> The main takeaway, again, is to keep things simple.

if true this seems like a bloated approach but tbh I wouldn't claim to know totally how to use Claude like the author here...

I find you can get a lot of mileage out of "regular" prompts, I'd call them?

Just asking for what you need one prompt at a time?

I still can't visualize how any of the complexity on top of that like discussed in the article adds anything to carefully crafted prompts one at a time

I also still can't really visualize how claude works compared to simple prompts one at a time.

Like, wouldn't it be more efficient to generate a prompt and then check it by looping through the appendix sections ("Main Claude Code System Prompt" and "All Claude Code Tools"), or is that basically what the LLM does somewhat mysteriously (it just works)? So like "give me while loop equivalent in [new language I'm learning]" is the entirety of the prompt... then if you need to you can loop through the appendix section? Otherwise isn't that a massive over-use of tokens, and the requests might even be ignored because they're too complex?

The control flow eludes me a bit here. I otherwise get the impression that the LLM does not use the appendix sections correctly by adding them to prompts (like, couldn't it just ignore them at times)? It would seem like you'd get more accurate responses by separating that from whatever you're prompting and then checking the prompt through looping over the appendix sections.

Does that make any sense?

I'm visualizing coding an entire program as prompting discrete pieces of it. I have not needed elaborate .md files to do that, you just ask for "how to do a while loop equivalent in [new language I'm learning]" for example. It's possible my prompts are much simpler for my uses, but I still haven't seen any write-ups on how people are constructing elaborate programs in some other way.

Like how are people stringing prompts together to create whole programs? (I guess is one question I have that comes to mind)

I guess maybe I need to find a prompt-by-prompt breakdown of some people building things to get a clearer picture of how LLMs are being used

How you see and use it is the same way I do. So interested to hear other replies