Clojurians Log v2

We keep adding to instructions normally and yes it's mentioned in the prompt, I see no issue with that and it should help LLM understand what you are referencing from the available contexts

It usually reads the file again, I am afraid that when it sees something like @index.html in the prompt then it doesn’t associate it with e.g. /eca_test/index.html in context. When you do this in Claude Code it just adds the content of the file into the prompt like:

What you see in this file:  content of the file  ?

I have two index files, first one: @index.html and @index.html, take the footer from the first one and add it to the second one

hum, yes we always had this problem, it's something to improve indeed

Are you saying that you're hallucinating homoiconically? 🤣

Yeah, maybe AI got into my dreams already 🫠

Yeah, this seems very promising! Keep it up, Eric!

I see the files are also being added to the upper context, but that only functions as a system message / instruction, right? Or are these files added to the prompt itself, as part of the chat history?