Father. Software Developer. Successful Businessman irl. Successful on-chain.
- Tweets 253
- Following 228
- Followers 13
- Likes 1,234
We recently released Privacy adversaries, which shows which adversaries each privacy protocol protects you from.
Every protocol page has a detailed section, where each rating comes with advice on what a careful user can do to stay private.
Here's are some of the top advices 👇
Every privacy protocol says "private."
But the question is: private from whom?
We just shipped a new feature on L2BEAT that answers that for every protocol we track, against 5 different adversaries.
8
7
44
3,716
Do you think people are actually going to adopt agentic automation for wallets?
You can definitely build features around cron jobs and automated actions, but I still don’t think I’d be comfortable giving an agent unrestricted approval to spend money directly from my main wallet.
For things like farming, airdrops, or other low-value activities, where I’m only putting a small amount of money at risk, it makes much more sense.
On the other hand, I think agentic systems could become very useful as a security layer on top of wallets. Humans can make mistakes while signing transactions, setting permissions, or delegating access.
If a user instead sets predefined rules, limits, and permissions inside a specific agentic sandbox, the agent could operate only within those boundaries. That could significantly reduce the chances of human error while still keeping the user in control.
Your AI agent says: "Deposited 100 to Morpho Vault 0xeE8F…"
Did it?
@CoinbaseDev's AgentKit returns exactly that string without checking whether the tx reverted. Every agent wallet I've read checks before signing. None of them check after.
I'm building the after.
2
2
40
As one of our Walrus Session 7 winners, @neopsih upgraded Markov using Walrus Memory to solve how agent handoffs handle unverified state.
When a fresh AI session blindly trusts notes from a prior run, it can easily resume work on a broken state.
This upgrade turns handoffs into self-proving objects, forcing receiving agents to validate integrity and terminal receipts before accepting a state.
Read on @Medium: medium.com/@anna.stolbovskaj…
10
18
127
10,027
You should try out our Popcorn Browser - the best way for humans and agents to collaborate.
- An AI Agent can use the browser programatically
- A Human can use a remote view to intervene
... All of this happening inside a a TEE, so your credentials are safe.
[New] Give your personal agents access to an isolated browser with a simple MCP. It can send you a link when it needs your help.
How I use this every day with my agent -
- I use @NousResearch Hermes over Signal
- I ask it to get me a reservation at Una Pizza
- Una Pizza reservations start at 9.00am sharp
- At 9.00, my agent fills up the form and finds an available slot
- Sends me a link to the browser where I can complete the reservation with my card
I still don't have a reservation because I'm too slow to respond to my agent. I'm the bottleneck.
3
4
461
Attack attempts continue to happen in DAOs with abandoned multi-million dollar treasuries.
The capture of decentralized organizations and protocols is a reality - and it will continue to happen if projects don't prioritize the security of their treasuries and contracts.
That's why we built @anticapture
@AmpleforthOrg was the target of a governance attack attempt this past weekend.
An address submitted a proposal requesting a supposed DAO payment of $2.5M to a grantee, the project's entire USDC treasury. The project exists and is discussed on the forum, so it could slip by at first glance.
However, in the last few hours, the proposal that would have executed the attack was canceled.
The address that submitted the proposal was receiving a delegation from another wallet. That wallet held 87K $FORTH, with 70K $FORTH being the minimum required to submit a proposal. Today, the owner of those $FORTH decided to sell their entire position in the governance token.
@AmpleforthOrg governance has a mechanism that allows anyone to cancel a proposal when the submitter holds less than 75K $FORTH. Seventy blocks after the $FORTH sale, an address canceled the proposal that would have stolen the $2.5M in USDC.
In parallel, a proposal to move treasury administration to the Ampleforth team's multisig entered voting. It was submitted one day after the attacker's proposal and will begin being voted on tomorrow.
If approved, Ampleforth would move administration of its treasury (via Timelock) to a 2-of-5 multisig, thereby preventing any $FORTH holder from moving the funds currently held in treasury.
