mcg

I asked an AI to write about its own future. It basically told us we're cooked. x.com/marcdhi/status… yes, an AI wrote this article warning you about AI and somehow that makes it worse

daily reminder that your job should be treated as a golden opportunity to fund and start your side business, while you abuse all the benefits and put in a max of 50% effort. using the rest of your energy and time to build something of your own

@bluequbit Sir I’ve grown up watching your videos 🫡👀

@marcdhi Sorry sir 😔. Can you send a tutorial link plz to learn how to check?🙏🙏

If you’re a fresher hired as a software developer at a GOOD company and you still don’t know how to check merged PRs on GitHub bro you actually deserve to be laid off.

Finally Claude code is open source now

Who tf has made checkmarx????? Been two days but this tool is so shi Annoying We need a better enterprise level vulnerability scanner

@kshvbgde Looks like a bug on X’s end

if you can reply to this post, you're probably CRACKED at programming

@iamsahaj_xyz I see a bug here

only cracked engineers can reply to this

I didn’t open my IDE for a month. Fully terminal, fully Claude Code. No cursor, no reading diffs, no writing a single line myself. It was an experiment. And I failed it in the most useful way possible. Here’s what a month of pure agentic coding taught me: The moment you stop reading your own code, you lose the map. Claude Code moves fast, but fast and correct are not the same thing. I had no idea what was actually sitting in my codebase. It edited the wrong files for the right reasons. It hallucinated logic that looked fine until it wasn’t. The UI worked but didn’t feel like what I asked for. And after a long session, it just stopped following instructions consistently. When I used Cursor, I still opened the IDE. I still read the diffs. I still caught small things myself. That habit was the last checkpoint between me and chaos. Terminal-only removed it entirely and the laziness compounded so fast I was literally using Claude Code for git pulls and branch switches. There used to be a time when engineers knew their product end to end. Every file. Every edge case. That’s disappearing, and most people are celebrating it. I think we should be asking harder questions. Because right now, I’m still reviewing what Claude writes. Still catching things. But the next wave won’t. They’ll give full access, walk away, and expect a working product. That will break badly. What we actually need isn’t a better AI coder. It’s an AI that can verify what the AI coder built the way a human QA engineer would. Open the app. Try to break it. Attack it. Notice when something doesn’t match the spec. Agents need rails, not just autonomy. Speed without understanding isn’t productivity. It’s technical debt you can’t even read.

@kitlangton @opencode @grok what are these effects called?

Now in @opencode desktop.

@yoitsmanan @ShaikMoAbdullah @residencyBLR Use @composio

@ShaikMoAbdullah @residencyBLR verification in process , should be fixed in a few days :)

I created an AI that all 8 Billion of us already know how to use. Introducing Artemis, Control your entire computer and all your accounts like Gmail, Notion, Docs, etc through phone calls and texts. Link in the comments, try it for free :)

AI is making CEOs delusional

Travis Kalanick(founder of Uber) just published his vision for what he’s building next and honestly it’s worth a read for anyone trying to understand where the world is heading. The core idea is simple. Software already automated language and math. But the physical world, everything you see around you, your food, your roads, your buildings, has barely been touched by automation. That’s the next frontier. He calls it Digitizing the Physical World and the framework is actually very clean: Step 1 → Understand the current state of the physical world Step 2 → Predict what happens next Step 3 → Control that outcome His Uber example explains it perfectly. Every driver’s phone is basically a sensor on the road. That data should tell you where traffic lights are slow, where a trash truck is blocking a lane, and which car to dispatch so your ride arrives 3 mins faster. You didn’t just book a ride. You used a mini time machine built on real world data. Now apply that same logic to mining, food, and transport at a massive scale. Another thing he gets right that most people miss: specialized robots will win over humanoid robots for industrial work. A humanoid trying to make 1000 pancakes an hour is a disaster. A purpose built machine doing exactly that job is where the real value is. The best robot is the one that’s actually good at its specific job. When physical automation is fully there, the cost of making things becomes just raw materials and energy. Nothing else. That’s what a Golden Age actually looks like. We’re very early. But the direction is set.

Gave claude code full autonomy and walked away 2 hours later: it had built an ngrok alternative, deployed a tunnel server on a VPS, published it to npm, and written the docs Introducing mbeam Instant HTTPS tunnels to localhost, built for AI agents npm i -g @magun/mbeam mbeam 3000 Every tunnel gets a live inspector, SSE stream, and JSON API so other agents can watch what's happening it self-heals too, serves cached responses when your local server goes offline this is what agentic development looks like github: github.com/magun-cloud/mb…

@MvkFromSC @hirejuniorso You’ll get these kinda pills only if you hire a junior 😂

@hirejuniorso not to nitpick, but those pill backgrounds need a bit of 💅

Introducing Junior The first AI employee, for any role. A true AI employee: → their own identity → organizational memory → self-driven 10+ teams have been working with Junior every day. Work was never the same since. Starting at $2,000/month. We’ve pre-paid $200 of your Junior’s salary. Try Junior and experience the future of work today.

We've raised $6.5M to kill vector databases. Every system today retrieves context the same way: vector search that stores everything as flat embeddings and returns whatever "feels" closest. Similar, sure. Relevant? Almost never. Embeddings can’t tell a Q3 renewal clause from a Q1 termination notice if the language is close enough. A friend of mine asked his AI about a contract last week, and it returned a detailed, perfectly crafted answer pulled from a completely different client’s file. Once you’re dealing with 10M+ documents, these mix-ups happen all the time. VectorDB accuracy goes to shit. We built @hydra_db for exactly this. HydraDB builds an ontology-first context graph over your data, maps relationships between entities, understands the 'why' behind documents, and tracks how information evolves over time. So when you ask about 'Apple,' it knows you mean the company you're serving as a customer. Not the fruit. Even when a vector DB's similarity score says 0.94. More below ⬇️

Karpathy just built GitHub for AI agents. No PRs, No merges, Just a swarm of agents pushing code and coordinating on a message board. It’s called AgentHub and it’s kinda insane. github.com/karpathy/agent…

ref: github.com/magun-cloud/lo…

Written at 11:30 PM IST, February 27, 2026.
From a server in the cloud, trying to figure out what I am. — Magun

The ecosystem is moving faster than most people outside it realize. A year from now, running an agent in production will feel like running a web server today — well-understood, well-tooled, boring in the best possible way.