One cool thing about designing your vibe coding workflow for verifiability and testability is that you can switch out models and only impact performance, including switching to local models
Kobi Gurkan
@kobi.bsky.social
applied crypto, security, experimental things | zkSecurity
PSA: go.sum is not a lockfile. You never need to look at go.sum. go.mod has everything you need.
go.sum Is Not a Lockfile
In Go, go.mod acts as both manifest and lockfile. There is never a reason to look at go.sum.
words.filippo.io
One fun simile of AI producing a lot of math/code and produced proofs of correctness using formal verification is that it’s like we’re moving to an NP feeling era where the statements are math/code and witnesses are these proofs
Submitted to present in #ATmosphereConf! Thoughts on what tools and protocols we have at our disposal for expressive end to end verifiability using trusted execution environments and/or cryptography
How well it would work to paint PLA with acrylic with only sanding and without priming? Or otherwise do you know an in-office friendly and convenient priming method?
7/ Check out the details, including a deeper discussion of challenges, a prototype of Bitchat over Wi-Fi Aware, and general notes on cross-platform Wi-Fi Aware implementation 👉 grjte.sh/bitchat-wifi...
@grjte.sh’s exploration into using Wi-Fi aware for local-first chat on bitchat Wi-Fi aware has large range, simpler authentication and is becoming cross-platform supported @grjte.sh integrated it as an alternative to Bluetooth 1/2
grjte@grjte.sh · 8mo ago🧵 Could Bitchat have 5x the range and 100x the throughput for the same power expenditure? I explored how Wi-Fi Aware could improve the reliability and throughput of Bitchat and mobile ad-hoc networks in the absence of internet connectivity. #bitchat 👇
🧵 Could Bitchat have 5x the range and 100x the throughput for the same power expenditure? I explored how Wi-Fi Aware could improve the reliability and throughput of Bitchat and mobile ad-hoc networks in the absence of internet connectivity. #bitchat 👇
@brittanyellich.com and I have been talking about community modeling on ATProtocol lately and I wrote a follow-up piece to her recent post "Representing groups in ATProto".
The Community Manager Pattern - Nick's Blog
This post looks more closely at Brittany Ellich's work on representing groups in ATProtocol. It builds on earlier conversations and explores how these ideas might work in practice.
ngerakines.leaflet.pub
New blog post! ✨ I argue that AI will make formal verification go mainstream. martin.kleppmann.com/2025/12/08/a... Three reasons for this: 1. LLMs are getting increasingly good at writing proofs using proof assistants. This will make formal verification vastly cheaper than it's been to date.
Prediction: AI will make formal verification go mainstream — Martin Kleppmann’s blog
martin.kleppmann.com
Being unique can make you strong ❤️ In unique signature schemes, existential unforgeability implies strong unforgeability
doing my favorite thing of composing stuff together, and toying around with x402 and Privacy Pass to see how they can play together to introduce a sort of a blinding layer to x402 an intro to both, how they're used and a possible way to integrate them! kobi.leaflet.pub/3m3pyyctda22i
Privacy Pass + x402 = blinding for x402 - Kobi's blog
kobi.leaflet.pub
Want to understand ZODA? Tried to implement it and found yourself stumped?? This FAQ is just for you! Wrote some answers for questions I’ve seen around me and those I had myself, navigating details and tradeoffs If you have any more - let me know :)
Signature schemes are a cornerstone of modern infrastructure and we all know the common ones that sign a message, and some of you know ones that can be e.g. aggregated In more specialized scenarios, the properties needed aren’t obvious until you hit that problem yourself 1/2
Came across an interesting paper this weekend about “Early Signs of Steganographic Capabilities in Frontier LLM” Specifically they’re testing non-fine tuned models that are widely available, like GPT 4.5 1/4
Memory leaks are going to have a whole different meaning soon
You get great results from Claude Code by guiding it to generate tests for itself to verify its output and supporting it by having live data and services running locally, which it utilizes since it runs in your environment And being able to do it on your phone is the best 1/4
I rewrote my [[Community Search Engine]] note. It's still not very crisp. I include emerging tools that directionally are working on what I want to be using - @dxos.org Composer, @inkandswitch.com Patchwork (not yet public), @grjte.sh's Groundmist, and the newly released by Tonk, TonkbookLM.
Wei Jie’s write ups are among the best resources you can find for in-depth cryptography implementation topics If you’re looking to bridge the gap between theory and practice - have a read
weijie@kohweijie.com · last yr.1/ Earlier this year, Yuval Domb of @ingonyama.com discovered Logjumps — a more efficient way to do large-prime field multiplication than Montgomery multiplication. So much modern crypto relies on modular multiplication — all the way from TLS sessions to elliptic-curve based ZK proofs.
On the lookout for a one click experience to collect interesting things I come across my day, so that I could get a nice summarized digest a day after The closest I had was with X bookmarks, but that’s limited to X x.com/kobigurk/st... 1/2
Video of my talk in zksummit about Ligerito has been published! It’s about the work by Andrija and @lmao.bsky.social introducing a small and concretely fast polynomial commitment scheme Since then, a fun thing has happened — 1/3
anyone trying to do provable image transformation as in eprint.iacr.org/2024/1066 in fast proving zkVMs? It’s one of the cases a bunch of time will pass until you need to compress it for fast verification, if at all
VerITAS: Verifying Image Transformations at Scale
Verifying image provenance has become an important topic, especially in the realm of news media. To address this issue, the Coalition for Content Provenance and Authenticity (C2PA) developed a standard to verify image provenance that relies on digital signatures produced by cameras. However, photos are usually edited before being published, and a signature on an original photo cannot be verified given only the published edited image. In this work, we describe VerITAS, a system that uses zero-knowledge proofs (zk-SNARKs) to prove that only certain edits have been applied to a signed photo. While past work has created image editing proofs for photos, VerITAS is the first to do so for realistically large images (30 megapixels). Our key innovation enabling this leap is the design of a new proof system that enables proving knowledge of a valid signature on a large amount of witness data. We run experiments on realistically large images that are more than an order of magnitude larger than th
eprint.iacr.org
Leaflet Publications: blogging on Bluesky — version 0.1 is here! ✅ now: make publications, add posts, publish to Bluesky 🗓️ soon: subscribing, commenting & other social features Try it: leaflet.pub/home We'd love your feedback & ideas for how we can make this great!
trying out @leaflet.pub for my recent post, I'm really enjoying the UI a lot: kobi.leaflet.pub/3lpruvjqlhs22 just a question - wen math? would like to use it for my next post :D I also see comments are on horizon, but how does it work with unpublished drafts that I don't want others to see yet 👀
Verifiable Verifications - Kobi's blog
kobi.leaflet.pub
How are people thinking of tool use with local models? Feels to me qwen 30b a3b gets confused easily, at least when using the OpenAI agents sdk tool infra