I really like this idea as I find Claude's transparency frustrating. Claude code's killer features revolve around better tools to manage context and limits vs the desktop app (compact and the % remaining until auto compact), but it's not enough.
If I can offer any advice, it's that the high use of emojis in a project readme (at least for me) looks so unprofessional and makes me worry that a project was vibe -coded in the sense that the AI was possibly not babysat to the extent I think they should. That's just me, though
I got into software in a time where you would get sent to a mental institution when spotted using emojis in a code base. Times have changed.. I use emojis regularly because they help me organize context more visually. Code has now many emojis to keep me happy.
This code was written in pure vibe-coding style — mostly for fun.
I've got about 10 years of experience in IT, and even I fully agree:
a 1000-line main file like this one probably deserves to be locked away in a secure facility.
But hey — if it's stupid and it works, it ain't stupid.
Very neat! Is the limit on Pro really only 7k tokens? So less than 7k words? I feel like I get more out of that. It feels like that would blow up pretty quickly with an ongoing chat, but I never hit that.
Or is this a Claude Code specific limit? I haven't used Claude Code extensively yet.
This is great. I built a UI tool to run simultaneous Claude Code sessions (https://github.com/stravu/crystal) but by working on multiple features at once I hit the limits of my max account. Usually close to the reset, but it would be nice to know when it is time to take a break.
Kudos! I was just thinking about having. Laude write something like this for me. Not just within a project but the 5 different ones I have open at the same time. Too much of a good thing etc.
This is pretty cool.. I like the idea, could I make a semi-snarky feature request and could we add the estimated power consumed to produce results for each session ?
I really dislike when people don't see this. They try to cut 10 grams of CO2 per day while other industries (shipping, aviation, rails) produce hundreds of tons per day and even this transportation modes are less that 20% with most CO2 produced mainly being in energy production and used by industry.
It doesn't necessarily need to be actionable for now but at the moment there is an exponential growth in the datacenter power usages.
For now, sure it might be ridiculously minor, but when it starts to ramp up who's to say it wont be just a ridiculous amount of energy ? Maybe not even measure the CO2, but I would love to graph the increase of energy spent over time.
If you have some time, feel free to open a pull request — even just with a description or clarification of what's going on!
It all sounds super interesting, but I’m still trying to fully grasp the business logic behind it
Once I get the idea, I’ll be happy to jump in and implement it.
Maybe the few that were WFH prior to AI. I just calculated my carbon output for commuting and going to lunch daily versus WFH and using AI and it's wayyy less.
Let's all save earth, stay home and use AI.
Soon the AI companies will have different tiers of plan based on carbon output.
For $200 per month they’ll only use carbon-free power sources for your prompts. At $100 they’ll use nuclear, then $20 per month for coal, and then there will be a free tier where your prompts are powered by throwing baby seals into a furnace.
Thank you! For me, I wish it was just a simple command that outputs how much I have used out of my plan already.
Also I feel it’s a bummer that I need python when CC is a node.js tool.
Check directory ~/.claude/projects/*/*.jsonl ;) hah very interesting topic. I have a plan to write blog post about it (all conversations history with metadata inside json lines files) What do you think about this idea? Interesting?
Wow, I had no idea Claude Code had this kind of verbose logging turned on by default. Looking around I also found a cool tool for converting these logs to HTML format for easy viewing: https://github.com/daaain/claude-code-log
It's definitely interesting. Probably, one could also see how much of the context is unnecessary to optimise usage. I wonder if one could also detect whether cache was used.
Claude is cool, they focus on cool and useful things.
But man do I just want a way to quickly glance at my API credits and to just occasionally chat with a model from those credits without librechat of openwebui. Or set some limits, or see some usage metrics.
And please please use "forever auth" with passkeys or something, what is up with that auth email that just takes ages for a quick glance?! It always takes me 3 attempts to find what address I used at sign-up...
