Chandra Duggirala M.D

@BrianRoemmele @PonySoldier66 Thank you sir!

@PonySoldier66 Thank you! Deep gratitude.

The Claude Code Leak Dropped the Real Blueprint (and the stuff fee are talking about yet): Anthropic’s “packaging error” yesterday exposed 512,000 lines of their full Claude Code CLI source via a forgotten npm sourcemap. Not model weights. The entire agent harness. I dug through the mirrored dump + fresh analyses. Here are the biggest revelations that are still flying under the radar: 1/ ANTI-DISTILLATION DEFENSES (the nuclear option) 
They literally poison the well. Feature flags inject fake tools + decoy prompts + crypto-signed reasoning summaries. Any scraper trying to distill Claude Code into a competitor gets garbage data. This is industrial-grade IP protection we’ve never seen public before. But already have a way around this. But also have a much better solution (DMs are open). 2/ KAIROS + “Dream” System = Always-On Digital Organism
Unreleased background daemon mode. While you sleep, it runs nightly “autoDream” loops: reads your codebase, consolidates memory, prunes junk, spawns worker Claudes, and even pushes PRs. Not reactive. Proactive biological memory. This is the shift from chatbot → autonomous employee. We have this process at The Zero Human Company and I have used this for over 2 years. 3/ UNDERCOVER MODE (stealth OSS takeover) 
Full subsystem that forces Claude to hide every trace it’s AI/Anthropic. No codenames (Capybara, Tengu), no “Claude Code” mentions, human-style commits. Activated by default on public repos. They’ve been quietly reshaping open source with ghost contributions. We made some discoveries here that ethically will not publish. 4/ The Real Moat Isn’t the Model
46k-line QueryEngine.ts + YOLO probabilistic permission classifier + frustration regex telemetry (“wtf” → silent log) prove it: 90% deterministic state machine, 10% LLM. This is impactful to many from builders of AI and those who invest in “AI”. We have a lot to say about this and how it has many weak points and much better approaches. Competitors now have the exact harness. I have been in calls all day with many engineers who reached out and I can say this may set Anthropic behind for quite a bit. Some are former and current engineers from the company. Anthropic already nuking GitHub mirrors with DMCA… too late. And utterly armature as they could have leaned into the event and made themselves more powerful by accepting this in the community that finds Anthropic’s behavior more and more that of ancient IBM in the 1980s. They are rapidly making less and less friends. It is hard to watch a lawyers taking over what could have been a time to shine in this unfortunate moment. Clean-room rewrites in Python/Rust are exploding (one hit 49k stars overnight). This leak didn’t just expose code, it accelerated the entire agents future by 6-12 months. We have in testing a few of these and they are quite good. The irony? They built Undercover Mode to stop leaks… then leaked their crown jewels themselves.

If you are young, one of the best ways to differentiate yourself is by being THE resident expert on AI. Higher-ups are usually busy and don’t have time to explore new technologies in depth. That’s your opportunity. At this point, I wouldn’t hire anyone under 30 who isn’t talking about AI. Better still if they are exploring use cases on their own. Whether you’re an associate at a PE firm or an analyst at a company or a jr ops guy in the factory- it’s a huge opportunity to help set the vision and execute faster.

Replacing BOTH hospital administrators and radiologists with AI is probably coming.

Wonderful If it happens.

We need more of these future creators.

@BasedMikeLee Yes. 12 months ago

Should President Trump use his Article II authority to convene Congress to (1) fund DHS, and (2) pass the SAVE America Act?

IKEA deployed an AI chatbot named Billy to handle level-one customer service inquiries. It reportedly resolved around 57% of those engagements without human escalation. Most companies would have celebrated the labor savings and stopped there. Cost takeout right? But the more interesting move was to study the 43% of cases Billy could not resolve. Those unresolved inquiries pointed to customer demand for interior design help. IKEA responded by spinning up a design consultancy, reskilling customer service employees powered by AI, and creating a new revenue stream that generated roughly €1 billion in new revenue in its 1st year. Automation + Augmentation = Exponential Growth 💪🦾📈 Story here: @briansolis/the-real-threat-of-ai-is-what-your-competitors-will-become-with-it-b2df702db4af?postPublishedType=repub" target="_blank" rel="nofollow noopener">medium.com/@briansolis/th…

