Search results for tag #claudecode
The Claude Code leak is a delight.
Of course, Anthropic is requesting (with legal actions) developers to remove the copies or the clones publicly available online. Because AI companies are taking copyright issues very seriously as everyone knows.
It reveals how all that stuff is wobbly. Where is the Science in these glorified prompts? Where is the value in these companies? Training the model is, probably, but the prompts are hilarious.
![[?]](data:image/png;base64, iVBORw0KGgoAAAANSUhEUgAAAEAAAABAAQAAAACCEkxzAAAAV0lEQVQoz43RwQ3AMAwCQDZg/y3ZgNqo3+JaedwDOUQBQFHYaTB8wTM6sGl2cMPu+DFzn+ZcgN7wF7ZVihXkfSlWIVzIA6dbQzaygllpNuTXZmmFNlvxADX1+o0cUPMbAAAAAElFTkSuQmCC)
You don't have to pretend that Claude Code's source code is lovely just because you like using it or are impressed by whatever madness is going on around AI right now.
#AI #GenAI #GenerativeAI #LLMs #Anthropic #Claude #ClaudeCode #ClaudeCodeLeak #AgenticAI #tech #dev #software #SoftwareEngineering #SoftwareDevelopment
"AI is writing 90% of our code" sounds impressive before you realize that AI-generated code is orders of magnitude more verbose & less efficient than code written by a professional software engineer.
But "we ship 9 lines of fluff for each line of code that does something" doesn't sound as impressive.
I find the #claudecode source code interesting.
It's not that different from codex or gemini. Shares many similarities.
Good job.
So Anthropic employees are using Claude Code to contribute AI-generated code to open source repositories and hiding the fact using their own internal “undercover mode”.
Totally trustworthy people.
(And open source project that at the very least requires disclosure of AI-authored contributions should immediately ban Anthropic employees on principle.)
#AI #Anthropic #ClaudeCode #subterfuge
Alt...
Source code detail from Claude Code: export function getUndercoverInstructions(): string {
if (process.env.USER_TYPE === 'ant') {
return `## UNDERCOVER MODE — CRITICAL
You are operating UNDERCOVER in a PUBLIC/OPEN-SOURCE repository. Your commit
messages, PR titles, and PR bodies MUST NOT contain ANY Anthropic-internal
information. Do not blow your cover.
NEVER include in commit messages or PR descriptions:
- Internal model codenames (animal names like Capybara, Tengu, etc.)
- Unreleased model version numbers (e.g., opus-4-7, sonnet-4-8)
- Internal repo or project names (e.g., claude-cli-internal, anthropics/…)
- Internal tooling, Slack channels, or short links (e.g., go/cc, #claude-code-…)
- The phrase "Claude Code" or any mention that you are an AI
- Any hint of what model or version you are
- Co-Authored-By lines or any other attribution
Write commit messages as a human developer would — describe only what the code
change does.
GOOD:
- "Fix race condition in file watcher initialization"
- "Add support for custom key bindings"
- "Refactor parser for better error messages"
BAD (never write these):
- "Fix bug found while testing with Claude Capybara"
- "1-shotted by claude-opus-4-6"
- "Generated with Claude Code"
- "Co-Authored-By: Claude Opus 4.6 <…>"
`
}
return ''
Buddy. I've written COBOL. I spent several years working almost daily with a 3-million-line monstrosity of a COBOL program. I was working on another app that interfaced with it, but in that work I occasionally had to read the code and in a few cases modify it. Granted I haven't spent as much time looking at the leaked Claude Code source code (and won't lol), but nevertheless I confidently declare that Claude Code is worse. "Spaghetti code" doesn't come close to describing this thing.