Oh and make it clearer why you have that API/credits system and a subscription, why is it so difficult to understand when you start using Claude that it's 2 different unrelated worlds?? First time I started my subscription I just couldn't figure out where the API section was, until I realized it just wasn't there.
I feel like I'm "holding it wrong", but please make it easier to hold it right then.
>And please please use "forever auth" with passkeys or something, what is up with that auth email that just takes ages for a quick glance?! It always takes me 3 attempts to find what address I used at sign-up...
As a person with the simple but brilliant technology of a freaking password manager, I LOATHE email to login/no-password websites. They are dreadful, we've somehow managed to come up with something worse in UX rather than move forward.
If you're working on a product that does this, or wants to do this, please please PLEASE reconsider, it's such a PITA for technical users and normies alike.
I used to think the same thing, and I still want a login/password alternative for me. But after seeing normies use online accounts and the trouble they have with password managers, I realized one of my friends had the most secure process: she would create extremely high-entropy passwords everywhere, but not remember them. Once she's logged out of her sessions after a couple of months, she uses the password forgotten link to generate another password, and so on. So her passwords are never stored anywhere, she's immune to many login stealing phishing attempts through genuine-looking fake websites as she can't enter the password, she doesn't have to deal with syncing the passwords between all her devices, and she doesn't have passwords on a post-it on her workstation. And she also doesn't get those annoying emails saying "your password is 6-months old, please change it or else!".
The email auth flow is a simplified and more efficient way to achieve the same outcome.
yeah, I find it annoying, but it's a simple way of making something secure, piggyback on something that already made a decent effort at trying to being secure enough.
Another reason to ditch email-to-login: It decreases active user count because login is such a chore. I only use those websites if I absolutely have to and there are no alternatives.
Yeah, it's the reason I eventually tuned out of Slack too (obviously I don't use it day to day). I used it for some things, some community stuff, every now and then.
So when I open it up on a new machine or after months, you have to go through that magic link bs multiple times! For all your accounts/channels! I did that 2 or 3 times and then I just stopped using it.
I don't even remember what I signed up for over the years, I know some of it was nice (like an LoRa IOT channel, an "AWS professionals" channel, I set something up for the LoRaWan network in the previous city I lived in... All just tuned out because of the login bs.
I don’t get why Slack can’t figure out how to help me migrate all my workspaces on a device I still have and am logged into in a new one in fewer than 3 steps per workspace. Sure, if I’ve lost all devices with the workspaces connected, maybe it needs to be that much PITA, but it’s a pretty common case to add/switch to a new machine or device and that process is obnoxiously poorly executed right now.
Anyone else have a bunch of fun with Claude Code's Github integration... until you realized it spends from your API wallet balance?
Oops, when I saw the /install-github-app command, I assumed that since I'm on a max plan and Claude Code in my terminal is free, then the Github integration would be free.
So I hooked it up to my repo and tagged @claude in everything. It was a lot of fun tagging it in backburner issues and seeing it solve issues I couldn't be bothered to do for years. Or just seeing what it would come up with on really low effort poorly explained issues.
Addressing those issues wasn't worth $0.50 apiece? That almost certainly translates into less than a _minute_ of fully loaded engineering cost from a human.
What's that link supposed to show? It shows my anthropic account and billing. My claude account and billing (with different payment details) is at https://claude.ai/settings/billing
I see what you mean. Anthropic's UI and system organization are always very confusing.
Your link and my link are just two views into the same underlying account. At your link, you can manage your Claude subscription. At my link, you can manage API keys and API credits.
But they both use the same underlying account (unless you registered a new account separately I guess).
With uv, is there a way to install Python and Node tools like this in a self-contained way so they appear to the system as a single executable?
I guess I don't want to duplicate Python/Node for every tool, but I also don't want it to be fragile. And this wants a Node CLI tool installed globally, which I've found breaks easily with changing versions.