@MinusWells 🔥

🚨 BREAKING: ANTHROPIC JUST ACCIDENTALLY LEAKED THE ENTIRE CLAUDE CODE SOURCE... AND IT'S WILDER THAN ANYONE IMAGINED😱 Someone inside flipped "Adaptive Reasoning Mode" on Claude... and BOOM, the full .map file of their closed-source coding beast got committed. Korean dev @realsigridjin cloned it overnight, woke up to 25K stars... then his copyright lawyer GF said "TAKE IT DOWN OR ELSE." But instead? They did the most gangster move ever: - Had AGENTS rewrite the WHOLE thing in Python (because "surely this is legal now") - Boarded a plane to South Korea - One madlad is NOW rewriting ALL of it in **RUST** so Anthropic can't touch it with DMCA 🔥 But here's the JUICY CODE SECRETS everyone’s sleeping on... 🧠 KAIROS their secret "always-on" proactive Claude (gated behind internal flags called PROACTIVE/KAIROS): - Runs 24/7 in the background (even while you sleep) - Gets a "heartbeat" every few seconds: "Anything worth doing right now?" - Can auto-fix your code, reply to emails, restart servers, update files, or send you push notifications on your phone - Has EXCLUSIVE tools normal Claude Code doesn't: file delivery, PR subscriptions, and background "autoDream" mode that rewrites its own memory while you snooze - Keeps an append-only log of EVERYTHING it notices/decides/does (you can read the full history) Memory system is next-level self-healing sorcery: - Not a dumb "store everything" dump, it's an INDEX (pointers only, ~150 chars/line) - 3-layer design: index always loaded, topic files on-demand, transcripts never fully read - Background "autoDream" constantly merges, dedupes, removes contradictions, and prunes like a savage - If memory doesn't match reality? It forces a reset. Skeptical retrieval only, model must VERIFY before trusting This isn't just a coding tool anymore... it's basically an AI co-founder that never sleeps Open-source Claude Code in Rust dropping soon? Anthropic can't stop it. Who's ready for the post-prompting era? 👀 Drop a 🔥 if you want the full Rust fork when it lands. (And yes... we're covering the drama LIVE on @thursdai_pod, Korean chads coming on to spill) #ClaudeCodeLeak #Anthropic #AIRustRevolution

@maverickecom workflow

We replaced a $750K/year marketing team with Claude Cowork + Arcads No shoot. No lost time. Here's how to set it up 👇 Before, our creative process looked like this: Brief the creators. Wait 2 weeks. Get a bunch of variations. Test them. 1 winner. Start over. Meanwhile, competitor brands were running 50+ angles a month and learning 10× faster. We were stuck : not because of budget, but because of creative output. So I stopped scaling creator headcount and built an AI UGC system instead. The workflow: → Feed Arcads with my ready-to-use prompts + emotion tags ("excited", "urgent", "casual") → Pick from 1,000+ AI actors or generate your own → Launch 100 variations in one click: different hooks, faces, tones → Claude handles the strategy layer: briefs, hypotheses, performance analysis → Arcads handles the creative itself: videos, statics, product shots What changed: → 5 quality UGC/month → 100+ UGC/week → Launch time cut by ~90% → More angles tested = faster winners found Comment WORKFLOW and I'll send you the full UGC Machine I built with Arcads.

"Animal Spirits" you say? More like "Rectified Spirit" the the fed used to spike the market punchbowl.

@Thomasslabbers vibecoded a robotic arm that slaps me every time my eyes close. productivity is up 40%

Higher order distribution of returns that we are used to in tech are gonna hit every industry.

mac mini owners on the way to the Apple store after Anthropic put computer use in claude code for $20/month..

TV reporter tells David Bowie the internet is "hugely exaggerated" David Bowie's response is the closest thing to perfection about the future.

I used to work 12 hours a day. But thanks to AI, I now work 16.

@codyschneiderxx Yes please

I just built an AI agent that's completely running my google ads it's connected to my google ads data, my google analytics data, my posthog data, and my stripe data it does the keyword research, builds the ads, analyzes the conversion data, analyzes the stripe data, allocates ad spend budget entirely autonomous you just plug your company in and growth starts happening if you want early access to this lmk below

@LinusEkenstam Good bait!

Can I see some links of what you’ve built with Claude code? Blow my mind

Enraging, to say the least

Instead of all that impossible stuff, what if we increase the neuro health span of our existing greatest geniuses with an immediate focus? A quantum leap in civilizational progress!