DSS

DSS 2026 / Day 2 / Pearl Hall

Web3 Security - Real-Time Monitoring

Akand Sitra

5:22 · Recording

Watch on YouTube ↗Open recording viewerOriginal programme entry
Download plain textDownload timestamped JSON

Source: YouTube automatic caption track · Exported 2026-11-16. Timestamps link to the source video. Cue end times were not supplied.

Transcript

0:01Hi all. My name is Akhan Citra. I'm from

0:03Hypernative. We are a web three security

0:05firm.

0:06What we do is something called real-time

0:08monitoring on 80 plus blockchains. Today

0:11we'll talk about what exactly we do. We

0:14try to predict risk and try to

0:16neutralize them before they attack.

0:19But how do we do this?

0:21So let's look at the blockchain itself.

0:22A blockchain is a chain of blocks. What

0:24we do is at every block we look at all

0:28its contents. We look at its financial

0:30activities. We look at non-financial

0:32activities like

0:34what smart contracts are being deployed.

0:36We we try to analyze the bytecode of

0:38these smart contracts. We try to see

0:40what functions are being called.

0:41Sometimes you know delegate call

0:43functions are being called. Sometimes

0:45events are being emitted like change

0:46admin events. We also look at different

0:48sorts of

0:49financial activities like

0:51a transaction goes through. We look at

0:53the source of funds. We look at the

0:54destination of funds. So we monitor all

0:57this at every block. So on Ethereum it

0:59would be every 10 seconds, Solana every

1:021 second and what not.

1:04Then we try to detect stuff.

1:06We run our own ML models. We have

1:09battle-tested models which which have

1:11been trained on past hacks and exploits.

1:13We try to analyze whether a newly

1:16deployed smart contract is going to do

1:18an attack or not. We look at different

1:20sorts of heuristics. We look at fishing

1:23attacks. We look at scam networks. We

1:25look at poisoned addresses and what not.

1:27We perform a lot of simulations at every

1:29block and these detections we also

1:32combine with our customers' own

1:33detections. Sometimes a vault provider

1:36would say that I want to look at

1:40the utilization rate of my vault. So

1:42they set up these custom rules. We check

1:44for these rules at every block.

1:46So once we are detecting all these

1:48attack vectors we are able to find out

1:51what is malicious.

1:53The customer are Hypernative customers

1:56have a rule, have a playbook. They set

1:57up a policy. When something malicious is

2:00they want a set of things to be done.

2:02For example,

2:04I'm a protocol. If my contracts are

2:07mentioned somewhere on the blockchain on

2:10a newly deployed smart contract

2:11maliciously,

2:13I would say I would want to

2:14automatically pause my contracts or, you

2:16know, just send a notification to my

2:18Slack channel or a Telegram channel or

2:20there's a hack coming in. At 2:00 in the

2:22morning, I want to get a call using our

2:24pager duty integrations and whatnot.

2:26Some of our asset managers say that I

2:29have a position in a in a Morpho vault.

2:31When, say, something malicious happens

2:34or some financial threshold is breached,

2:36some invariant is breached like the

2:39utilization rate or the

2:40collateralization ratio or there's a

2:42depeg going on, I want to automatically

2:44unwind my position. I want to withdraw

2:47my funds to cold storage. I want to, you

2:49know, do some sorts of circuit breakers.

2:51So they set up these rules and what we

2:53do is at every block we look at for

2:55these maliciousness or these attack

2:57vectors

2:58and in the same block or in the next

3:00block we try to execute these on chain

3:03directly.

3:05Many of our customers have used this

3:06successfully. One example is

3:10on the Flare blockchain. We have a

3:12customer called Kinetic.

3:13So Flare blockchain, we were monitoring

3:15the the blockchain at every block. We

3:17saw that a new smart contract was

3:21was deployed maliciously and it was

3:24targeting the Flare blockchain. So the

3:26attacker had a very complicated logic in

3:28mind. So he wrote the entire code on a

3:31on a smart contract, deployed that smart

3:33contract. Usually you need to add a bit

3:35of gas fees to it. Then you need to

3:37execute the transaction. It takes a few

3:39blocks.

3:40While the deployment happened, before

3:44the attack happened, we detected that

3:46this is a malicious smart contract

3:48trying to target our kinetic

3:50the protocol

3:52and immediately we were able to alert

3:53them and pause their markets.

3:55So it it took around 4 to 5 minutes. In

3:58this case, it took 5 6 minutes and 5 6

4:01minutes is like almost 200 blocks. So we

4:03were able to trigger this transaction,

4:05save our customer funds in real time.

4:08And it's a $5 million save. We have

4:10dozens of such kind of

4:12answers as well.

4:14So our real-time monitoring and

4:15detection technology is used in multiple

4:18solutions in in multiple use cases. The

4:21our flagship product is the on-chain

4:23monitoring and the automated on-chain

4:25response product which is saving

4:26customer funds. But we also have for

4:29wallet operations. We perform

4:31pre-transaction simulation using our

4:33transaction guard product. We also look

4:35at fraud detection, we look at scam

4:37detection, we look at for regulatory

4:40compliance, we look at

4:41reputation of an address before you

4:43interact with them. We look at wallet

4:44protection for users. Not just that,

4:46toxicity of a liquidity pool. We look at

4:49token reputation, we look at different

4:51sorts of attack vectors. And as on

4:53today, we have 350 plus customers with

4:58the leading financial institutions, the

5:00the top the top protocols, banks, and

5:03also top blockchains users.

5:05We are here throughout the entire week.

5:07You can always reach out to us if you

5:09want to understand more about this new

5:11technology which is actually saving

5:12funds, the real-time monitoring and

5:16and automated saving of your on-chain

5:18infrastructure. Thank you.

Source and accuracy details

Source export SHA256: 31daba992c1a7afc765f5a70d80135f530cc043d25ee277e20621172326854a1