Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The new rules of context engineering for Claude 5 generation models

claude.com

Post reply on HN
Print view View original

171–180 of 434 posts

Re: The new rules of context engineering for Claude 5 generation models

#171

Post by margalabargala »

tossandthrow wrote: post #167
sothatsit wrote: post #150

Similarly, I recently disabled auto-memory in Claude Code, and performance improved. Managing the context that agents have available to them is far too important to leave to the agents themselves. Agents tend to write far too much into their memory, they are terrible at trimming it down, and their choice of what to include is very poor. I have had much more predictable results by disabling auto-memory and actively sh…

This article is about the Claude 5 series of models. My impressions is that they have overhauled the auto memory system. You might want to re assess how it works with the new generation of models.

Some update was made a while ago such that Opus and Sonnet 4 series both were more aggressive about storing memories, starting in April 2026 or so.

None of the 5 series models have appeared to have remotely different memory behavior.

Re: The new rules of context engineering for Claude 5 generation models

#172

Post by backwardsponcho »

jghn wrote: post #166
chaboud wrote: post #163

Earlier quoted context omitted.

I have heard so many of my co-workers use "load-bearing" over the last couple of months. It's truly comical. Maybe this is a way that we can make "fetch" happen.

Why is that funny. I've had that as part of my professional lexicon for over 20 years

The thing is that some people start to adopt certain "mannerisms" from their LLM of choice. It's not funny in and by itself, but it tends to be unnecessarily pompous words/expressions as well. Relevant: https://www.vice.com/en/article/youre-not-imagining-it-peopl...

Re: The new rules of context engineering for Claude 5 generation models

#173

Post by jghn »

backwardsponcho wrote: post #172
jghn wrote: post #166

Earlier quoted context omitted.

Why is that funny. I've had that as part of my professional lexicon for over 20 years

The thing is that some people start to adopt certain "mannerisms" from their LLM of choice. It's not funny in and by itself, but it tends to be unnecessarily pompous words/expressions as well. Relevant: https://www.vice.com/en/article/youre-not-imagining-it-peopl...

I've seen this too and I get it. However, we should not assume that certain phrases are AI tells. And that was my point. There are all kinds of things I see described here as "AI slop" that are things I just do, and have done, for decades.

Re: The new rules of context engineering for Claude 5 generation models

#174

Post by chickensong »

boron1006 wrote: post #161
chickensong wrote: post #159

Earlier quoted context omitted.

People make fun of the language, rightfully so in some cases, but also it's often quite effective language. "load-bearing seam" communicates quite a lot in very few characters.

Or in a way, nothing at all.

On its own, yes, but not in context. I think one of the problems with Claude's stock output is it assumes the reader has a firm grip on the context of the output, which is often untrue. Stock output is exhausting to read because you have to unwind metaphors in an unstated context.

Re: The new rules of context engineering for Claude 5 generation models

#175

Post by boorang »

m3h wrote: post #36

Is there a specific list of changes they made to the system prompt? They're claiming they removed 80% of it. That's quite substantial. It would be good to know what the model knows to do by training and what we need to avoid over-specifying in our system prompts. Saying that "give Claude judgment" is too vague for agent implementors. Given the lack of specific details, my takeaway is that we need to go and review all…

I spent some time running mitmproxy and watching the system prompts and it's what drove me to codex. the main issue for me was their system prompt wrapped the CLAUDE.MD with a "IMPORTANT: this context may or may not be relevant to your tasks. You should not respond to this context unless it is highly relevant to your task." https://github.com/anthropics/claude-code/issues/18560

Anyhow- if anyone is sufficiently curious and has access- just tell the agent to setup an mitmproxy to watch the traffic and see what the system prompt looks like.

Re: The new rules of context engineering for Claude 5 generation models

#176

Post by customguy »

pdimitar wrote: post #140
customguy wrote: post #133

Earlier quoted context omitted.

