Luka
548 posts

Luka
@apisurfer
🇪🇺 Software engineer ⬩Web⬩AI⬩Apps⬩Automation
Croatia, Europe Katılım Mart 2023
568 Takip Edilen68 Takipçiler

@lastgoodhandle @mhilmi @boristane I hope you have Claude Max plan at a minimum, this is serious business!
English

@boristane I have some ideas and plan on tackling this. Seems like a foundational thing in the era of AI and once devs move up a ladder. What are your biggest friction points?
English

But the fact that you start debugging your vibe coded app only when it hits real production issues is that additional flavor nobody mentions, and you get to experience it for longer since you don't understand the codebase. What a trhill!
Shikhar@shekhu04
Before LLMs: - Coding :3 hours - Debugging: 1 hour After LLMs : - Coding : 3 minutes - Debugging : 1 week
English

@housecor Same here. There's only few scenarios where I run multiple agents at once.
Mostly I do it real-time, and do due dilligence on the output of the core pieces, a bit less so on the leaf nodes. But just letting agents roam and do whatever they want accumulates debt very fast
English

I see many people running multiple agents simultaneously.
I avoid it.
Instead, here’s what I typically do while my *one* agent is running:
- Review the code it generates in real time
- Watch its terminal output and course correct
- Do code reviews for teammates
- Reply to email / Slack / Teams
- Plan my next steps
- Pair with a dev on a different task
- Pee
So, for me, running a second parallel agent is a last resort.
English

@anitakirkovska Rarely happens if you bend it and invert the control of who owns the prompts, context, and what actually works on implementation
English

@theandreboso @forgebitz They really don’t need to be, everything is upside down now, everything is solved
English

@forgebitz I’m still waiting for them to fix this insane limit issue, I have to say they’re not very good at communication!
English

@Perpetualmaniac Bug? It’s a feature. If you look it through the Anthropic’ lens
English

Anthropic is committing suicide right now.
Everyone is exhausting their AI credits in hours. I just got back from vacation and exhausted nearly all my credits in 3 hours.
This bug has been up for over 3 days and no fix has been issued. No AI credits refunded.
reddit.com/r/Anthropic/co…
github.com/anthropics/cla…

English
Luka retweetledi

🚨 CRITICAL: Active supply chain attack on axios -- one of npm's most depended-on packages.
The latest axios@1.14.1 now pulls in plain-crypto-js@4.2.1, a package that did not exist before today. This is a live compromise.
This is textbook supply chain installer malware. axios has 100M+ weekly downloads. Every npm install pulling the latest version is potentially compromised right now.
Socket AI analysis confirms this is malware. plain-crypto-js is an obfuscated dropper/loader that:
• Deobfuscates embedded payloads and operational strings at runtime
• Dynamically loads fs, os, and execSync to evade static analysis
• Executes decoded shell commands
• Stages and copies payload files into OS temp and Windows ProgramData directories
• Deletes and renames artifacts post-execution to destroy forensic evidence
If you use axios, pin your version immediately and audit your lockfiles. Do not upgrade.
English

@cramforce Doing the same thing with claude and zed. But I miss just a couple of autosuggestion features that would make this an absolute go to combo
English

A while ago I got so frustrated with cursor-the-editor that I churned back to VSCode.
Now that agents write most of the code I don't want fancy autocomplete second-guessing what I want to do. I literally only go to the editor when I know exactly what to do.
Turns out VSCode was even worse at this. It constantly refuses to save files. Like it has one job. And this is for trivial edits like package.json or .gitignore.
So, I moved on again. Giving zed a try.
English