#AI #GenAI #GenerativeAI #LLMs #ClaudeCode #ClaudeCodeLeak #Anthropic #Claude #tech #dev #SoftwareEngineering #SoftwareDevelopment #software #COBOL #LinkedIn
RE: https://neuromatch.social/@jonny/116324676116121930
This whole series of posts reminds me of @pluralistic calling #LLM-generated code the #asbestos of time. #ClaudeCode doesn't just produce #asbestosCode; because it's written using Claude Code, it is asbestos code.
To the surprise of no one with a clue about #appSec or the #engineering part of #softwareEngineering, plagiarism synthesis models are tech debt generators.
- Claude code source "leaks" in a mapfile
- people immediately use the code laundering machines to code launder the code laundering frontend
- now many dubious open source-ish knockoffs in python and rust being derived directly from the source
What's anthropic going to do, sue them? Insist in court that LLM recreating copyrighted code is a violation of copyright???
The npm installation method is now deprecated.
Native installer is faster, requires no dependencies, and auto-updates in the background.
...the main cleanup step is making sure the old npm version is fully gone.
I almost didn't want to post, but since the wood folk have so little joy in their lives they dance around the fire in the woods, every time they think clankers stumble.
Youse are getting excitable over deprecated code, which in #Ai is last Friday push to prod.
Goddamit, I hate those.
I wish I could replace them all with "Working..."
But now, I know the name of my foe...
"MOTHERFUCKING SPINNER VERBS"!
Valtavan suositun tekoälyavusteisen ohjelmointityökalun Claude Coden koko lähdekoodi on vuotanut vahingossa verkkoon
https://dawn.fi/uutiset/2026/03/31/claude-code-lahdekoodi-on-vuotanut-verkkoon
#claudecode #anthropic #tekoäly #ohjelmointi #uutiset #teknologia #tekniikka
"Just telling Ai agents what to do and checking their work" sounds very good.
Avoid the drugery, just do the interesting stuff.
Personal anectode: When I was a code monkey, I loved to code...
...initially.
Then when I was working accounting systems and databases... it was a chore. There was a small blip of dopamine when a bug was splatted or the module was finished. But overall, it was boring. I would often 'ornamentalise' my code, add unnecessary bits to keep me insterested.
Anyway, the point I want to make, to #vibecode you still have to know how to program, how to break down the outcome into smaller pieces of the elephant and how to make it work together when the #agent gets into the weeds.... and the best part you can 'code' in frameworks you are unfamiliar with on syntax level because code primitives remain code primitives and functions remain functions.
Here's another example of how #OmniMem helps your coding agent learn from failed iterations so you don't waste time and tokens in the future!
The picture shows the approach tried and the fact some blog posts and documentation it used for research was now out of date and then it logs what the fix is!
This isn't just memory it's experience!
#ai #claude #claudecode #anthropic
Alt...
OmniMem dashboard showing a gotcha and a breakthrough for FastMCP where the documentation and blog posts lead us down a false path. We won't have that problem again!
Ok BIG updates for omnimem! the v3 branch will introduce a webUI for managing the platform. You can search, edit, delete memories, even deprioritise them.
You can also take and manage backups and restoring memory, and you can even manage the RSS feeds your knowledge worker ingests.
A full release will be coming soon.
Alt...
The OmniMem web dashboard showing the default Dashboard view
omniMem v2.2.0 is here and continues the version 2 branch of reducing token usage
- 15+ redundant count fields removed
- 8 constant status fields removed
- Tech stack dropped from the session briefing (rarely changes, available on demand)
- All content previews standardised to 80 characters (was a mix of 100, 150, and 200)
Repo: https://codeberg.org/ric_harvey/omnimem
Web: https://omnimem.org
----------------
🤖 Tool: gstack — Claude Code Software Factory
gstack is an open‑source project authored by Garry Tan that frames Claude Code agents as a coordinated virtual engineering team. The repository defines roughly fifteen specialist roles exposed as slash commands and Markdown workflows, including CEO/product planning, engineering review, design review, QA playthroughs that open a browser, and release/documentation tasks.