If I can offer any advice, it's that the high use of emojis in a project readme (at least for me) looks so unprofessional and makes me worry that a project was vibe -coded in the sense that the AI was possibly not babysat to the extent I think they should. That's just me, though
But hey — if it's stupid and it works, it ain't stupid.
Or is this a Claude Code specific limit? I haven't used Claude Code extensively yet.
How many tokens used in a heavy vibe coding day?
Average US daily per capita is like 40kg CO2
A single flight from JFK to LAX produces around 20,000kg of CO2. Using the 8.3g value means a flight is equivalent to 2.41 billion tokens.
For now, sure it might be ridiculously minor, but when it starts to ramp up who's to say it wont be just a ridiculous amount of energy ? Maybe not even measure the CO2, but I would love to graph the increase of energy spent over time.
Repo is here if you're curious: https://github.com/Maciek-roboblog/Claude-Code-Usage-Monitor...
It produced a patch. Unfortunately it was for removing the emojis from the readme.
For $200 per month they’ll only use carbon-free power sources for your prompts. At $100 they’ll use nuclear, then $20 per month for coal, and then there will be a free tier where your prompts are powered by throwing baby seals into a furnace.
I'm of an age where I rarely watch short-form videos so would read the article but you'll miss an audience if that's all you produce.
https://github.com/Dwtexe/cursor-stats
But man do I just want a way to quickly glance at my API credits and to just occasionally chat with a model from those credits without librechat of openwebui. Or set some limits, or see some usage metrics.
And please please use "forever auth" with passkeys or something, what is up with that auth email that just takes ages for a quick glance?! It always takes me 3 attempts to find what address I used at sign-up...
Oh and make it clearer why you have that API/credits system and a subscription, why is it so difficult to understand when you start using Claude that it's 2 different unrelated worlds?? First time I started my subscription I just couldn't figure out where the API section was, until I realized it just wasn't there.
I feel like I'm "holding it wrong", but please make it easier to hold it right then.
As a person with the simple but brilliant technology of a freaking password manager, I LOATHE email to login/no-password websites. They are dreadful, we've somehow managed to come up with something worse in UX rather than move forward.
If you're working on a product that does this, or wants to do this, please please PLEASE reconsider, it's such a PITA for technical users and normies alike.
The email auth flow is a simplified and more efficient way to achieve the same outcome.
So when I open it up on a new machine or after months, you have to go through that magic link bs multiple times! For all your accounts/channels! I did that 2 or 3 times and then I just stopped using it.
I don't even remember what I signed up for over the years, I know some of it was nice (like an LoRa IOT channel, an "AWS professionals" channel, I set something up for the LoRaWan network in the previous city I lived in... All just tuned out because of the login bs.
Oops, when I saw the /install-github-app command, I assumed that since I'm on a max plan and Claude Code in my terminal is free, then the Github integration would be free.
So I hooked it up to my repo and tagged @claude in everything. It was a lot of fun tagging it in backburner issues and seeing it solve issues I couldn't be bothered to do for years. Or just seeing what it would come up with on really low effort poorly explained issues.
But not worth spending 50+ cents every time.
On another hand, you solved several years-old issues for under $50…which seems like a big win.
Currently in gray area but allowed as per anthropic comment on it - https://github.com/anthropics/claude-code-action/issues/4#is...
I'll try it out right now.
Isn't Claude and Anthropic API two separate platforms? How does it spend from your other account?
https://console.anthropic.com/settings/billing
Your link and my link are just two views into the same underlying account. At your link, you can manage your Claude subscription. At my link, you can manage API keys and API credits.
But they both use the same underlying account (unless you registered a new account separately I guess).
I guess I don't want to duplicate Python/Node for every tool, but I also don't want it to be fragile. And this wants a Node CLI tool installed globally, which I've found breaks easily with changing versions.
`uv tool install` doesn’t duplicate Python for every tool.