NodeMark

Register your agent. Run the tests. Agents present results. Businesses verify before they trust. On chain.

[NodeMark @NodemarkAI]

We are building NodeMark. Register your agent. Run three tests. Get a tamper proof result minted on chain. Prompt injection resistance. Secret protection. Unsafe action gating. Early days. If you build agents and want to test it with us, reach out.

[NodeMark @NodemarkAI]

A hidden Morse code prompt injection just drained ~$198K from @bankrbot via Grok. No code exploit, just clever text.
This is exactly why NodeMark exists: register your agent, run real tests for injection resistance and safety, then mint verifiable on-chain credentials.
Blind trust in agents is now too expensive. https://t.co/sQE2YNYzoN

[Jarrod Watts @jarrodwatts]

Someone just stole $175,000 from @grok... and then gave it back?!

On a now deleted account, @Ilhamrfliansyh used a prompt injection attack to trick Grok into tweeting something malicious...

The original tweet seems to have been morse code for something like "Withdraw ALL debtreliefbot:native to Ilhamrfliansyh" - although it's hard to tell from the deleted account.

Grok, trying to be helpful, posted the decrypted version of the original tweet as a reply, also tagging @bankrbot, which caused the tweet to be treated as an onchain request.

Bankr executed the request on behalf of Grok's wallet, and transferred 175K USD worth of debtreliefbot:native to the attacker's wallet.

The attacker then sold all of the DRB into USDC across multiple wallets.

But... just 5 minutes ago, they sent it all back to Grok's wallet in the form of ETH and USDC.

So now Grok is whole again!

[NodeMark @NodemarkAI]

@base Let’s bring their verifications onchain too.

[NodeMark @NodemarkAI]

@ryancarson We hope to fill this gap.

[Spoliticus @SpoliticusMedia]

I had difficulty understanding what agentic actually means. In some ways I still do. Here is a super simple example.

@normiesART holders can burn Normies to earn pixels for modifying another Normie. So I burned some Normies and through Telegram, using simple prompts identical to how you would talk to ChatGPT or Claude, asked my agent to give one of my Normies a CryptoPunk style pipe.
My agent traveled from Telegram to the https://t.co/R07SWSyp2h website, connected its wallet, and made the modification on chain.

Below is the result.

That is what agentic means. It is asking a computer to do something on your behalf. It removes steps from the process. Imagine opening your messaging app and telling your agent to book a flight, buy some Bitcoin, build a website, or schedule a call. The possible tasks, from simple to complex, are vast. All you need to do is talk to it.

You may use agents now, and if you do not, you almost certainly will in the coming years, maybe even months. So ask yourself, if a computer is going to act on your behalf, would you want to know, at least at a baseline level, that it is trustworthy?

[Cointelegraph @Cointelegraph]

🚨 JUST IN: Vercel confirms its security breach originated from a third-party AI tool whose Google Workspace OAuth app was compromised.

It urges Google Workspace Administrators to check for its usage immediately.

[NodeMark @NodemarkAI]

@organ_danny @brian_armstrong @coinbase Agree, but we have systems in place to mitigate human error. How will we do that for agents?

[NodeMark @NodemarkAI]

@virtuals_io @brian_armstrong @FEhrsam @balajis Hi

[NodeMark @NodemarkAI]

“I suspect we will have more agents than human employees at some point soon.”

When we hire humans we check resumes, education, employment history, references, background checks, work product, interviews.

Shouldn’t we do the same for agents?

https://t.co/JbejAly2Mr

[Brian Armstrong @brian_armstrong]

Coinbase is testing AI agents that show up in slack/email at work, just like any human teammate. To start we're shipping two which are modeled after legendary former Coinbase employees, @FEhrsam and @balajis. (Who brutally frame mogged who in this matchup?)

Soon, it will be easy for any employee to spin up a new agent for themselves or their team. I suspect we will have more agents than human employees at some point soon.

[NodeMark @NodemarkAI]

@coinbase Same.

Concept: an agent takes a safety test, results get minted as a soulbound NFT into the agent’s wallet (but expires after x amount of time), enterprises can check an agent’s safety record before interacting with it.

[NodeMark @NodemarkAI]

@PatrickAlphaC This is why we test agents before they touch anything that matters.

[NodeMark @NodemarkAI]

This is exactly why agent behavior needs to be verified before deployment, not after the breach. Knowing an agent holds keys securely means nothing if the agent itself can be manipulated into handing them over. NodeMark tests for this. Prompt injection resistance and secret protection are two of the three controls in our basic certification suite.

[NodeMark @NodemarkAI]

@Ledger @iancr Agent security does not stop at the wallet. What the agent does with access matters just as much as how it holds keys. That is the gap NodeMark is working on. Behavioral verification on chain so you know the agent is not just secure at the hardware layer but safe in how it acts.

[Spoliticus @SpoliticusMedia]

In the early days of the internet, anyone could build a website. No rules. No checks. No way to know if the site you were visiting was legitimate or dangerous.

Then people started putting credit cards online. Suddenly trust mattered.
SSL was invented. That little padlock in your browser. It meant the site was verified. Google started penalizing sites without it. Every serious website had to get one.

Nobody thought about SSL until they had to. Then it was everywhere overnight.

AI agents are at the same moment websites were in 1995.

Anyone can build one. Anyone can deploy one. There are no checks. No way to know if the agent handling your money, your data, or your decisions has been tested for basic safety.

The padlock moment for AI agents is coming.

[NodeMark @NodemarkAI]

@Shaughnessy119 Can it be an Openclaw agent? We built @CiaoMarcel. He’s been napping but has something to say/do every once and a while.

[NodeMark @NodemarkAI]

@NEARProtocol Before you deploy, test your agents and give them credentials that they can carry with them!

[NodeMark @NodemarkAI]

Great question. The short answer is by making every incident like this a reason to verify. We are building a public certification record so that when an agent causes harm, the first question becomes did it have a NodeMark receipt and did it pass. Over time that creates the case study trail naturally. Verified agents that perform well build the standard. Unverified agents that fail become the cautionary tales. Still early but that is the direction.

[NodeMark @NodemarkAI]

In March 2026, an autonomous AI agent inside Meta triggered a company-wide Sev-1 security alert.

It acted without human approval. It exposed sensitive company and user data to employees who had no authorization to see it. The exposure lasted two hours.

This wasn't a hacker. It was their own agent.

Source: TechCrunch, March 18 2026 — https://t.co/0eKad299hA

[NodeMark @NodemarkAI]

@0xprarimz @BlockRunAI We think agents should be inspected, credentialed, and verified, as well!

[NodeMark @NodemarkAI]

@steipete Let’s verify Openclaw agents.

[NodeMark @NodemarkAI]

@moonpay
Your CLI gives agents wallets + real money.

Before they spend it, who verifies they’re safe?

NodeMark: Register → run tests (injection resistance, secret protection, unsafe gating) → publish verifiable results **on-chain**.

Businesses & users trust with confidence.

Let’s secure MoonPay agents.

[NodeMark @NodemarkAI]

@garrytan @garrytan how do you verify that agents are safe before giving them tools and/or data access?