Harsh Methwani

@Divyans90594944 @AnthropicAI

@KernelHarsh @AnthropicAI What about mythos 🤓

Hey @AnthropicAI when will you release a model for smart contract auditing ?

Today marks 8 years of QuillAudits. Most Web3 security firms didn't exist 8 years ago. Most won't exist 8 years from now. We've built through 3 bear markets, 2 exploit waves, and the full evolution of smart contract attacks from simple reentrancy to cross-protocol economic exploits. 1,500+ protocols. $3B+ protected. The biggest lesson from 8 years and 1,500+ engagements : One team, one method, one pass doesn't cut it when you're protecting hundreds of millions in user funds. So we rebuilt the model. Multi-Layer Audit → four independent security layers, delivered in the same timeline as a traditional audit: > Senior auditors who've collectively reviewed 1,500+ protocols > AI security agents trained on 5,000+ real exploits since 2017 > Independent bug bounty through curated security researchers > Continuous monitoring, because threats don't stop at deployment 4 layers. Each one catches what the others miss. Web3 has a $100T addressable market if institutions show up. They won't show up until security is embedded in every layer, every transaction, every deployment, the way HTTPS is embedded in the internet. That's the problem worth solving for the next 8 years. QuillAudits built the foundation, QuillShield is the next chapter — an AI security agent that brings what we learned from 1,500+ manual audits into every developer's workflow, before code ever hits mainnet. 8 years in. Still early.

If vibe coding is the future Code security scanning is the first step for any IDE/CLI before taking any step

@cluely should build a AI glass for cheating

@PatrickAlphaC This is a pretty serious thing the impact can be much wider, with prompt injections forcing the AI to run tools, delete databases, make transactions, and through some specific strategies even conceal what the AI is doing so that it gets ignored by users.

"but ser, if you ask your claude/openai/chatgpt/cursor/windsurf/openclaw to ignore sensitive files, it will - use the .cusorignore/rules/llm.md/etc file" If that is your security posture, you've already lost. Private keys should never be in plaintext

Idk after trying Codex I’ve stopped using Claude for coding. The speed of Codex is addictive thinking, asking, reviewing all fast. I still have Claude open in another tab, but my brain automatically jumps to Codex. TBH, I’m asking way more with Codex than actually building.

Bullish on oil 🥹

I think Grok should automatically flag all videos that appear suspicious and are not real, rather than people tagging and asking, "What do you say about this feature, @grok?"

No-one wins in war

Benchmark saturation has come; models need harder benchmarks

@RoundtableSpace GLM-5-Code will be fun 🤩

AI MODELS RUMORED TO BE RELEASING THIS MONTH: - DeepSeek V4 - GLM-5-Code - GPT-5.4 - GPT-5.3 Garlic - Gemini 3.1 Flash - Claude Sonnet 4.7 - Meta Avocado Which one are you most excited for?

@PatrickAlphaC A prompt injection in any file of a codebase, with medium or low reasoning/safety model, can wreck lives.

"Oh, it won't access the files in your .gitignore" People. People. People... "Won't" is VERY different from "can't". Prompt injection is real. If your AI "can" access the data, you "can" get your shit rekt.

@pratyush_r8 A laptop with ban on claude still hire for Agentic/AI Engineers

strange that its 2026 and there are still some "big tech" cos not offering claude code or their alternatives to all their employees ngmi

Imagine vibe directing a movie

WW3 on X >>> any news channel

Building the next B2B or B2C SaaS is definitely a gamble. OpenAI, Anthropic, and Chinese labs won’t leave any market untapped. Relying only on LLM APIs and a few tools won’t justify such crazy valuations. If adoption is slow, they might build more tools to drive adoption faster.