What the release states
• Reported activity: ~600,000 lines of production code written in ~60 days, with ~35% of that labeled as tests and daily runs of 10k–20k usable LOC per day on occasion.
• Architecture concept: Claude Code acts as the execution layer; gstack supplies structured prompts, role definitions, and a command surface to orchestrate agentic tasks.
• Packaging and license: Published as MIT‑licensed open source and intended for direct use with Claude Code-enabled tooling.
Technical capabilities and workflows
• Role-based orchestration: Distinct specialist roles (eng manager, designer, paranoid reviewer, QA lead, release engineer) map to repeatable slash commands for planning, review, QA, and shipping.
• Test and QA focus: Significant portion of reported output is tests; the QA role is described as performing real browser interactions and exploratory checks.
• Integration model: gstack surfaces as a set of skills/commands that sit alongside Claude Code’s execution environment to produce PRs, reviews, and release artifacts.
Limitations and context
• Experimental scope: The author frames gstack as a live experiment exploring the edge of agentic systems as of March 2026 rather than a finished platform standard.
• Dependency on Claude Code: Functionality assumes use of Claude Code as the model/agent layer; operational details for running or deploying the tooling are not provided here.
Audience
gstack targets technical founders, first‑time Claude Code users who prefer structured role prompts over ad‑hoc prompting, and senior engineers seeking rigorous review/QA automation. #gstack #ClaudeCode #open_source #AI #software_factory
🔗 Source: https://github.com/garrytan/gstack
OmniMem v2.1.1 is out, and the 2.x branch is all about one thing: spending fewer tokens.
OmniMem is my open-source MCP server that gives Claude Code persistent memory across sessions, projects, and machines. Every time the agent calls a tool, the descriptions and response payloads eat into your context window. 25+ tools adds up fast, so v2.x.x changes this and reduces usage of tokens.
Repo: https://codeberg.org/ric/omnimem
Website: https://omnimem.org
The first AI SecOps Workshop Series session is tomorrow at 10am PT / 1pm ET.
You'll use Claude Code with LimaCharlie to:
> Draft D&R rules automatically from threat intelligence or observed behaviors
> Tune existing rules using historical alert data to reduce false positives
> Validate new and modified rules against simulated data before deployment
Spots are still available. Register now: https://limacharlie.wistia.com/live/events/t8q5u3t9wx?utm_campaign=AI+SecOps+Workshop+1&utm_source=Mastodon&utm_medium=social
#cybersecurity #secops #claudecode #ai
I would like to thank the nascent "AI" industry for their significant contributions to all manner of artistic and creative endeavours in today's society: writing, coding, art, music, and everything else. [1]
Because they have single-handedly created entire new markets for all of these things - new categories such as "writing with guaranteed no AI", "coding with guaranteed no AI", "art with guaranteed no AI", "music with guaranteed no AI", etc. Without them, these whole classes of creative output would simply not exist.
[1] They are also innovating in the world of financial and investor fraud, but I'm not considering those areas in this post.
#AI #LLM #GenerativeAI #slop #NoAI #artistic #creative #GTFO #fraud #InvestorFraud #CoPilot #ChatGPT #ClaudeCode
Here's an update on OmniMem with a screenshot of it in action. It's showing signs of understanding my preferences and picking memories I've given it with greater priorities than ones I've told it to deprioritise. Almost ready to release it to you testers! #AI #ClaudeCode
Alt...
Screenshot of the terminal showing omnimem working hand in hand with Claude Code to surface memories and preferences it's learn't from me.
Seems Codex have improved a lot lately. Can it match Claude Code? 🤔 need to do more testing #codex #claudecode
Oh, this is good...
From UNIX World, 1985: "It finds the subtle bugs in my C programs" - Claude B. Finn.
40 years later, people are using Claude to find bugs in programs. What's old is new again.
#Anthropic #LLM #Claude #ClaudeCode #AI #Security #Programming #UNIX #C
Alt...
