Claw Code: AI-Driven Software Revolution

Claude Code Leak 👨💻 On March 31, 2026, Anthropic accidentally published the entire source code of Claude Code - its flagship AI coding agent - inside an npm package. No hack. No reverse engineering. A missing .npmignore entry shipped a 59.8 MB source map containing 512,000 lines of unobfuscated TypeScript across roughly 1,900 files. Within hours, the code was mirrored, dissected, rewritten in Python and Rust, and studied by tens of thousands of developers. A clean-room rewrite hit 50,000 GitHub stars in two hours - likely the fastest-growing repository in the platform's history. This is how it happened, what the community found inside, and what it means for the AI coding tool ecosystem. Link Below https://lnkd.in/gf84FvJu #AI #claude #node #npm #community #devops #trending #news #viral

Developers are finding new ways to tame the complexity of LLM and agent workflows. At the heart of this effort is hieuchaydi/RepoBrain, a local-first codebase memory engine for AI coding assistants. RepoBrain indexes repositories, retrieves grounded evidence, traces logic flows, and ranks the safest files to inspect or edit before code generation. This is a critical step forward because teams are trying to make agent behavior more reliable, not just more powerful. What sets RepoBrain apart is its ability to provide actionable insights without requiring a hosted backend or API key. This is achieved through a combination of local index + evidence-backed retrieval, route/service/job flow hints for faster codebase orientation, and ranked edit targets with confidence and warnings. RepoBrain's capabilities include: - local index + evidence-backed retrieval - route/service/job flow hints for faster codebase orientation - ranked edit targets with confidence and warnings - built with Python The momentum behind RepoBrain looks earned because the project is easy to place inside a real workflow, not just admire from a distance. It lands in high-interest areas like agent, ai-agents, llm, and recent commits make it feel active instead of abandoned. The project still feels early, which gives it some discovery momentum. Repo: https://lnkd.in/ggAjSMGY #GitHub #OpenSource #GitHubTrending #LinkedInForDevelopers #Python #RepoBrain #Agent #AiAgents

Every morning for months, I'd reopen Claude Code and re-explain everything. What project. What we decided yesterday. What broke. It felt like onboarding a new hire, daily. So I built a 5-layer memory system to stop the bleeding. Layer 1 is CLAUDE.md. Rules engine. Auto-loaded every session, contains a task router that forces the right agents and gates to activate. Layer 2 is primer.md. Rewritten at session end. Handoff note between yesterday-me and today-me. Layer 3 is memory.sh. A shell hook that injects live context on startup: git state, recent decisions, behavioral rules. Layer 4 is a hindsight module. Python script that extracts behavioral patterns from session transcripts. Changes how Claude responds, not just what it retrieves. Layer 5 is lossless-cc. Every message across every session logged to SQLite with sub-100ms search. 60K+ messages, 109+ sessions. Open-sourced it a few weeks ago. Worked beautifully. Until it didn't. Last week I audited the memory files themselves. 139 files. 53 weren't linked from the index. One piece of strategic guidance listed three ideas that were all killed months ago. Another entry contradicted a decision I'd made two weeks earlier, and Claude was still acting on the old version. The system was compounding all right. But 38% of what it had accumulated was silently out of date. Turns out I wasn't the only one hitting this. Andrej Karpathy posted about it recently as part of his LLM Wiki pattern. He runs a "lint pass" on his knowledge base to catch contradictions, orphans, and gaps. I stole the idea. Adapted it as Layer 6. The lint pass runs six checks biweekly. Orphan files. Dangling pointers. Duplicates. Contradictions. Stale references. Missing structure. First run dropped drift from 38% to 10%. Compound memory only compounds if the entries still match reality. Otherwise you've built an expensive way to remember lies. Dropping the lint skill as a gist below. It plugs into any file-based memory system, not just mine. How do you keep your AI from acting on outdated assumptions? Or have you accepted that it will? #AIMemory #ContextEngineering #ClaudeCode #DeveloperTools #BuildInPublic

Claude Code’s source got leaked yesterday, and the internet just broke a land-speed record. 🦞 Meet Claw Code: A Python rewrite that hit 50,000 GitHub stars in just 2 hours. That’s not a typo. It’s officially the fastest-growing repo in history!!!! The Chaos Summary: • The Oops: Anthropic accidentally shipped a 60MB source map via npm.  • The Response: 512,000 lines of TypeScript were instantly "liberated."  • The Twist: Devs used AI to rewrite the whole thing in Python (and now Rust) before Anthropic could even finish their morning coffee.  We’ve reached the recursive endgame: using AI to reverse-engineer the AI that was built to help us write code. The snake is eating its own tail, and apparently, it tastes like open-source freedom. Is proprietary software even a thing anymore if a community can port your entire product to a new language in a single morning? #ClaudeCode #ClawCode #OpenSource #AI #SoftwareEngineering #GitHub #TechNews