That would prevent further attacks, but it would also end the utility of $FORTH - using the DAO's money.
1
4
235
We've been notified of an issue blocking voting on select deployments related to last month's roll out of @CoinbaseWallet 's support for EIP-7702.
We are looking into the issue and expect an update before Noon ET.
We suspect the problem is limited only to new wallets that opted into their smart-account feature, which became mandatory for new accounts created after 2026-08-15.
If you hold your private key, any other wallet is expected to work as normal.
2
2
168
Introducing the Platåberget testnet! Its our Glamsterdam focused testnet to help the community test all the upcoming changes. Please get involved with the testing efforts!
blog.ethereum.org/2026/08/17…
5
36
116
22,943
We spend most of our time on the parts of yield that don't fit in a headline number.
@definikola discussed that with @stable_summit on Stable School during @ethbelgrade: APY disclosure, curator incentives, and why we think capital alignment is non-negotiable.
Careful risk management is a hard position to hold when the market rewards whoever prints the higher number.
In this episode of Stable School, we sit down with @definikola of @BlockAnalitica to discuss what an APY number hides, why the burden of disclosure sits in the wrong place, and why curators should have their own capital behind the markets they run:
piped.video/watch?v=85RolHJm…
3
3
13
842
Equities have 32.5 regular trading hours a week. Onchain lending runs 168.
Price data doesn't stop outside those hours; 24/5 feeds exist and overnight venues trade. But regular hours carry the vast majority of volume, depth thins sharply once they end, and the primary markets sit closed all weekend. Information keeps arriving; reliable prices get harder to come by.
That gap tends to clear at Monday's open in one large step, and large single steps are what liquidation engines struggle with.
The price jump problem, in full:
2
3
12
878
Indexing stablecoins on @solana is never one filter. Three transfer instructions, two token programs, and most transfers happen inside another app like a swap or a lending deposit rather than at the top of a transaction.
So we wrote a tutorial and indexed a full day of USDC, USDT, PYUSD and USDG to go with it. It's three files, and we've included a prompt your agent can run.
docs.envio.dev/blog/index-so…
1
2
6
185
Over $6.2B in tokenized assets, powered by Proof of Asset.
See what’s in the wrapper chroniclelabs.org/dashboard/…
1
4
17
790
For institutions, not all browser extensions are built equal 🔍
⛓️ Supporting a chain ≠ connecting to its dApps
🔎 Enrichment should come from the wallet, not the dApp
🧪 Simulation that feeds policy is a control, not a display
📋 8-point evaluation framework for institutional teams
Read the guide → medium.com/fordefi/a-guide-t…
1
2
390
🗳️ New Proposal for Safenet Aegis is now live on the Safe{DAO} forum. Here is a breakdown ↓
It asks to fund the first production-ready protocol version built to secure @safe Multisig and the wider ecosystem in Q4.
Threats to self-custody are only increasing, and most defences today are still just warnings, not enforcement. Safenet Aegis is changing that paradigm.
🔰 Safenet is a decentralized network of independent Sentinels and Validators
📜 Sentinels check every transaction against a public, DAO-owned charter using their own proprietary threat detection systems
🧾 Sentinels pass verdicts. Verdicts are attested by validators onchain making security enforceable
🛡️ Enforcement happens at the account level via Safenet Guards
Proposed Independent Sentinels checking transactions: @OpenCover, @IntentGuard , @hackenclub, @hexens, @BlockSecTeam and @candidelabs
Proposed Validators: @gnosisdao, Core Contributors, @greenfield_cap, @SafeLabs_, @RockawayX and @bcap
Safe{DAO} through this proposal, owns every fee parameter on the Safenet protocol. Safenet Aegis will ship inside @SafeLabs_ multisigs in Q4 2026.
Read More ↓
Draft Proposal for Safenet is live on the Safe{DAO} forum for discussion:
It asks to fund Safenet Aegis: The first production-ready protocol release to secure the @safe ecosystem and the first, with onchain fees paid to participants in the network.
Read More: forum.safefoundation.org/t/d…
8
8
31
8,118