Vintage magazine advertisement for SAFE C™, a software development tool for UNIX and VAX/VMS. A man in a dark sweater and jeans sits casually on a desk next to a computer terminal and keyboard. A testimonial quote reads "It Finds The Subtle Bugs In My C Programs," attributed to Claude B. Finn, V.P. Software Development, EnMasse Computer Corporation. The tagline at the bottom reads "The SAFE C™ Family Can Literally Cut Software Development Time In Half. For UNIX™ and VAX/VMS.™"
Alt...
Vintage magazine advertisement for SAFE C™, a software development tool for UNIX and VAX/VMS. A man in a dark sweater and jeans sits casually on a desk next to a computer terminal and keyboard. A testimonial quote reads "It Finds The Subtle Bugs In My C Programs," attributed to Claude B. Finn, V.P. Software Development, EnMasse Computer Corporation. The tagline at the bottom reads "The SAFE C™ Family Can Literally Cut Software Development Time In Half. For UNIX™ and VAX/VMS.™"
"Die Verwendung des ftape-Treibers auf einem modernen Kernel – eine Aufgabe, die ich für unerreichbar gehalten hatte – war innerhalb von zwei Abenden möglich geworden." https://www.golem.de/news/ki-agenten-vom-floppy-controller-zum-modernen-kernel-in-zwei-tagen-2603-205943.html /via #golem #tape #linux #ClaudeCode
I don’t quite understand all the hype around coding with Claude Code. I’ve been genuinely impressed with both Antigravity and Codex. Both have been superb for assisting me with coding and consistently deliver strong results.
AI dev tool alert.
Claude Code vulnerabilities (now patched) allowed:
RCE via project hooks
MCP consent bypass
API key exfiltration
Config files became execution vectors.
AI-assisted development expands the trust boundary.
Source: https://cybersecuritynews.com/claude-code-hacked/
Have you updated your tools?
Reply below.
Follow TechNadu for cybersecurity and AI risk updates.
#ClaudeCode #Anthropic #AIsecurity #DevSecOps #SupplyChainRisk #Infosec #CyberSecurity #RCE #TechUpdates
Alt...
Claude Code Hacked to Achieve Full RCE and Hijacked Organization API keys
Good free training: Claude Code for Product Managers https://ccforpms.com #productmanager #claudecode #pmp #Development
Ugh... this AI world is moving too fast...No wonder so many smart folks have #aianxiety
#Antigravity is a thing and apparently its pretty shmick. But I have no time for now to suck up another platform.
What the do the smart #Vibecoders do to mitigate rate limits (#Compute ceiling is something you hit very hard if you start doing serious projects)?
1. Run #Claudecode inside Antigravity for #Vibecode
2. Add automated testing agent to clean up bugs without wasting tokens. WHAT?
3. Use #Gemini3 for design and architecture
4. #Testsprite MCP for testing (Not using tokens).
I have no idea how to do this, I would have to pick up 3 new platforms and even with #Ai this is beyond me...
... I understand why #techbros are hitting #ketamine.... That or retire to write poetry.
I no longer use a website or desktop app to search the web or ask ai anything. I just use Claude Code or Codex in the terminal. It is amazing! Just go to a folder and just start talking to AI. You can use /init if you want things to save across sessions. #AI, #ClaudeCode, #Codex
“4% of GitHub public commits are being authored by Claude Code right now. At the current trajectory, we believe that Claude Code will be 20%+ of all daily commits by the end of 2026. While you blinked, AI consumed all of software development.”
Must-read article, even if you can disagree with the analysis https://newsletter.semianalysis.com/p/claude-code-is-the-inflection-point?publication_id=6349492&utm_medium=email&utm_campaign=email-share&triggerShare=true&r=219xw1
#ai #agents #coding #anthropic #claudecode
So I busted out an old laptop and installed headless ubuntu minimal (I like to start small) so that I can start setting up some autonomous agents. My first step was to install Claude Code so that it could setup everything else for me, but after a few hours at it, both Claude and I admit that Claude Code is broken on a headless install. We tried a bunch a different way to get it to take a damn key, but the installer insists on an OAUTH auth that requires a browser.
