How Claude Code works in large codebases • Hacker News Board

Re: How Claude Code works in large codebases

#81

Post by bryanrasmussen » Fri, May 15, 2026, 6:58 AM UTC

suzzer99 wrote: post #63

nopinsight wrote: post #57

Earlier quoted context omitted.

This is already the case for many startups. In fact, the figure might be closer to 100%. The work shifts to requirements analysis, high-level specifications, and final review instead (after AI code review).

Yeah I'm working on one of those now that a 3rd-party vendor cranked out for us. I spent all day ripping out an endpoint that did 98% of what another endpoint did and should never have existed. I also ripped out 80 lines of code that looked like this: const sqlStatement = (!params.mostRecentOnly) ? {giant SQL statement} : {identical giant SQL statement + 'LIMIT 1' at the end} AI never met a problem that can't be solv…

I mean quite frankly I have seen enough code that was definitely written by humans that had exactly this "style".

Then again I don't want to pay for AI to give me the coding style of the worst I ever worked with either.

Re: How Claude Code works in large codebases

#82

Post by MaxikCZ » Fri, May 15, 2026, 7:00 AM UTC

jameson wrote: post #79

Why can't Claude Code generate effective harness for us by inspecting the code base? I tried defining CLAUDE.md (or AGENTS.md), skills, plugins, but I'm not getting the effectiveness others claim to be. LSP plugin for example, CC doesn't to use LSP's symbol renaming and edits file one by one slowly, or it does not invoke the skill when I explicitly ask to remember to invoke when prompt contains a specific clue. Am I…

This is the pain point that existed for years now and its still not solved at all.

"If A, do X. Do B,C,D. Do A" - and it just never uses X because "it forgot".

You just cant trust that the time you spend building rules will actually pay off, in fact you can trust that it will fail you sooner or later.

RAG, Harness, Skills... all was supposed to fix this, but in reality it never had.

Re: How Claude Code works in large codebases

#83

Post by Kwpolska » Fri, May 15, 2026, 7:02 AM UTC

mystifyingpoi wrote: post #68

martypitt wrote: post #55

I don't have any LSP's hooked up to CC yet (going to fix that today), or particularly sophisticated CLAUDE.md files. So, if I've read this post correctly, that means that CC is navigating my codebase today by sending lots of it up to a model, and building an understanding. Is that correct? Did I misunderstand it? I kinda suspected there was more local inference going on somehow -- partly because the iteration times a…

I think that's correct. Which is kinda funny, I remember 10y ago that I was heavily relying on IntelliJ features to understand new codebases (jump to definition, find all usages of a function, navigate from SQL to the table in database tab etc.). It turns out, that for a machine, find and grep is all that's required.

A human could get by with just find and grep too. And in both cases, find and grep will be slower and less precise than an IDE's code navigation features.

Re: How Claude Code works in large codebases

#84

Post by lukaslalinsky » Fri, May 15, 2026, 7:02 AM UTC

jwilliams wrote: post #26

> Claude Code navigates a codebase the way a software engineer would: it traverses the file system, reads files, uses grep to find exactly what it needs, and follows references across the codebase. It operates locally on the developer’s machine and doesn’t require a codebase index to be built, maintained, or uploaded to a server.... > Agentic search avoids those failure modes. There's no embedding pipeline or central…

It works exactly the way I'd work. I have learned to navigate large codebases before LSPs existed. I used vim for many years and would grep to find the relevant files. When I first tried Claude Code last year, I was like WTF, it's going exactly what I'd be doing.

Re: How Claude Code works in large codebases

#85

Post by mewpmewp2 » Fri, May 15, 2026, 7:03 AM UTC

suzzer99 wrote: post #63

nopinsight wrote: post #57

Earlier quoted context omitted.

This is already the case for many startups. In fact, the figure might be closer to 100%. The work shifts to requirements analysis, high-level specifications, and final review instead (after AI code review).

Yeah I'm working on one of those now that a 3rd-party vendor cranked out for us. I spent all day ripping out an endpoint that did 98% of what another endpoint did and should never have existed. I also ripped out 80 lines of code that looked like this: const sqlStatement = (!params.mostRecentOnly) ? {giant SQL statement} : {identical giant SQL statement + 'LIMIT 1' at the end} AI never met a problem that can't be solv…

That is a skill issue though. I have rules for my agents to write compositional, reusable, modular, small files and to avoid any sort of boilerplate etc. Being config driven, single source of truth, having other agents review that rules are followed, etc. Any API or UI or any sort of entry points very light, just proxying to the modular logic basically, so this logic could be reused by any entrypoint easily.

