Arka Lal Chakravarty @arka_codes

@ClaudeDevs The unhandled object object error check that as well and fix it

@ClaudeDevs please have a look in this and fix this ASAP!

Andrew Hart @AndrewHart

2 months ago

Yesterday VS Code and Claude Code were working perfectly fine. Then it updated overnight. Now there's a new sidebar, I'm getting "MCP errors", Claude isn't responding, i'm being gaslit this is an "issue on your end". I feel like the guy in Her when his AI gets deleted.

3 1 7 1K 0

Yesterday VS Code and Claude Code were working perfectly fine. Then it updated overnight. Now there's a new sidebar, I'm getting "MCP errors", Claude isn't responding, i'm being gaslit this is an "issue on your end". I feel like the guy in Her when his AI gets deleted.

@AndrewHart I am facing the same issue since today morning

Hiring too early kills startups. Hiring too late stalls them. The real solution 👇 Before you hire: • Validate the problem • Ship a lean MVP • Watch real users • Fix what actually hurts Most startups don’t fail from lack of effort. They fail from premature scaling. Small teams win early because they learn faster. If you’re building right now, ask yourself: Are you solving problems — or adding people? That answer decides your runway.

Most companies aren’t hiring “coders.” They’re hiring people they can trust with outcomes. That’s the real gap 👇 Owners ask: • Why are we building this? • What happens if this fails? • How fast can users see value? Task-doers ask: • What’s the ticket? • What’s the deadline? Skills get you interviews. Ownership gets you offers. If you want better roles or clients, start thinking like the product is yours. That mindset changes everything.

AI didn’t replace SaaS products. It exposed weak ones. What actually matters now 👇 • Reliability over novelty • Systems over prompts • Speed over perfection • Retention over hype Anyone can demo an AI feature. Very few can make it work at scale. The future belongs to builders who treat AI like infrastructure — not magic. If you’re building right now, ask yourself: Will this still work when users actually rely on it? That answer decides everything.

Writing more code won’t make you senior. Thinking better will. The biggest shift in my career 👇 I stopped asking: “How do I implement this?” And started asking: “What breaks when this scales?” Now I focus on: • Data flow • Failure cases • User behavior • Shipping speed • Real outcomes That mindset matters more than any framework. If you want better roles, better clients, better pay — build like the product depends on you. Because eventually, it will. What mindset shift helped you most as a dev?

Most MVPs don’t fail because of bad ideas. They fail because founders overbuild. Here’s the MVP rule that saves months 👇 Your first version needs: • One user • One problem • One workflow • One outcome That’s it. No dashboards. No edge cases. No future-proof fantasies. MVPs aren’t about proving how smart you are. They’re about proving someone cares. If you’re building right now, ask yourself: What can I remove today and still launch?

Everyone talks about prompts. Almost no one talks about AI architecture. That’s why most AI SaaS products feel flaky. Here’s what actually matters 👇 • Clean data ingestion • Proper vector chunking • Deterministic retrieval • Guardrails for failure cases • Feedback loops for improvement Models change every few months. Architecture decisions last for years. If your AI app breaks when users scale, it’s not an AI problem — it’s an engineering problem. Serious builders know this. Do you?

“AI will replace developers” That fear disappears once you actually build with it. Here’s what AI really changed for me 👇 I don’t spend time on: • Boilerplate • Repetitive CRUD • Basic wiring code I spend time on: • System design • AI workflows • Product decisions • Shipping faster AI didn’t reduce my value. It removed the boring parts. The devs getting replaced aren’t bad. They’re just refusing to evolve. If you’re a builder, ask yourself: Are you fighting AI — or using it to ship faster?

Most SaaS founders don’t fail because of tech. They fail because they build too much, too early. I’ve shipped multiple AI-powered MVPs fast, and here’s the truth 👇 Speed isn’t about rushing code. It’s about ruthless scope control. What works every time: • Backend & AI logic first • Frontend only for real flows • No “nice to have” features • Launch → observe → iterate Fast shipping doesn’t reduce quality. It reveals what quality actually means. If you’re building your first SaaS, remember this: Users don’t care how elegant your architecture is. They care if it works. Would you ship faster if you removed 50% of features today?

“Full-stack” doesn’t mean knowing more tools. It means understanding impact. Real full-stack engineers think like this 👇 • UI affects backend load • APIs affect iteration speed • Deployment affects growth • Data affects everything • AI affects cost & reliability It’s not about writing more code. It’s about making fewer bad decisions. Do you think in features — or in systems?

Shipping isn’t the win. Iteration is. I learned this after wasting weeks on a “perfect” first version. Here’s what actually works 👇 • Ship early • Measure behavior • Ignore opinions • Fix friction • Iterate weekly Your v1 is a question. User behavior is the answer. Products don’t fail because they’re bad. They fail because no one listens after launch. Are you building for version 1 — or version 10?

I didn’t link my GitHub profile. I rebuilt GitHub inside my website. • Live contribution heatmap • Real-time commits & repos • Year-wise activity switch • Top repos ranked by actual contributions • Powered by GitHub GraphQL API No screenshots. No static numbers. Just real work, live. I’ve added a short demo so you can see it in action 👇 Portfolios should show proof, not promises.

Most AI products feel impressive. Very few feel useful. The difference isn’t the model. It’s the system. Here’s how I design AI SaaS 👇 1️⃣ Start with the user decision 2️⃣ Map the workflow without AI 3️⃣ Find the slowest manual step 4️⃣ Use AI to remove that 5️⃣ Add feedback + logs early LLMs don’t fix broken flows. They amplify them. AI works best when it disappears. Would you build AI this way?

I redesigned my Projects section to feel like a product — not a gallery. Here’s what users get now 👇 A clean bento-style grid Clear project descriptions Tech stack visible instantly Live Demo + View Code CTAs Hover → play icon appears Click → short preview video plays inside the card Smooth fade-in animations on scroll No guessing. No scrolling fatigue. Just a fast way to understand what I build and how I think. This section is optimized for one thing: great UX for busy recruiters, founders, and clients. Watch the video below to see the full interaction in action. And try it live here 👉 arkalalchakravarty.com What’s one thing most dev portfolios get wrong?

Writing good code didn’t level up my career. Understanding how products scale did. Here’s the shift that changed everything 👇 • Code is replaceable • Systems are not • Features come and go • Architecture compounds • Speed creates leverage AI didn’t change this. It amplified it. The best engineers aren’t faster typers. They’re better decision makers. What do you think matters more: code quality or system design?

I turned my About Me section into a story. Not text. Not cards. A sequence. When the section comes into view 👇 The paragraph types itself — explaining who I am and how I help. When typing finishes, the cards load one by one. Each card animates based on its content. The header keeps typing infinitely, adapting my message. All built with Framer Motion. All optimized to stay lightweight and fast. Why? Because users don’t read. They experience. If a recruiter or founder understands you in 5 seconds, you’ve already won. Watch the demo video below to see it in action. And try it live here 👉 arkalalchakravarty.com What’s one section on your site that could tell a better story?

Your MVP didn’t fail because users didn’t care. It failed because it took too long to ship. Here’s what kills MVP velocity 👇 • Backend-first obsession • No clear deployment path • Features without validation • AI slapped on without purpose • Zero iteration plan Winning MVPs do this instead: • Ship fast • Learn faster • Iterate weekly • Optimize after traction Speed isn’t reckless. Speed is survival. If you had 21 days to launch, what would you cut first?