If you believe that the Ethereum ecosystem deserves high-quality, independent, open-source security tooling — built by people who care about the craft and commit to maintaining what they ship — then we're building the same thing you want to see in the world. Please share 🫶

Anyone have any questions about formal verification? Post them below and I'll get your answers from the experts ;-)

We’re starting in 1 hour! Formal verification can get technical fast, so we brought the right people to unpack it. Joining us: @RaoulSaffron and Everett from @rv_inc @Fricoben from @veritylab @bennylada from @vyperlang @zoep_42 and @clonker3141 from @argotorg @vilhelm_sjoberg from @CertiK Come listen in ↓ nitter.net/i/spaces/1YGNrbXQkQyGw…

Looking forward to discussing formal verification in the Ethereum space today! Join us in 2h

Ethereum security cannot only depend on finding bugs after the fact. This Thursday, TheDAO’s Ethereum Security Space is back to talk formal verification, and how proving critical code behaves as intended can help make Ethereum safer. Join us ↓ nitter.net/i/spaces/1YGNrbXQkQyGw…

Formal verification where it matters: the Linux kernel.

Join me next week for a stellar debugger walkthrough 💛

Onchain security is everyone’s problem and nobody’s job. ETHSecurity Initiatives is how we are changing that. Propose the work. Fund the work. Build the work. initiatives.thedao.fund/

We’re hosting Stellar "Security Office Hours" over the next 5 weeks, starting September 15th. Open to all @BuildOnStellar teams, join us for free 1:1 sessions with our Soroban experts to go over any security, protocol design or tooling questions! Limited spots, RSVP below 👇 luma.com/gss14bcg

Our stellar debugger is now available on the VSCode and VSX marketplaces. Love to hear feedback from Soroban devs. We're currently planning some office hours around it. I'll keep you posted. In the meantime, feel free to reach out directly with any questions or issues you may face.

I think it's hilarious how tech billionaires keep selling us AGI when in reality "made by chatty" has become the meme for attesting stupidity. - "Man dies of eating a peanut butter sandwich" - "Oh, you're completely right. Peanut butter is indeed made of real peanuts, and you should be careful given your allergy."

Severity classification in web2: This regular expression could potentially, maybe, under the right circumstances cause non-polynomial compute time. CRIT!!! Call the FBI! Severity classification in web3: This billion dollar smart contract's upgrade mechanism is controlled by single private key held by a cool dude we danced with at a conference. Usually I'd rate this informational, but because the protocol team said governance attacks are out of scope, I'll call it invalid. Also, our dude had sick dance moves. Yolo. LOL.

Me: "Rewrite everything in LEAN4, let LLM find the proofs". The LLM:

One underrated feature of chatbots is their exceptional user interface - and I hope that many traditional apps based on graphical user interfaces will follow. I haven't seen this coming, yet it's not surprising that a conversational model that works for humans, also works well for human-machine communication. Here is a wild guess: In the next years, we will see a accelerated decline of graphical "chrome" components (e.g. buttons, tabs, paginators). What will stay (and take more screen space) are visual components that actually carry information content (e.g. diagrams, charts, info graphics, videos). Taking the speculations one step further, I can imagine we'll see a decline of keyboard+mouse devices, and an even bigger shift towards smartphones. The desktop computer will eventually become the new fax or something only used for deep-tech jobs.

EVM formal methods are hot right now🔥just added multiple bangers here: - equiVM by @argotorg - evm-asm by @pirapira @dhsorens - evm-sail by @TauLepton_ - {evm,yul}-semantics & verified yul-c by us @powdr_labs github.com/leonardoalt/ether…

Excited to announce the first release of EquiVM! ✨ EquiVM is a Lean 4 framework for formally verifying that deployed EVM bytecode faithfully implements a formal, human-readable specification, with machine-checked proofs guaranteeing the two behave the same. The proofs are generated by autonomous LLM agents and checked by the Lean kernel, so nothing about the agent has to be trusted. The important part: you can audit a contract by reading the spec instead of the bytecode, and reason formally about its behavior through the high-level representation alone. This research direction grew directly out of community feedback on act, and we're excited to see it take shape. The 0.1.0 release ships proofs for real deployed contracts: WETH9, MakerDAO Dss contracts, OpenZeppelin contracts, and more, across a range of solc versions and optimizer settings. We'd genuinely value your input at this stage. If you work on verification, security, or tooling, take a look at the examples and the trusted-base section of the README. Tell us what’s unclear, what do you think is missing, or what you’d implement differently. Suggestions for meaningful next target contracts are very welcome. Link to the release below ↓

If you are working on security tooling / services that aren't in some way scaling formal methods you are building something that is *already* or will quickly become obsolete.

Formal verification just became inevitable.

"How do you know it's vibe coded?"

Everyone's bullish on AI + Formal Verification