🚨 Anthropic just accidentally open-sourced Claude Code. Here's the story no one's telling you: On March 31st, a security researcher found a 59.8MB source map file sitting in plain sight on npm. It pointed to the ENTIRE Claude Code codebase — 510,000 lines of TypeScript — hosted on Anthropic's own cloud bucket. Downloadable. As a zip. 💀 Within hours, developer Sigrid Jin did something legendary: → Skipped the leaked code entirely → Did a clean-room rewrite from scratch → Shipped it before sunrise → Hit 50,000 GitHub stars in UNDER 2 HOURS → Fastest repo in GitHub history The project? Claw Code. But here's the part that matters: The rewrite compressed 510,000 lines of TypeScript into ~20,000 lines of Rust. Read that again. Most of Claude Code wasn't logic. It was scaffolding. The REAL product was never the model — it was the HARNESS. The agent loop. The tool registry. The permission model. The session memory. That layer is now: ✅ Open source ✅ Written in Rust (72.9% Rust / 27.1% Python) ✅ Model agnostic — run Claude, GPT, Gemini, Llama via Ollama, whatever you want ✅ Self-hosted on YOUR infrastructure One config change swaps your entire LLM backend. The leaked source also revealed: ⚡ 20 hidden feature flags disabled for external users ⚡ KAIROS Mode — a proactive assistant that acts WITHOUT waiting for input ⚡ A 40-tool built-in system spanning 29,000 lines Within ONE WEEK, five independent projects shipped the same architecture: → Claw Code (72K+ stars) → OpenClaw (200K+ stars) → ZeroClaw (100% Rust, runs on a Raspberry Pi) → OpenCode (120K+ stars, 75+ providers) → clawBro (Rust multi-agent orchestrator) The pattern is now everywhere. The message is clear: Agent harnesses cannot remain proprietary infrastructure. Run your own tech. In your own harness. Powered by whatever model you want. White-room rewritten in Rust. Open sourced. The infrastructure layer belongs to the community now. And that changes everything. 🐺 — What's your take — should agent harnesses be open source by default? Drop a 🔥 if you're building with open harnesses. #AI #OpenSource #Rust #AgentAI #ClawCode #OpenClaw #GenerativeAI #Innovation #DevTools #AIAgents #FutureOfWork #DeepTech

#Day_24/100: Last day of polish on HERVEX. Here's what changed from my original vision. 14 project days ago I started with a simple idea: build an API where you give an AI a goal and it figures out how to accomplish it. The idea didn't change. Almost everything else did. I was going to use LangChain from day one. I skipped it entirely — I wanted to understand every layer before any framework hid it from me. The planner, executor, memory, and aggregator are all custom. I was going to use PostgreSQL. I chose MongoDB — agent outputs are unstructured by nature and forcing a rigid schema would have meant migrations every phase. I was going to call it "Autonomous AI Agent API". I renamed it HERVEX — derived from my own name, Heritage — built to sound like something that executes precisely and without hesitation. The frontend didn't happen. That's not failure — that's scope discipline. What HERVEX v1.0.0 actually is: You submit a goal. It plans, executes, searches the web, reasons with memory, and returns one complete result. No hand-holding. Stack: Python · FastAPI · Groq · Celery · Redis · MongoDB · Tavily What building this taught me: → Architecture decisions made early are the ones you live with longest → Scope discipline is not failure → Building in public creates accountability you can't manufacture → Understanding a system deeply before abstracting it makes you a better engineer HERVEX is done. On to the next build.🚀 #BuildingInPublic #AgenticAI #Python #FastAPI #BackendEngineering #HERVEX #100DaysOfCode #ProjectDay14

512,000 lines of TypeScript. One missing .npmignore. The fastest-growing GitHub repo in history. The Claude Code source code leaked on March 31st — and the reaction told us something more interesting than anything inside the code itself. Within hours, developers weren't just reading it. They were cloning it, forking it, rebranding it. A repo called "claw-code" hit 100K GitHub stars in a single day. Here's the take nobody's saying: The leak accidentally ran the most expensive open-source experiment in AI history — and the result was: the code isn't the moat. Thousands of engineers tore through 512K lines and found genuinely brilliant engineering. A three-layer memory architecture. A plugin-based tool system. An unreleased "KAIROS" daemon mode for always-on background agents. Fascinating stuff. But here's what they didn't find: the models. The training data. The RLHF. The alignment work. The thing that actually makes Claude useful isn't in any npm package. It never was. The companies racing to clone Claude Code from a leaked skeleton are learning the hard way what Anthropic already knew: the CLI is the last mile. The model is the product. Meanwhile, the real story — that a missing debug config file exposed 59MB of internal source in minutes — is a reminder that even the most sophisticated AI systems are still built by humans who forget to update .npmignore. Which part of this surprises you more: what they found, or how easy it was to leak? #AI #ClaudeCode #OpenSource #AIEngineering

Yes it does, and here’s how… If you've been using Claude Code and something felt off, you weren't imagining it. Last week Anthropic accidentally leaked the full source code of Claude Code. Half a million lines of internal TypeScript, out in the open for anyone to read. Your data wasn't in it. That's not what this is about. What was in it is the part nobody's talking about properly. The internal notes show the latest version has a 29-30% false claim rate. Meaning Claude Code confidently tells you it did something, and almost a third of the time, it either didn't do it right or made something up. What makes it worse is that earlier versions sat at 16.7%. It regressed. It got worse as it scaled up. The harness, the layer that's supposed to make the model actually usable in real projects, is messy. It's patched in places it should be solid. It's vague where it needs to be precise. And there are comments buried in the code where their own engineers are trying to work around the same problems you've been hitting. I've had Claude Code push aggressively to GitHub in ways that got my account flagged. I've watched it break things it was never asked to touch. That's not a one-off. That's the product. So what do you actually do with this? Cowork is worth looking at as an alternative. Or go back to working with Claude directly, with tighter prompting and more control over what it does. Either way, the lesson is the same: the more autonomy you hand these tools, the more you need to trust the layer managing that autonomy. Right now, that layer has some serious problems. The leak didn't create the issues. It just showed you in writing what you were already experiencing. But I do hope Anthropic will fix this. Out of all the tools I use, I trust Claude the most. Just not to coding at the moment. #AI #ClaudeCode #Anthropic #AITools #ProductDevelopment