Alexey

Technical Director @ Klang Games
ex. Head of Engineering @chainsafeth

[Alexey @OnChainAlexey]

@carla_muc_j @boredbae98 @yourmomsnftacct @SuperteamDE @bitblondy @ApeClubDach @herdaogermany @Bjornout1 @NikeSchmidt1 @yesjess @Pizza_DAO @RarePizzas We will 🍕🤌

[Vladimir S. | Officer's Notes @officer_secret]

Just spotted one of the best resources around to understand the native design constraints (and schools) of bridge design!

Good read: https://t.co/Gws0R5xXdd

Much thanks @odysseas_eth for a nice find!

[Alexey @OnChainAlexey]

Arbitrum: froze $71M this week.
Me: freezing outside a parking garage, having just discovered its closing time the hard way.
Different scale. Same lesson about reading the fine print.

[Alexey @OnChainAlexey]

Landed in Cannes for EthCC[9] 🇫🇷
If you’re here this week - let’s grab a coffee. DMs are open.

#ethcc

[ChainSafe @ChainSafeth]

Security is our top priority. Huge thanks to the @hexens team for their rigorous work stress-testing our defences 💛

[ChainSafe @ChainSafeth]

Forest is now the only Filecoin client to support trace_call 🤯
Trace transaction execution for debugging and testing without ever hitting the network.

Read the full update👇

[Alexey @OnChainAlexey]

Really like the framing of security as minimizing divergence between intent and execution.

One missing angle is that a large part of user intent exists outside the transaction itself. Users don’t reason about raw addresses - they reason about publishers/recipients, domains, and apps. Many exploits are technically valid transactions that are semantically wrong because the action diverges from the user’s mental model of who they are interacting with.

A model I’ve been exploring is issuer-level attestations: organizations bind a domain identity to signed sets of contract addresses, adding a human-semantic projection of intent alongside simulation, permissions, and economic bounds. Not a source of truth - just another independent signal that can disagree.

Identity is socially defined and imperfect, but it encodes expectations that purely cryptographic checks cannot.

If multiple projections of intent diverge - address, simulation, behavior, issuer context, that divergence itself becomes the security signal.

[Alexey @OnChainAlexey]

Benchmarks, Forest vs Lotus: https://t.co/2t5d9nfuiw

[Alexey @OnChainAlexey]

No trust without transparency, so in this video, I explain how we conducted Forest for #Filecoin's benchmark tests.

You can follow these methods to confirm our Forest vs Lotus benchmark stats, shared in the 🧵

[Alexey @OnChainAlexey]

How do we measure Forest against Lotus?

In this video I demonstrate how we benchmark #Filecoin nodes using our Filecoin Benchmark Suite. I’ll explain the setup, resource monitoring with Grafana, and how to generate performance reports.
https://t.co/fN9jhA4AAw

[Alexey @OnChainAlexey]

While we're talking about recent benchmark numbers, it's worth noting that @getblockio had already integrated ChainSafe’s Forest @Filecoin client into their RPC infrastructure long before that.

They moved to a Rust-based stack early to lock in better performance and long-term stability for their users.

Proud to see top-tier providers hardening their stack with Forest. 🤝

[Alexey @OnChainAlexey]

Bottom line:

For general node duties, Lotus remains excellent.
For RPC API workloads, Forest leads. Forest is often dramatically faster while staying extremely lightweight.

Full Compatibility & Benchmark Report — February 2026 🌲
https://t.co/68sbsEKtko

[Alexey @OnChainAlexey]

Yesterday we benchmarked Forest vs Lotus.
And results are mind-blowing 🧵⬇️

[Alexey @OnChainAlexey]

CPU pattern follows the same trend.
Lotus maintains a heavier baseline load. Forest spikes during execution and returns close to idle afterwards.