My point is that a percentage of deterministic makes no sense. Just like being a bit pregnant doesn't. Take "proper" UI. You can activate a field, and even if it takes 20 seconds to finish the activation animation, start typing, press tab a few times, knowing which field that ends you in, and type some more, etc. hit enter, hit enter again to confirm the dialog you know will pop at that point, and make tea, knowing t…

Here are a few examples of what I thought was clearly understandable from my "1-10% non-determinism in LLM coding output": - Variations in code patterns used. Might be a chain if if/else-s and not a case/switch statement; - Different decomposition of a hierarchy of functions/modules/classes; - Uses RED->GREEN test discipline, or not; - Writes the tests before the code, or not; - Different saga patterns (call 3rd part…

My point isn't to talk you out of doing what works for you, but simply to disagree that you can "approach determinism". You can get more and more predictable, but with determinism, you don't need to predict and make educated bets; you know, since it all follows from the input which you provide.

Subtraction and addition are deterministic, so you can add and subtract the same number from 0 ten or or a million times, with the same outcome. You never need to double check if a stray "coin flip" threw a wrench in it. To me that's more a property of the thing in question, not so much a practical matter. If for you in practice, it's as good as a deterministic tool, but better, that's great, but it's still fundamentally based on probabilities, that's kind of in the nature of it.

Re: The new rules of context engineering for Claude 5 generation models

#177

Post by nickm12 »

During the rise of agentic assistance, I've gotten the advice that "If the generated code isn't what you want, fix the steering, not the code". I've also been advised to not change the sometimes verbose comments and documentation that agents write because "it will help them the next time they look at the code".

I've been skeptical and these guidelines validate this. I continue to document code and write specs as I've always done. If an agent produces poor output or misunderstands, I use that as an opportunity to improve the docs, but in a way that that aims to be accessible for human peers, not the quirks of the current generation of models.

Re: The new rules of context engineering for Claude 5 generation models

#178

Post by ckolkey »

nickm12 wrote: post #177

During the rise of agentic assistance, I've gotten the advice that "If the generated code isn't what you want, fix the steering, not the code". I've also been advised to not change the sometimes verbose comments and documentation that agents write because "it will help them the next time they look at the code". I've been skeptical and these guidelines validate this. I continue to document code and write specs as I've…

One of the oddities of the comments Claude makes is that they are very referential to the current session, or reference now dead/removed code... Which I consider useless. Things like "implement this X way because it's faster than Y" ... But a human reading the code wouldn't know what Y is, or that it even existed necessarily. So, I find myself constantly fixing comments to be less temporarily dependant.

Re: The new rules of context engineering for Claude 5 generation models

#179

Post by qmmmur »

chrisweekly wrote: post #138
chmod775 wrote: post #113

Earlier quoted context omitted.

It's language and speech patterns that seem designed to trick readers into believing that claims are correct, even when the claims aren't based on anything and are possibly wrong. It was rewarded for this during training for some reason. Alternative theory: The LLMs only way to "think" about abstract concepts is through language, and this leaks into into conversation it has with humans. But humans generally prefer to…

"There is a depth of thought untouched by words, and deeper still a depth of formless feeling untouched by thought." - Rilke Your assertion that we don't think in language is questionable. It runs counter to the lived experience of developing thoughts through writing ("writing isn't capturing thinking -- it is thinking"). I believe there is more to thought than language alone, but I also feel quite sure that language…

Just because you can think through writing doesn't mean language is the essence of thought itself.

Re: The new rules of context engineering for Claude 5 generation models

#180

Post by fzaninotto »

I would love to follow this advice, expect there is no proof that it leads to better results. Maybe it’s only better on Anthropic codebases (e.g., Bun). Or maybe it’s only marginally better but substantially more expensive.

Other comments in this thread show that your mileage may vary. But we spend so much money on Claude Code and give it so many responsibilities that we deserve at least some undeniable proof that it’s bringing value.

Where is the evidence that this new type of prompting is better on real life examples?

Post reply on HN