Anthropic's Claude Code's full source code leaked.
-
Anthropic's Claude Code's full source code leaked. Claude is seen by many to be the best coding LLM on the market with Anthropic proudly stating that Claude Code itself is mostly written by the LLM.
Now this sounds good as long as nobody can see the code which is quite the trash fire. Detecting "code sentiment" via regular expressions, variable and functions names containing prompt parts trying to influence the bot, a completely intransparent mess of a control flow that makes actual maintenance and debugging functionally impossible and the prompts ... of the prompts. All the begging and pleading to the chatbot not to do this or not to do that or please to do this.
It is fascinating but it is as far away from actual engineering as drunkenly pissing your name in the snow. Dunno what you call the people prompting software at Anthropic but "engineer" is not it.
Now it is fun to look at the currently hyped product striped bare and showing its pathetic quality but that is the future of software if we let those companies continue to undermine every good practice software engineering has tried establishing.
The software we have to use will be bad, insecure, unmaintainable, expensive with nobody having the skills or resources to build something better. As I wrote a few months ago: LLM based software production is equivalent to saying that fast fashion should be the only way to produce clothing. A tragic degeneration of the quality of the artefacts we rely on build for maximum profit on the backs of people in countries from the global majority.
-
@tante I’m very glad you posted this to LinkedIn so I could repost it there. It’s like a zombie apocalypse over there, but instead of “braiiiins” they’re all drooling “aiiiiiii”.
@aegir I am basically just trolling on LinkedIn.
-
Anthropic's Claude Code's full source code leaked. Claude is seen by many to be the best coding LLM on the market with Anthropic proudly stating that Claude Code itself is mostly written by the LLM.
Now this sounds good as long as nobody can see the code which is quite the trash fire. Detecting "code sentiment" via regular expressions, variable and functions names containing prompt parts trying to influence the bot, a completely intransparent mess of a control flow that makes actual maintenance and debugging functionally impossible and the prompts ... of the prompts. All the begging and pleading to the chatbot not to do this or not to do that or please to do this.
It is fascinating but it is as far away from actual engineering as drunkenly pissing your name in the snow. Dunno what you call the people prompting software at Anthropic but "engineer" is not it.
Now it is fun to look at the currently hyped product striped bare and showing its pathetic quality but that is the future of software if we let those companies continue to undermine every good practice software engineering has tried establishing.
The software we have to use will be bad, insecure, unmaintainable, expensive with nobody having the skills or resources to build something better. As I wrote a few months ago: LLM based software production is equivalent to saying that fast fashion should be the only way to produce clothing. A tragic degeneration of the quality of the artefacts we rely on build for maximum profit on the backs of people in countries from the global majority.
@tante The difference from fast fashion is that low quality clothing doesn't make decisions about who to bomb, who to put in prison, who gets benefits, how a bridge or skyscraper gets built, what medical treatment someone gets, whether their stalker or abusive partner can search their phone, etc.
Low quality software is fundamentally a threat to human life and safety.
-
@hopeless@mas.to @tante@tldr.nettime.org yes, as I read this, I found myself questioning whether the author has ever even used coding agents before. They are very competent, and I am working on a project for the fediverse using a variety of them. I have a programming/engineering background though, so I can design the architecture for them to implement and just use them as basically a super charged search engine/code monkey. I do also plan on full code reviews and outside collaboration so we can sanity check the results, but from what I have seen so far, the results it can achieve with thoughtful guidance is really amazing.
Some people are experiencing AI as a very unpleasant thing... they want it to just go away. If Anthropic's code leaked and it was super awesome, it would in no way improve their opinion. The topic of the next meeting of Anti AI Anonymous would simply be something else negative.
If you tell them it's good for coding assist (as it is) it's like telling them Cthulhu is good at coding assist. It's important to keep them in touch with reality. But it's not going to solve it.
