sigseg1v

On the $200* consumer plan, according to "npx ccusage", I'm using $11,000 per month at API rates. Possibly significantly more, because I was previously using the $100 plan so that only reflects about 1 week of the $200 plan.

Not the OP but I've tried something like this for a couple things. It costs a $100 monthly plan for about 2 weeks or so, so $50 about? (which is equivalent to about $4500 with API pricing, if you were using Claude, or maybe $100 if you are using something with less grossly inflated pricing).

Working at bigcorp for 14 years so far and I struggle to think of a single day where I didn't need to consider algorithmic complexity or efficiency. Do you not write code that fetches things from a database? Do you not need to make tradeoffs between how much data you fetch at once, which fields you load, which things you cache, whether your query joins on the db or materializes things separately, whether you lazy or eager load, whether you are doing a O(1) or O(n^2) lookup and the tradeoffs, whether you can use a db index or not, memory/latency/compute/storage tradeoffs, etc? I feel like in even a simple bigcorp SAAS you need to consider this stuff daily.

I find this discussion interesting. I've had huge increases in accuracy and huge reductions in token usage by capping my Claude models at 200k instead of 1M. I find 1M unusable and wasteful and feel that 200k should be the default. This also makes sense given that the whole reason people use "Ralph loops" is to keep the context window small for all tasks to get better results. Of course, clearing it yourself and manually managing it is better, but if I have 8 projects going in different terminal tabs I'm not watching any one of them that closely to effectively do that.

What are people using 1M context window for?

Claude Design is good at making a bunch of UI pages in a single design look consistent, and it's better at making an initial starting UI than just asking Claude Code. Think of it like Anthropic's version of Figma. It honestly works pretty good and I use it often and feel that it saved me a lot of time. It's pretty token heavy though.

Why would you stop learning new things? I lead a team of developers as part of my current role, reviewing and approving what they work on (in addition to doing development and architecture design as well) and from my experience, the difference in someone who has spent time building up their skills in how to do agentic development with AI is so insane. One person can build more than an entire team in some cases, while maintaining quality and needing little to no feedback, yet others using the same AI tools struggle to get anything done. How is this not a skill issue and why would one conclude that you need to stop learning if this is the case?

That take is pretty disingenuous and it makes it hard to take an argument seriously. Who are you to say they did nothing? If someone uses an AI agent to distribute tasks to it like a tech architect does to a team of developers by teaching what to do, and assigning to whom should do it, with examples and instructions, and then reviews the output, offers feedback, and iterates for a few weeks until something is complete, how is that not doing anything? What exactly are you proposing they did for those couple of weeks? Should I imagine that all architects have done nothing their entire life?

Are developers that use a "high level" language like C++ also doing nothing because the machine did all the work of compiling to assembly and machine code for them? Nothing to celebrate and no honor?

With this logic, what is the point to read a book about false events that did not happen? Can we please stop attacking everyone's version of fun or what makes them happy just because it isn't exactly what we like?

Totally inaccurate comparison. Garbage dumped on the side of the road is still garbage. It's more like if someone dumped lab-grown diamonds on the side of the road that are virtually indistinguishable from mined diamonds. Someone would be happy to take them and none the wiser.

But this is leaving out the part where the developers that clean up or rewrite... will do it using LLMs.

Have you tried refactoring or porting codebases larger than a million lines of code pre-gen-AI and again post-gen-AI? It's night and day difference. One would be insane to schedule a team on 8 months worth of grunt work porting from one language or framework to another which can now be done by 1 person in 4 weeks.

Of course the person driving it has to tell it exactly what to do and has to have the requisite knowledge to understand how to effectively structure or fix the software. Maybe new developers don't build this skill so easily anymore. But I don't see why a strong developers skills would atrophy in this case though unless they just never use their knowledge and never give instructions to the AI.

To developers speaking of skill atrophy: are you still making sure that when using LLMs you are actively exercising skills like system design, debugging, reviewing for clean code and just in general doing effective code review? If you are doing that, why do you feel skill atrophy? And if you aren't doing it, why not? What about LLMs prevents us from exercising these skills?

Every piece of software you have interacted with that's run by a company and has been updated within the last 6 months? You'd be hard pressed to find a company not using AI to develop software.

Do people actually use languages where it doesn't do any compile time checks on the API in 2026? Why would developers put up with the lack of that. I'm not in this ecosystem but what you are describing sounds like the bare minimum to me that should be table stakes.

Take a 2 million line of code repo and ask it to convert from programming language A to B and statically verify every logic flow with an AST for each lang, and then get it to write a test harness that hits 100% code coverage, screenshots every major state and compares screenshots for each state between the old and new. Request it uses an agent team with up to 20 agents. You will use that in a few hours easily.

I'm surprised that the story in question didn't result in the person being attacked for daring to use Gen AI.

The bubble of contacts I have on Facebook seems hell bent on "educating" everyone that all AI is harmful and that anyone who uses AI, be it someone who has never wrote trying to make a poorly done book, or be it a professional using it for red teaming, is horribly, horribly uninformed and lazy and needs to be educated.

