).
@ kagi team
In this situation it's a limitation of the model so the best solution would be to embed the files instead of parsing it as actual text
