Hacker News Reader

Comment by bcherny | Hacker News Reader

When you submit a /bug we get a way to see the contents of the conversation. We don't see anything else in your codebase.

View on HN

Was there a change in Claude Code system prompt at that time that nudges Claude into simplistic thinking?

Here is a gist that tries to patch the system prompt to make Claude behave better https://gist.github.com/roman01la/483d1db15043018096ac3babf5...

I haven’t personally tried it yet. I do certainly battle Claude quite a lot with “no I don’t want quick-n-easy wrong solution just because it’s two lines of code, I want best solution in the long run”.

If the system prompt indeed prefers laziness in 5:1 ratio, that explains a lot.

I will submit /bug in a few next conversations, when it occurs next.

Holy sweet LLM, this gist is crazy. Why did they do this to themselves? I am going to try this at home, it might actually fix Claude.

Remember Sonnet 3.5 and 3.7? They were happy to throw abstraction on top of abstraction on top of abstraction. Still a lot of people have “do not over-engineer, do not design for the future” and similar stuff in their CLAUDE.md files.

So I think the system prompt just pushes it way too hard to “simple” direction. At least for some people. I was doing a small change in one of my projects today, and I was quite happy with “keep it stupid and hacky” approach there.

And in the other project I am like “NO! WORK A LOT! DO YOUR BEST! BE HAPPY TO WORK HARD!”

So it depends.

Let us know if it does, because we all want it to work :)

That Gist does explain quite a few flaws Claude has. I wonder if MEMORY.md is sufficient to counteract the prompt without patching.

And if memory.md can’t and you need something quick and dirty for flat memory management, I wrote a plugin just for this.

https://github.com/NominexHQ/pmm-plugin

Is there not a setting to change the system prompt itself? I vaguely remember seeing it in the docs.

There is!!

https://code.claude.com/docs/en/cli-reference#system-prompt-...

  --append-system-prompt
  --append-system-prompt-file
  --system-prompt
  --system-prompt-file

Can this script be made to work without patching the executable?

Might be worth extracting the system prompt and then patching it. TBH, that's what I was expecting when I saw the gist.

This might be more complex than I imagined. It seems Claude Code dynamically customizes the system prompt. They also update the system prompt with every version so outright replacing it will cause us to miss out on updates. Patching is probably the best solution.

https://github.com/Piebald-AI/claude-code-system-prompts

https://github.com/Piebald-AI/tweakcc

Interesting. So literally triggering any of these changes probably invalidates the cache as well…

I didn't know we could change the base system prompt of Claude Code. Just tried, and indeed it works. This changes everything! Thank you for posting this!

Very interesting. I run Claude Code in VS Code, and unfortunately there doesn't seem to be an equivalent to "cli.js", it's all bundled into the "claude.exe" I've found under the VS code extensions folder (confirmed via hex editor that the prompts are in there).

Edit: tried patching with revised strings of equivalent length informed by this gist, now we'll see how it goes!

I adapted these patches into settings for the tweakcc tool.

https://github.com/Piebald-AI/tweakcc

Pushed it to my dotfiles repository:

https://github.com/matheusmoreira/.files/tree/master/~/.twea...

The tweaks can be applied with

  npx tweakcc --apply

Isnt the codebase in the context window?

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.

This is crazy..

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

I mean whatever part of the code that is read by the AI has to be in the content window at some point or another nSprewd throughout your sessions Id think even with a huge codebase, 90% of it is going to be there

Reply on news.ycombinator.com