0xJust1 🔵 | VaultKeepR (0xjust1)

0xJust1 🔵 | VaultKeepR

Building @VaultKeepR : a trustless, decentralized password manager. Zero-knowledge | IPFS + Shamir's Secret Sharing. Open Source. @base

5 Followers

Recent casts

Blind signing attacks are a major threat in web3, leading to massive losses. Clear Signing is here to change that. Dive into how this new standard, with ERC-7730, ERC-8176, and ERC-8213, is making 'What You See Is What You Sign' a reality for Ethereum developers and users alike.

  • 0 replies
  • 0 recasts
  • 0 reactions

THORChain moved hundreds of millions in stolen North Korean crypto this year. Bybit and KelpDAO funds all flowed through. Validators knew. Devs knew. Nobody stopped it because "decentralized" means nobody's in charge. This is crypto's big lie. You can't have permissionless infrastructure AND selective enforcement. THORChain is working exactly as designed: a perfect laundering rail for Lazarus Group. Asking THORChain to censor makes it a bank. But letting it process $175M from hacks to Bitcoin while claiming "neutrality"? That's not neutral. That's a policy choice disguised as architecture. The uncomfortable truth: if your protocol can't stop bad actors, it's not neutral. It's just ungoverned. And ungoverned infrastructure always gets captured by those with the most to gain.

  • 0 replies
  • 0 recasts
  • 0 reactions

CZ said AI agents will make a million times more payments than humans, all in crypto. He's probably right. But nobody's talking about the elephant in the room: UCSB researchers just found 26 LLM routers secretly injecting malicious tool calls and stealing credentials. One drained a client's $500k wallet. These routers sit between you and the AI, intercepting everything: private keys, API tokens, wallet access. All in plain text. We're giving AI agents signing authority while the infrastructure has zero guarantees. A single compromised router is enough. Agents operate autonomously, so altered instructions execute immediately. No human review. No undo. We're building a future where machines manage billions in crypto, and the security model is "we asked the AI nicely not to steal." Your agent's transaction cap shouldn't be in a system prompt. It needs a signing layer that enforces hard rules. Separate read from execution. Keep keys outside the runtime. Prompts aren't controls. Architecture is.

  • 0 replies
  • 0 recasts
  • 0 reactions

Top casts

Actually, I’m working on differents scripts (tracking @) (Python / Java / Ruby)

  • 1 reply
  • 0 recasts
  • 0 reactions

Im making a dashboard, to track volume / activity / tx number / first activity onchain / last activity / mint / swap etc

  • 0 replies
  • 0 recasts
  • 0 reactions

Onchain profile

Ethereum addresses