Ricky Flim
128 posts
Ricky Flim
@0x180db
in-the-trenches dev, event-driven enjoyer, poking at gpgpu
Katılım Nisan 2022
168 Takip Edilen68 Takipçiler
Ricky Flim retweetledi
If you're like me and learn through katas, check out github.com/0x180db/cuda-z…. First five exercises are ready. No local GPU needed, thanks to @modal. Extra challenge: finish without burning through your full starter credit 🙂
English
i did the evm debugger with symbolic execution (not oss) and the open source usage metering engine for billing - unimeter.io on zig, the next goal is CUDA kernels. My work uses rust but heart is with #zig
English
People frequently ask me for visualization tools for Quint. But I don't believe in a good one-size-fits-all solution.
The ITF format is great tho, so it's easy to build custom solutions.
The software industry is changing and we need more people to be aware of what is possible.
Quint@quint_lang
Ever asked an agent to visualize a Quint trace your way? The best visualization is always going to be something domain-specific. And LLMs find it easy to understand ITF traces produced by Quint (see --out-itf). Unicode, image, HTML... just ask for whatever you prefer.
English
@alexzoid @Certora github.com/0x180db/awesom… Let's combine the lists. Feel free to copy over any entries from my repo that you don't have, and then I'll delete mine
English
Just kicked off my curated list of premium resources for Formal Verification with @Certora Prover (EVM, Stellar, Solana).
Hand-picked, always updated, all in one spot. Bookmark this treasure, anon!
github.com/alexzoid-eth/f…
English
built a symbolic evm debugger + fuzzer (poc). funny how last year this felt like magic
English
@muratdemirbas Formal verification scales poorly due to the exponential state explosion and the undecidability of non-linear arithmetic. It is only effective for critical components, not for the entire system as a whole.
English
If agents generate code 100x faster than humans, we need a verification harness that runs 100x faster for realizing Agentic AI.
But integration testing is slow and serial. Doesn't this imply that Formal Verification (not just unit testing) is a hard prerequisite for Agentic AI?
English
Ricky Flim retweetledi
Yo, I decided to learn a bit more about Quint (quint-lang.org), but I missed something like the GraphViz integration with TLA+
I vibecoded a tool that let you explore Quint's trace with a UI, showing the state transition and other related data.
quint-visualizer.noghartt.dev
English
the reason i’m deep in evm and symbolic execution: black hats don’t hunt manually anymore.
pashov@pashov
🚨A smart contract exploit from 1hr ago just drained ~8536 ETH, worth around ~$26M The code was live on Ethereum for almost 5 years already. Source code unverified - only bytecode is public.
English
lmk when the mainstream train arrives. ticket's ready
martin.kleppmann.com/2025/12/08/ai-…
English
@cantinaxyz @BNBCHAIN @lista_dao @HashDit I have not received my competition reward after five months. How can I contact support?
English
.@BNBChain’s low fees and high throughput change security assumptions in ways many teams still underestimate.
On December 11, 2025 at 10:00am ET, join Cantina, @Lista_dao and @Hashdit on X for a live Space on what “ready to ship” security looks like.
English