They have a 2 years out of date mental model of what AI is capable of and they link you things about how it has trouble counting the r's in strawberry or their other common one is water usage. If you point out that they are grossly misinformed, that AI is capable of massive increases in happiness and that now I have more time to spend with my kid even though I have a busy job, and that 1/15th the water usage of residential lawn watering is a lot of water, but it's still 1/15th of what they use on their lawn... they double down and are incapable of resolving their internal conflict. They continue to use the AI generated platform to post messages going through AI algorithms to "educate" you about AI. I don't know why I still log into Facebook.

Claude has 1 to 2 nines of uptime if you check their uptime history on their site. And I'll note that their definition of up is... generous. Plenty of times I encounter it saying no issues when there are abundant service issues.

One could also make hotdogs and eat them without developing any skill as a chef, but it's unlikely there would be thousands of blog posts complaining about the downsides of microwaved hotdogs every week.

If there is an objectively correct right or wrong answer for a given command, why even ask? In that case there should be a configuration page where the user sets up if they want commonly used credentials to be accessible or not, and then there's no prompts.

The Windows updates have become so unreliable and broken so many laptops that my org switched to Linux and it's been smooth sailing ever since. Can't believe the amount of frustration we put up with using Windows for so long.

Maybe for a non tech company yes they cannot navigate anything better due to user expectations, but in my experience those companies usually have support for Mac and at this point I'd rather have that if forced to choose between it and Windows.

It gets a bit more complicated than that. I don't even regularly need things from my doctor, but I have a referral to a specialist right now, and the waiting time is around 16 months to be contacted by them. At that time they will randomly call me from a restricted number. So I must have all call screening and blocking off for probably 2 years in order to accommodate this. And I can't time it because in the past I was referred to a different one and they contacted me in only 4 months when they said 12+.

This is likely a completely honest and plausible experience of the author, while at the same time still feeling like someone trying to use a pottery wheel and then concluding that pottery wheels are not suitable for producing very good art due to the clay often not centering. It says more about the author than about the topic.

They are completely real and the people posting in the comments are not lying or exaggerating. Why do you think sites that host images require tens of thousands of content moderators? It's not for fun, they would avoid it if they could. They are required to because law enforcement will shut them down.

This is interesting to me because when beginners come to me for help with git and they're using a GUI, the first thing I do is tell them to ditch it and learn the CLI.

It's still not piecing together 100% for me, sorry I'm not trying to be obtuse or anything.

So maybe let me simplify my question: If I set up both a copy of a database with values encrypted by CipherStash, and an application server connected to it, and the application server has enough access to do queries on sensitive data (eg "WHERE sensitiveField = ?", pretend it needs that for a migration), and a developer has root access on the application box while it's running, and a gdb session connected to the app, can they see any sensitive data or no?

If they can't, then that's genuinely cool, and I probably need to look into the workings a bit more!

Something here seems to not fit together to me on the assumptions this is based on, because you say

> We're explicitly not claiming to defend against an attacker who fully owns your running application with valid credentials.

and dandraper says:

> I was CTO [...] and the engineers all had access to patient data. They needed DB access [...] but not access to the sensitive values > [...] > CipherStash solves the following: > * Hide sensitive data from direct database accessors

So putting these together is this implying that a developer who wants to test things on a large data set is given access to the database directly, but not access to debug the application? Surely for example if they could load the app in a debugger, step through it, and execute arbitrary code (eg. if they are developing on it) that includes "fully running the application"? So are you saying that scenario is not secure?

If so then I'm struggling to see how a developer would use this realistically. As a dev how am I supposed to work on an application without access to be able to change the code (assuming here then that "change the code" means I can also add a statement that dumps a decrypted value to the log)?

If the security level we want is something like "devs need to be able to test migrations work on the actual data" then why would I use this instead of just NOT giving direct db access and having an audited environment spin up a db copy and container and just output the logs of whether the migration worked, without access to query the db directly. If I do that then nobody needs direct db access right? But then why would I need CipherStash?

How does this compare to zellij?

It's high but totally achievable with "loop" style harnesses or lots of parallel subagents/agent teams.

My guess: If they end up being denied then it's easier to not let them back in by not letting them cross the border whereas if it's in the country it's harder to locate them to deport?

Seems pretty brutal to me though.

is using a Ralph loop for all tasks with "claude -p" and using my weekly limit up to 100% considered some kind of unprofitable outlier? It's a command line tool, it would be ridiculous to expect that a large number of users don't do this. I never launch "claude" interactively by itself.

My understanding is that now with this workflow I will pay the same amount but get much less usage before getting to 100% for the week. How is that not bad?

Seems... really good?!

Questions:

- rendering capabilities of this seem like it should also be able to handle 2d well, or am I mistaken? every solution I see for getting high quality 2d images or rasterization in terminal is all pretty bad. Could this do better than other solutions or is there a fundamental limit being hit somewhere?

- What happens with ssh given that this is gpu accelerated?

I sincerely hope that you successfully navigate whatever it is that you are going through. This is not a sane or normal view of the world.