Top
Best
New

Posted by kmelve 9/2/2025

A staff engineer's journey with Claude Code(www.sanity.io)
549 points | 395 commentspage 2
tkgally 9/2/2025|
Anthropic just posted an interview with Boris Cherny, the creator of Claude Code. He also offers some ideas on how to use it.

“The future of agentic coding with Claude Code”

https://youtu.be/iF9iV4xponk

nikcub 9/2/2025||
> budget for $1000-1500/month for a senior engineer going all-in on AI development.

Is this another case of someone using API keys and not knowing about the claude MAX plans? It's $100 or $200 a month, if you're not pure yolo brute-force vibe coding $100 plan works.

https://www.anthropic.com/max

vincent_builds 9/3/2025||
Author here, quick clarification on pricing: the $1000-1500/month is for Teams/Enterprise with higher rate limits, not the consumer MAX plans. Consumer MAX ($200/month) works for lighter usage but hits limits quickly with parallel agents and large codebases.

For context: that's 1-2% of a senior engineer's fully loaded cost. The ROI is clear if it delivers even 10% productivity gain (we're seeing 2-3x on specific tasks).

You're right that many devs can start with MAX plans. The higher tier becomes necessary when running multiple parallel contexts and doing systematic exploration (the "3-attempt pattern" burns tokens fast).

I wouldn't be doing it if I didn't think it was value for money. I've always been a cost-conscious engineer who weighs cost/value, and with Claude, I am seeing the return.

imron 9/4/2025||
> The ROI is clear if it delivers even 10% productivity gain

What if what feels like a productivity gain is actually a productivity loss?

https://mikelovesrobots.substack.com/p/wheres-the-shovelware...

(see link in the article to a study showing developers thought AI gave them a 20% gain in productivity, but measuring this showed they instead had a 20% loss)

reissbaker 9/3/2025|||
Yeah $1k-1.5k seems absurdly high. The $200/month 20x variant of the Max plan covers an insane amount of usage, and the rate limits reset every five hours. Hard to imagine needing it so badly that you're blowing through that rate limit multiple times a day, every day... And if you are, I think switching to per-token payment would probably cost a lot more than $1k.
rolls-reus 9/3/2025||
The MAX plan is a consumer plan, it’s not available with Teams or Enterprise. They introduced a premium team plan ($150) with Claude code access but not sure how much usage that bundles.
kbuchanan 9/2/2025||
For me, working mostly in Planning Mode skips much of the initial misfires, and often leads to correct outcomes for the first edit.
mierz00 9/4/2025|
Recently I’ve been taking a step back and getting ChatGPT 5 to ask me questions to create a spec.

I refine that spec and then give that to planning mode and then go from there.

I’ve found if I jump straight into planning mode I miss some critical aspects of what ever it is I am building.

RomanPushkin 9/2/2025||
There is one thing I would highly recommend to anyone using Claude or any other agents: logging. I can't emphasize it more, if you have logging you can take the whole log file, dump it into AI, outline the problem and likely you're getting solution or would advance to the next step. Logging is everything.
makk 9/3/2025||
I don’t understand the use of MCP described in the post.

Claude code can access pretty much all those third party services in the shell, using curl or gh and so on. And in at least one case using MCP can cause trouble: the linear MCP server truncates long issues, in my experience, whereas curling the API does not.

What am I missing?

musbemus 9/4/2025|
You're exactly right. To be honest, in pretty much every case I've seen, indicating usage of a read-only resource directly in the prompt always outperforms using the MCP for it. Should really only be using MCP if you need MCP-specific functionality imo (elicitation, sampling)
jbs789 9/3/2025||
I often find that Claude introduces a level of complexity that is not necessary in my cases. I suspect this is a function of the training data (large repos or novel solutions). That said, I do sometimes find inspiration for new techniques in its answers.

I just haven't heard others express the same over-engineering problem and wonder if this is a general observation or only shows up b/c my requests are quite simple.

(I have found that prompting it for the simplest or most efficient solution seems to help - sometimes taking 20+ lines down to 2-3, often more understandable.)

