0x2d66...640d

People often crap on Perplexity, but it is very aggressive in how it uses search. Sonnet 4.6 on both Perplexity and Claude, and Perplexity gave us way better results, and I let Gemini be the judge :)

so @dethective isn not really active here, but on x, he posted something interesting https://x.com/dethective/status/2025542867989475706 couldn't help but run the chap thru repnet, and of course, not surprised https://x.com/bytebot/status/2025564710301835469

also, farcaster isn't the same without @0xbaran.eth also known as juliettemeon her x account has definitely been somehow taken over. she's the only iranian i know that didn't come back online - so consider this worrying :(

don't see any talk here about flying tulip or sonic - $ft - or andre cronje anyway, the repnet council of LLMs have some analysis also, tokens are pricey. https://byte.github.io/repnet.github.io/ https://x.com/bytebot/status/2025508292794192098

repnet thoughts 1. lazy use of search + ai can lead to massive false positives. imagine being *node vs *block? so we have to be real careful to verify and never get confused. 2. how truly anon are you or do you think you are? and if you're running a claimed agency as an anon, do people hire you, or do they just take your word for it? "Moderate risk driven by anonymity and unverifiable claims; no documented financial harm or criminal activity." is currently what we see 3. in a world where we can now really go deep into someone, do we see more anons? the longer you've been around, there are bound to be skeletons in your closet. burn notice?

We all know @dwr is good, and the council of LLMs agreed. RepNet. A working name, but as mentioned, the better Ethos. AI-first reputation intelligence. Evidence-backed. A multi-LLM council. https://byte.github.io/repnet.github.io/

interesting to see wikipedia ban archive .is because the site itself tried to ddos Jani Patokallio's blog. it is strange, but i actually know jani, even made a video of him back in the day! now, even i wonder who's behind this archive site many use https://arstechnica.com/tech-policy/2026/02/wikipedia-bans-archive-today-after-site-executed-ddos-and-altered-web-captures/

Great people are not born great; they grow great.

ok, this super sucks, lost 6 figures due to a fake google ad for uniswap https://x.com/ika_xbt/status/2024513850133840183 ad in question: https://x.com/ika_xbt/status/2023884594119119240 zachxbt is not wrong, these ads are not just google, but facebook, and there was a recent study about how much facebook profited from bad ads https://x.com/zachxbt/status/2024707598419915004 i'm happy to say that Mitsukeru LinkSentinel protects you against this We look at adurl which is treated as a redirect param in heuristics. Ad scanning uses the full risk pipeline, also does RDAP/MX fallback - ad wrappers always get strong checks like regular links

Very good read on the annoyance economy - plenty of tooling can be built to solve for this https://groundworkcollaborative.org/work/taking-on-the-annoyance-economy/

honestly, this is what social media is all about - and what farcaster was to me, and still is to me. celebrate the moments! don't let the algorithms grind you down.

sometimes we get carried away with building, and business talk, we fail to realise that people are having real life changes, and you know, we should be celebrating it, not just focusing on building (and i can be incessant with messages about tech+strategy). i for one can say there have been a few things lately - quite a few funerals (last week, my mom's cousin was only 45 - his final rites were even on valentine's day, what an expensive season to go), some life moments, etc. but there is a disease to somehow wake up early or push thru the night to still sling something productive. probably should stop that and just be present in the moment(s). its not about touching grass, its about stopping to smell the roses.

RepNet demo: live, deterministic trust scoring from evidence (projects + people), explainable score drivers, real-time updates, and static dossiers for deep review. AI-first reputation layer for humans, projects and agents: multi-LLM council reads everything (on-chain, socials, courts), builds living dossiers, verifies creds, tracks rebrands/wallet rugs, weekly re-ratings, on-chain evidence + real fight-back (not just denial). Roadmap: V1 demo, V2 hiring/interview trust, V3 broader trust infra.

Mitsukeru has a new website https://mitsukeru.ai/

EVMBench from OpenAI is exciting. It will expand to other forms of smart contracts/programs. Beauty of open. Work alongside Tempo. So thanks @dwr and team https://openai.com/index/introducing-evmbench/

♻️ I just transformed my Warplet into a Buttlet #4703 🍑 Transform and mint yours 🍑 Only 0.00037 ETH

Why compromise cline to install openclaw? There is clearly more waiting to happen https://github.com/cline/cline/security/advisories/GHSA-9ppg-jx86-fqw7

Reproducible builds are important, but what about reproducible videos? No surprise video is about a better ethos, reputation layer stuff that I've been hacking on in my spare time. No manual clicking. No editing timeline by hand. How it works: - Playwright drives the real app flow end-to-end - Narration generated locally using macOS say - On-screen actions synced to segment durations - ffmpeg composes the high-resolution video, voiceover and transcript automatically Basically, I can get this output on every single run. A demo that can be reproducible and regenerated deterministically every time.

Happy Lunar New Year. Mitsukeru coded update. Trezor scam via QR code... https://x.com/cointelegraph/status/2023457537941549057?s=12&t=iLSaYO2O7t6yPdmEXeqMdA Very common parking scam via QR code (millions lost this way yearly) https://x.com/officer_secret/status/2023593200032629040?s=12&t=iLSaYO2O7t6yPdmEXeqMdA Good news? Mitsukeru LinkSentinel on iOS Safari and the Chrome plugin protects you, even if Google didn't also pick it up.

read about this bad chrome plugin (lol) good news, mitsukeru lerg protects against attack. on macos anyway. some gaps on windows, still alpha, so will fix that. got claude opus 4.6 to help w/the review :) https://www.foxnews.com/tech/fake-ad-blocker-breaks-pcs-new-malware-extension-scam