UI components always presentational only logic abstracted modularly, etc...

Re: How Claude Code works in large codebases

#86

Post by fragmede » Fri, May 15, 2026, 7:04 AM UTC

jwilliams wrote: post #26

> Claude Code navigates a codebase the way a software engineer would: it traverses the file system, reads files, uses grep to find exactly what it needs, and follows references across the codebase. It operates locally on the developer’s machine and doesn’t require a codebase index to be built, maintained, or uploaded to a server.... > Agentic search avoids those failure modes. There's no embedding pipeline or central…

That's the question, innit? Dumped into a codebase and given a ticket, what's the fastest way to get your bearings and do the ticket? It's gonna depend on the codebase and the ticket, but it would be an interesting contest to see what tools people have. Some form of grep, sped up using an index, is going to get a skilled operator pretty far, but more complex tools for more complex tickets, eg fix something subtle, like a bug that only manifests on Tuesdays in 2% of requests from Poland, I imagine more advanced tools would help the programmer figure it out faster.

Re: How Claude Code works in large codebases

#87

Post by cdnsteve » Fri, May 15, 2026, 7:04 AM UTC

Small plug for what I built:

You need a code dependency graph: https://github.com/roboticforce/remembrallmcp Ask "what breaks if I change this?"

Saves 98% token usage. Saves 95% tools being called.

Runs as an MCP server, works for 8 languages.

It just works, you need to try it.

Re: How Claude Code works in large codebases

#88

Post by lukaslalinsky » Fri, May 15, 2026, 7:04 AM UTC

hibikir wrote: post #33

jwilliams wrote: post #26

> Claude Code navigates a codebase the way a software engineer would: it traverses the file system, reads files, uses grep to find exactly what it needs, and follows references across the codebase. It operates locally on the developer’s machine and doesn’t require a codebase index to be built, maintained, or uploaded to a server.... > Agentic search avoids those failure modes. There's no embedding pipeline or central…

Even if there is first principles traversal of some parts of the codebase, there are other bits that definitely not change, and where exploring every time is a massive waste of tokens. My arguments with claude often have to do with making it explore a lot less, because I know better, and faster, than its slow, expensive navigation of things that basically never change. And it just goes into the same kind of rabbit ho…

I still think the best process with Claude Code is: 1) ask it to gather context that you know is relevant 2) only then ask it to do whatever you want it to do. If you do it the other way around, it will over research, over think and generally make more of a mess.

Re: How Claude Code works in large codebases

#89

Post by Kwpolska » Fri, May 15, 2026, 7:06 AM UTC

polotics wrote: post #28

zarzavat wrote: post #16

Earlier quoted context omitted.

Apologies but what is a Boris Cat?

Boris Cherny and Cat Wu are the lead devs of CC at Anthropic who unsurprisingly talk their book and find so many ways to justify tokenmaxing. As the product they deliver is greenfield and in the newest of domain spaces, there is a serious halo-effect to consider. On a side note, at a company I know the devs are split between Stick to Copilot inside visual studio - suspiciously cheap Opus quotas there + they read thei…

Copilot's extreme subsidies end this month. Starting in June, you'll be paying API rates for all models.

Re: How Claude Code works in large codebases

#90

Post by batshit_beaver » Fri, May 15, 2026, 7:07 AM UTC

cbg0 wrote: post #66

suzzer99 wrote: post #63

Earlier quoted context omitted.

Yeah I'm working on one of those now that a 3rd-party vendor cranked out for us. I spent all day ripping out an endpoint that did 98% of what another endpoint did and should never have existed. I also ripped out 80 lines of code that looked like this: const sqlStatement = (!params.mostRecentOnly) ? {giant SQL statement} : {identical giant SQL statement + 'LIMIT 1' at the end} AI never met a problem that can't be solv…

If only we had this tech back when managers were looking at how many lines of code you were committing weekly as a performance metric.

Now they're looking at your token consumption, which is even more gameable (and stupid).