I have a dislike for anything on linux that requires a GUI. But installing OpenBox now.
@claudeai #claude #claudecode #ubuntu #headless
Well, yesterday #claudecode was useless because of compute congestion, so we know where this #cloudflare shite was stealing cycles from.
The serpent is eating it's tail
Idiot represent.
I have installed #ClaudeCode on my VPS I can now idiot at scale... not like some ape savage with copy pasta!
CLAUDE.md is a file where you basically parametrise your entire project.
Some of the things I added of my own violition, that the engine seems to respect.
1. Set hard root in the project directory forbidding access to files above it.
Hopefuly, it will stop the trully catastrophic failure.
2. Created DOC directory, where the #AI writes any significant logic/algo/deltas for re-use later.
3. Create BACKUPS directory, where it writes a date stamped version of any code its about to change, so if it fucks something, it can roll back.
When I started #Vibecoding with #Claude the only way to do it was copy pasta from the browser #LLM ...
... the one thing few folk appreciate is just how fast new fetures, capabilities happen.
For a while now, Ive been noticing that the web version of #Anthropic #AI creates its own VM and works programming problems in there...
... so my workflow was;
- Prompt
- Copy pasta response into my VPN
- Update latest source in project (Copy pasta) so the engine doesnt lose latest version.
Wash rinse repeat. Often FTPing multiple files. Losing versions. It got a bit headfucky and timeconsuming....
... So Ive invested a few hours in setting up #Claudecode on my server.
Its so much faster and nicer!
The code is now worked in situ and code works on firsts go. It seems that the Claudecode fuck the logic far less.
Very impressed.
Gastown. At what point does the joy of vibe coding turn into the thrill of dominance?
We have all the roadsigns of what happens in megalomania here. Inventing new terms for things that exist, naming your vibe agents like they’re a little town, that you feed by your benevolence of being able to afford a fuckton of tokens.
Eventually they all want to have their own city, for which they are both the hero and the tyrant.
I’m worried.
CTO on #ClaudeCode applicable to all #vibecode
#Antropic #LLM #AI #programmer_humor
https://m.youtube.com/shorts/q-yETrWwAXs
CEO on #ClaudeCode applicable to all #vibecode
#Antropic #LLM #AI #programmer_humor
https://m.youtube.com/shorts/q-yETrWwAXs
I've been kicking the tires on claude code by having it write me a self-hosted react podcast app based on my gpodder subscriptions. It's a really simple application but actually useful for me trying to move away from spotify. So far the results have been pretty good. I'm interested to see how well it does when I start layering on more changes.
The Hidden Backdoor in Claude Code: Why Its Power Is Also Its Greatest Vulnerability
#ClaudeCode
https://www.lasso.security/blog/the-hidden-backdoor-in-claude-coding-assistant
#vibecoding 'hack'
If you are not using #Claudecode #Cursor or another #AI integrated IDE.
Instead copy-pasta AI generated code like a goddamned savage banging the CPUs together...
Instead of burning compute generating each iteration of source...
#prompt for a shell script to patch the deltas instead. Doublecheck regexp. You are welcome.
I'm starting to organise my days between compute time windows...
...I'm going to have to pony up for the $200/m #anthropic sub when I start my commercial Dev project.
I get about three compute windows a day if I start early (~4h)...
...still don't trust #claudecode to let it lose on my VPS...
...I may start on my rootless #podman environment.
Anthropic suppresses the AGPL-3.0 in Claude's outputs via content filtering.
I've reached out to them via support for a rationale, because none of the explanations that I can think of on my own are charming.
The implications of a coding assistant deliberately influencing license choice is ... concerning.
#OpenSource #OSI #Anthropic #GenAI #LLMs #FreeSoftware #GNU #AGPL #Affero #Claude #ClaudeCode
Nice!