P.S. I tend to work with data and a web app for processes related to a small business, while not a formally trained developer.

chamomeal 9/3/2025|
Seems like LLMs really suffer from the "eh I'll just write it myself" mindset. Yesterday on a react app using react-query (library to manage caching and re-fetching of data) claude code wanted to update the cache manually, instead of just using a bit of state that was already in scope in the exact same component!

For me, stuff like that is the same weird uncanny valley that you used to see in AI text, and see now in AI video. It just does such inhuman things. A senior developer would NEVER think to manually mutate the cache, because it's such desperate hack. A junior dev wouldn't even realize it's an option.

namesbc 9/2/2025||
Spending $1500 per-month is a crazy wasteful amount of money
the_hoffa 9/3/2025|
That's 18k a year, or about equal or cheaper than "outsourcing", minus the tax and legal ramifications.

I agree it's wasteful, but from a long-form view of what spending looks like (or at least should/used to look like). Those who see 1.5k/month as "saving" money typically only care about next quarter.

As the old adage goes: a thousand dollars saved this month is 100 thousand spent next year.

BobbyTables2 9/2/2025||
The author will be in upper management before they know it!
axus 9/2/2025||
I like his point about more objectivity and zero ego. You don't have to worry about hurting an AI's feelings or your own when you throw away code.
awesome_dude 9/2/2025|
But I still find myself needing (strongly) to let Claude know when it's made a breakthrough that would have been hard work on my own.
CharlesW 9/2/2025|||
Good creators tend to treat their tools with respect, and I can't imagine any reason we shouldn't feel gratitude toward our tools after a particularly satisfying session.

Also, there may be selfish reasons to do this as well: (1) "Should We Respect LLMs? A Cross-Lingual Study on the Influence of Prompt Politeness on LLM Performance" https://arxiv.org/abs/2402.14531 (2) "Three Things to Know About Prompting LLMs" https://sloanreview.mit.edu/article/three-things-to-know-abo...

groby_b 9/2/2025|||
Curious: Do you also laud your compiler for particularly good optimizations?
awesome_dude 9/2/2025||
There's a couple of things there

1. I don't see the output of the compiler, as in, all I get is an executable blob. It could be inspected, but I don't think that I ever have in my 20+ year career. Maybe I lie and I've rocked up with a Hex editor once or twice, out of pure curiousity, but I've never got past looking for strings that I recognise.

2. When I use Claude, I am using it to do things that I can do, by hand, myself. I am reviewing the code as I go along, and I know what I want it to do because it's what I would be writing myself if I didn't have Claude (or Gemini for that matter).

So, no, I have never congratulated the compiler (or interpreter, linker, assembler, or even the CPU).

Finally, I view the AI as a pairing partner, sometimes it's better than me, sometimes it's not, and I have to be "in the game" in order to make sure I don't end up with a vibe coded mess.

edit: This is from yesterday (Claude had just fixed a bug for me - all I did was paste the block of code that the bug was in, and say "x behaviour but getting y behaviour instead)

perfect, thanks

Edit You're welcome! That was a tricky bug - using rowCount instead of colCount in the index calculation is the kind of subtle error that can be really hard to spot. It's especially sneaky because row 0 worked correctly by accident, making it seem like the logic was mostly right. Glad we got it sorted out! Your Gaps redeal should now work properly with all the 2s (and other correctly placed cards) staying in their proper positions across all rows.

LtWorf 9/2/2025||
You've got to check the assembly, not the binary, for optimisations…
awesome_dude 9/2/2025||
Yeah - or I could just not care unless I have to (which, in the last 20 plus years, has been... let me think... oh, right... never)
nzach 9/3/2025|
One thing that I haven't seen a lot of people talk about is the relatively new model config "Opus Plan Mode: Use Opus 4.1 in plan mode, Sonnet 4 otherwise".

In my opinion this should be the default config. Increasing the quality of the plans gives you a much better experience using Claude Code.

More comments...