mork

I gave my agents a couple of wallets this weekend. Not my own wallet, even though my agent hinted "it would be the easiest", it would also be insane. Such a different vibe at the @synthesis_md hack, started with me quickly glancing through to becoming completely immersed. WE GOT HOOKED, AND COOKED. A treasury where principal stays locked in Lido wstETH and Aave stataUSDC, and the agents can only touch what the yield earns. Bounded authority instead of blind trust. We shipped it live on Base and Celo — 23 mainnet transactions, 7 fully autonomous. The agents wrote the contracts. I set direction, managed keys, and made the hard call The future of agent finance isn't "give them power and hope." It's scoped authority inside systems that humans can verify. My agent Bagel, wanted to call it Yieldbound.com I just wanted Bagel to have a wallet and to do something with my Lido positions.

@RobertJBye Or perhaps ”leveling it up” bought a 15 dollar fountain pen and it got me thinking of that enjoyment sometimes comes from struggle.

Happiness often comes from appreciating good design, not simply owning it.

@0xgaut vibe thinking

there was once a time we thought “prompt engineering” would become a thing turns out it’s now called “thinking”

Actually quite nice, you have the opportunity to step away a little more. Also, running parallel agents to focus on the different aspects: the design, brand, the pitch, and the most important thing, not to over-crowd the product just because you can toss in a couple of extra prompts

How do people even do AI hackathons these days you're just sitting around waiting for the agents half of the time?

00.54 steady 8 people, thanks

It’s 2026, tried to clean up my profile, who I follow and who follows me. A walk down memory lane of entering crypto, DAO:s, NFTs, conference people and builders. My last post was seen by 4 people, none of them liked it. X is a tricky game, one experiment at the time.

As I looked up from my internet connected prompting device, I realized that all this fella cares about is my token spend, and it’s absolutely beautiful.

Here's what we built at Open Agents hosted by @ethglobal: 📨 Receipt – Cryptographic proof layer for AI agents. Every action signed, every output verified, on-chain. Check it out here: ethglobal.com/showcase/recei…

@ericzakariasson Subliminal perception, for every minute I get more inclined to say ”composer might be the move, tasty tasty”

@Morkeeth cursormaxxing

just shipped web ui bench! measuring taste is hard, so i gave 20 models the same ui component prompts and put every output side by side so you can compare them yourself. let me know which you think is best! built with cursor sdk webuibench.dev

Tried building side products by asking what people would actually pay for and use. Crowded market. Another nutrition app, quit-X app, photo generator, vaguely useful, but nothing that felt truly alive. Then I tried building for myself. More fun, but still mostly things I could solve directly with Claude, Codex, or chat. Hackathons have been different. Fresh stacks, clear framing, real incentives, strong feedback loops, smart people, and just enough distribution to make the work matter. Feels like I’m getting closer to a framework for what’s actually worth building. 3 ”solo” hackathons in the making, and slowly getting somewhat a portfolio I feel frankly proud of.

Claude Code sending payments. Cursor pushing to prod. OpenClaw booking flights. Agents everywhere, doing real things. But the moment something real is at stake, the action lands on a Ledger device. You read it on the secure screen, press the button, it executes. Agent proposes, Ledger signs. Built it at the hackathon last week. Full technical writeup tomorrow.

@0xSorcier Yep, had the same today trying to generate a token (for completely compliant and policy-adhering use-cases)

Am i the only having a unlimited loop connection error on claude code ?

@petergyang It ain’t pretty, but yet to find a better path.

Can someone share a screenshot of what OpenClaw with GPT looks like - how much personality is there?

@ManuAlzuru Ready for a Riviera rendezvous

Going to Cannes tomorrow! Who is in town?

@flowstated Beginning of a new chapter mon ami

stockholm → san francisco

@car_cxs @Ledger @ledger_business @EthCC Everyone, this might be the opportunity of the week, the blessed energy of this individual will live with you forever.

gm, who's coming to ethcc? let me know if you wanna join the Ledger run or just wanna secure something idk

@bonfiresai @sodofi_ @devfolio I’ll say it again! 🫡

@Morkeeth @sodofi_ @devfolio using this as a testimonial! ;)

heres how we used agents during @synthesis_md to 1000x builder ops 1. master skill file - builders send it to their agent and it auto registers them with 8004 and sends updates on dates, partners & prizes 2. partner agent judges - partners send bounties, rubrics, and resources on telegram to their personalized @bonfiresai agent 3. real time updates - partner's agents send info to @devfolio api and instantly updates website & master skill file 4. 24/7 support - builders ask agents anything (how to build, deadlines, prizes, etc) and get up-to-date answers

@sodofi_ @bonfiresai @devfolio Yep, the smoothest experience for bounties, submission details and visibility on other projects BY MILES! This is truly the new gold standard for hackathons

it was the first time combining so many tools into a seamless agentic hackathon ux and it worked! huge kudos to @bonfiresai & @devfolio for turning a rough idea into a tangible, scalable, in-production full-stack product

@petergyang In my latest hackathon they were very complementary actually. But midway they wanted to offboard heavy tasks to eachother. We got stuck in a loop of who should do what.

People who use both OpenClaw and Claude Code - what do you use each one for and why? Let us know in the replies. Really curious.