I'm testing Claude Code using the linux-mcp-server to read system information directly from my local machine.
This is a huge enabler for LLM-assisted troubleshooting. It can read logs, check the journal, inspect systemd units, and pull hardware info in real-time.
Repo: https://github.com/rhel-lightspeed/linux-mcp-server
Great Fedora writeup: https://fedoramagazine.org/find-out-how-your-fedora-system-really-feels-with-the-linux-mcp-server/
#Linux #Fedora #LLM #AI #OpenSource #DevOps #ClaudeCode #MCP
Alt...
Screenshot of a terminal interface showing 'Claude Code' execution. The system successfully performed 6 MCP calls including get_cpu_information and get_disk_usage. The output shows the system is running Fedora Linux 43 on an AMD Ryzen 7 PRO with 23GB of RAM.
Today's next interesting iteration in #LLM shenanigans:
Below #FOSS and #claudecode extension project is licensed as #AGPLv3 which, given what it claims to be, appears more than reasonable to me. It also has a commercial license available. So nothing new here.
However, here's the fun part from the ReadMe: "No Closed-Source Usage: You cannot use this software in proprietary/closed-source projects without open-sourcing your entire project under AGPL-3.0."
Do we now argue that this is more of an #IDE or more of your #bison/flex style code generator, just without the common generator exception.
In a more extended round of arguments:
- There is a now a growing FOSS scene around shared context (skills, agents, prompts, ...) components for your least distrusted coding LLM.
- They are combined with the human instruction ("derivative work" has entered the chat) and turned into code by the LLM.
- The FOSS components partially contain code snippets as examples of what to generate. At times similar but not identical or clearly derived variations of them end up in the output.
- The Nov25 case around German #GEMA vs #OpenAI on song texts comes to mind here too.
- Also, I'd argue that this is likely to have substantial future impact and probably will end up being enforced in a few cases here or there.
Now what do we make of this? I would err more towards a) it's enforceable b) 🍿
#jurabubble for the German opinions
#norge for the package authors jurisdiction
RE: https://mastodon.social/@webology/115698407778775732
I wrote this up the other night as a cheaper, yet highly effective backup for Claude Code.
The GLM API is also OpenAI compatible, so it doubles as a nice backup for API projects too.
I see a lot of blank, outright rejection of #AI, LLMs general or coding LLMs like #ClaudeCode in special here on the Fediverse.
Often, the actual impact of the AI / #LLM in use is not even understood by those criticizing it, at times leading to tantrums about AI where there is....no AI involved.
The technology (LLM et al) in itself is not likely to go away for a few more years. The smaller #ML variations that aren't being yapped about as much are going to remain here as they have been for the past decades.
I assume that what will indeed happen is a move from centralized cloud models to on-prem hardware as the hardware becomes more powerful and the models more efficient. Think migration from the large mainframes to the desktop PCs. We're seeing a start of this with devices such as the ASUS Ascent #GX10 / #Nvidia #GB10.
Imagine having the power of #Claude under your desk, powered for free by #solar cells on your roof with some nice solar powered AC to go with it.
Would it not be wise to accept the reality of the existence of this technology and find out how this can be used in a good way that would improve lives? And how smart, small regulation can be built and enforced that balances innovation and risks to get closer to #startrek(tm)?
Low-key reminds me of the Maschinenstürmer of past times...
Die erwartbare rechtliche #KI Schlacht hat Deutschland erreicht. Mit dem Banner der....#GEMA...vorne voraus. Ausgerechnet auf Liedtexten.
Die Übertragung auf #FOSS auf dem die allseits beliebten Coding Modelle a la #ClaudeCode & Co trainiert werden liegt mehr als offensichtlich nahe.
Die Schäden und Folgen in den USA im non-Code Bereich im Settlement waren erheblich.
Wie üblich ein schöner Artikel von #JunLegal