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

Universal Claude.md – cut Claude output tokens

github.com

Post reply on HN
Print view View original

71–80 of 192 posts

Re: Universal Claude.md – cut Claude output tokens

#71

Post by peacebeard »

skeledrew wrote: post #44

Strange. I've never experienced verbosity with Claude. It always gets right to the point, and everything it outputs tends to be useful. Can actually be short at times. ChatGPT on the other hand is annoyingly wordy and repetitive, and is always holding out on something that tempts you to send a "OK", "Show me" or something of the sort to get some more. But I can't be bothered with trying to optimize away the cruft as…

My experience is that Sonnet can be a bit verbose and prompting it to be more succinct is tricky. On the other hand, Opus out of the box will give me a one word answer when appropriate, in Claude Code anyway.

Re: Universal Claude.md – cut Claude output tokens

#74

Post by flashgordon »

sillysaurusx wrote: post #7
btown wrote: post #4

It seems the benchmarks here are heavily biased towards single-shot explanatory tasks, not agentic loops where code is generated: https://github.com/drona23/claude-token-efficient/blob/main/... And I think this raises a really important question. When you're deep into a project that's iterating on a live codebase, does Claude's default verbosity, where it's allowed to expound on why it's doing what it's doing when it…

I wrote a skill called /handoff. Whenever a session is nearing a compaction limit or has served its usefulness, it generates and commits a markdown file explaining everything it did or talked about. It’s called /handoff because you do it before a compaction. (“Isn’t that what compaction is for?” Yes, but those go away. This is like a permanent record of compacted sessions.) I don’t know if it helps maintain long term…

I've actually been doing this for a year. I call it /checkpoint instead and it does some thing like:

* update our architecture.md and other key md files in folders affected by updates and learnings in this session. * update claude.md with changes in workflows/tooling/conventions (not project summaries) * commit

It's been pretty good so far. Nothing fancy. Recently I also asked to keep memories within the repo itself instead of in ~/.claude.

Only downside is it is slow but keeps enough to pass the baton. May be "handoff" would have been a better name!

Re: Universal Claude.md – cut Claude output tokens

#76

Post by fwipsy »

airstrike wrote: post #50
dataviz1000 wrote: post #31

Earlier quoted context omitted.

Oh, so the word 'gate' is probably in the documentation also! I see. So this isn't as scary. Claude is helping me understand how to use it properly.

Why would it be scary? Claude is just parroting other human knowledge. It has no goal or agency.

By that logic, nothing computers do is scary.

Re: Universal Claude.md – cut Claude output tokens

#77

Post by whattheheckheck »

hatmanstack wrote: post #20
btown wrote: post #4

It seems the benchmarks here are heavily biased towards single-shot explanatory tasks, not agentic loops where code is generated: https://github.com/drona23/claude-token-efficient/blob/main/... And I think this raises a really important question. When you're deep into a project that's iterating on a live codebase, does Claude's default verbosity, where it's allowed to expound on why it's doing what it's doing when it…

Seems crazy to me people aren't already including rules to prevent useless language in their system/project lvl CLAUDE.md. As far as redundancy...it's quite useful according to recent research. Pulled from Gemini 3.1 "two main paradigms: generating redundant reasoning paths (self-consistency) and aggregating outputs from redundant models (ensembling)." Both have fresh papers written about their benefits.

No such thing as junk DNA kinda applies here

Re: Universal Claude.md – cut Claude output tokens

#78

Post by david_allison »

sillysaurusx wrote: post #12
david_allison wrote: post #11

Earlier quoted context omitted.

Is this available online? I'd love documentation of my prompts.

I’ll post it here, one minute. Ok, here you go: https://gist.github.com/shawwn/56d9f2e3f8f662825c977e6e5d0bf... Installation steps: - In your project, download https://gist.github.com/shawwn/56d9f2e3f8f662825c977e6e5d0bf... into .claude/commands/handoff.md - In your project's CLAUDE.md file, put "Read `docs/agents/handoff/*.md` for context." Usage: - Whenever you've finished a feature, done a coherent "thing", or oth…

Oh wow, thank you so much!!!!!

Re: Universal Claude.md – cut Claude output tokens

#80

Post by niklassheth »

So many problems with this:

The benchmark is totally useless. It measures single prompts, and only compares output tokens with no regard for accuracy. I could obliterate this benchmark with the prompt "Always answer with one word"

This line: "If a user corrects a factual claim: accept it as ground truth for the entire session. Never re-assert the original claim." You're totally destroying any chance of getting pushback, any mistake you make in the prompt would be catastrophic.

"Never invent file paths, function names, or API signatures." Might as well add "do not hallucinate".

Post reply on HN