-
@tante genuine question: How is this different from furniture from Ikea? Shoddy quality, probably good enough for most people while most people never have the ability to maintain or repair this furniture and yet more people have access to it than ever before?
@jmovs @tante But it's as people are trying building airplanes, medicinal devices or nuclear reactors using IKEA parts only. And people shrug and are OK with that. Because some lying Techbro promises AGI with just a few billion dollars more.
And it's clear where this will end: with LLM vibecoded voodoo programming deciding over life & death, running our critical infrastructure (probably into the ground) and guessing which apartment block to bomb first in some dictator's private war.
-
@aegir I am basically just trolling on LinkedIn.
@tante They deserve it. I’m sure it’ll give them “a teachable moment about B2B sales” or something.
-
Anthropic's Claude Code's full source code leaked. Claude is seen by many to be the best coding LLM on the market with Anthropic proudly stating that Claude Code itself is mostly written by the LLM.
Now this sounds good as long as nobody can see the code which is quite the trash fire. Detecting "code sentiment" via regular expressions, variable and functions names containing prompt parts trying to influence the bot, a completely intransparent mess of a control flow that makes actual maintenance and debugging functionally impossible and the prompts ... of the prompts. All the begging and pleading to the chatbot not to do this or not to do that or please to do this.
It is fascinating but it is as far away from actual engineering as drunkenly pissing your name in the snow. Dunno what you call the people prompting software at Anthropic but "engineer" is not it.
Now it is fun to look at the currently hyped product striped bare and showing its pathetic quality but that is the future of software if we let those companies continue to undermine every good practice software engineering has tried establishing.
The software we have to use will be bad, insecure, unmaintainable, expensive with nobody having the skills or resources to build something better. As I wrote a few months ago: LLM based software production is equivalent to saying that fast fashion should be the only way to produce clothing. A tragic degeneration of the quality of the artefacts we rely on build for maximum profit on the backs of people in countries from the global majority.
@tante Whenever I buy fast fashion, I buy it expecting it to fall apart because I'm broke. When it *doesn't* it's an unexpected surprise, a production glitch.
A future where youll be saying "Oh wow, my banking app/age verification app/bluetooth-integrated blood sugar monitor app *didn't* shit itself and delete all my saved data/log me out of my accounts/publicly leak my personal data to the internet today!" is... not quite the impressive feat that OpenAI & Co think it is. 🥲
-
Anthropic's Claude Code's full source code leaked. Claude is seen by many to be the best coding LLM on the market with Anthropic proudly stating that Claude Code itself is mostly written by the LLM.
Now this sounds good as long as nobody can see the code which is quite the trash fire. Detecting "code sentiment" via regular expressions, variable and functions names containing prompt parts trying to influence the bot, a completely intransparent mess of a control flow that makes actual maintenance and debugging functionally impossible and the prompts ... of the prompts. All the begging and pleading to the chatbot not to do this or not to do that or please to do this.
It is fascinating but it is as far away from actual engineering as drunkenly pissing your name in the snow. Dunno what you call the people prompting software at Anthropic but "engineer" is not it.
Now it is fun to look at the currently hyped product striped bare and showing its pathetic quality but that is the future of software if we let those companies continue to undermine every good practice software engineering has tried establishing.
The software we have to use will be bad, insecure, unmaintainable, expensive with nobody having the skills or resources to build something better. As I wrote a few months ago: LLM based software production is equivalent to saying that fast fashion should be the only way to produce clothing. A tragic degeneration of the quality of the artefacts we rely on build for maximum profit on the backs of people in countries from the global majority.
@tante
although, any coincidence with the weapons argument?Now I wonder where can the Claude code be seen? Tor somewhere?
some background:
https://www.theguardian.com/technology/2026/apr/01/anthropic-claudes-code-leaks-ai -
@tante
although, any coincidence with the weapons argument?Now I wonder where can the Claude code be seen? Tor somewhere?
some background:
https://www.theguardian.com/technology/2026/apr/01/anthropic-claudes-code-leaks-ai
