CallMeGwei.eth

2.2K posts

CallMeGwei.eth banner
CallMeGwei.eth

CallMeGwei.eth

@CallMeGwei

Director of Security Tooling R&D @OpenZeppelin Researching, discussing, developing and protecting web3 & the open economy Expediting the blockchain revolution

The Internet Katılım Ekim 2017
329 Takip Edilen1K Takipçiler
CallMeGwei.eth retweetledi
Shaw (spirit/acc)
Shaw (spirit/acc)@shawmakesmagic·
Speaking from much experience here I run 8-10 agents simultaneously but most are on cleanup duty You can have 1-2 generative features But you also need to deslop, deduplicate, implement strong typing, run tools like ruff, knip, Madge, slither, remove fallbacks and unnecessary code paths etc So yeah you shouldn’t run more than 1-2 building stuff or you’re gonna flood with slop, but you should have a lot of janitors
English
9
1
30
2.2K
CallMeGwei.eth retweetledi
Shaw (spirit/acc)
Shaw (spirit/acc)@shawmakesmagic·
99% of business don’t need an agent, they need an automation But they don’t know how to do that So they need the agent to make the automation for them 🤷‍♂️
English
42
12
138
7.9K
CallMeGwei.eth retweetledi
Gerard Chetrit
Gerard Chetrit@chetrit_·
Your editor shouldn’t care which model you run. Your agent shouldn’t be tied to your editor. Your code shouldn’t train the company selling you the tool. But when editor, agent, and model all ship from one company, all three happen. hob is the tool you have been looking for🤌
hob@hob_app

hob is a workspace for AI coding agents. Only the workspace — not the model, not the agent. You bring those. Because when workspace, agent, and model all ship from one company, your tools stop working for you. Your code becomes their training data. Models change. Agents change. Your workspace shouldn’t. one surface, all agents.

English
0
1
0
173
CallMeGwei.eth retweetledi
jchetrit.eth
jchetrit.eth@josepchetrit·
Three years ago I called my dad and told him coding had just changed forever. Since then, I've changed my whole coding setup three times.
hob@hob_app

hob is a workspace for AI coding agents. Only the workspace — not the model, not the agent. You bring those. Because when workspace, agent, and model all ship from one company, your tools stop working for you. Your code becomes their training data. Models change. Agents change. Your workspace shouldn’t. one surface, all agents.

English
2
5
21
2.2K
CallMeGwei.eth
CallMeGwei.eth@CallMeGwei·
Spent the last two years watching AI coding tools consolidate toward exactly the outcome developers should be most worried about: your workspace, your agent, and your model all coming from the same company. Built hob because I want the opposite: an independent workspace that keeps developers in control. Alpha today. DM for access.
hob@hob_app

hob is a workspace for AI coding agents. Only the workspace — not the model, not the agent. You bring those. Because when workspace, agent, and model all ship from one company, your tools stop working for you. Your code becomes their training data. Models change. Agents change. Your workspace shouldn’t. one surface, all agents.

English
2
2
3
328
CallMeGwei.eth retweetledi
banteg
banteg@banteg·
guys don't tell me your answer to 1/1 multisigs getting hacked is 2/2 multisigs. sometimes i feel this industry is incapable of learning.
English
57
66
824
55.6K
CallMeGwei.eth retweetledi
tim-clancy.eth
tim-clancy.eth@_Enoch·
Absolutely not. To sustained disappointment, not a single major Ethereum L2s is properly trustless. It has always been, and continues to be, a risk to use these trustful L2s. There are many users who stay exclusively on L1 because of exactly this risk. Were AAVE to socialize L2 losses to L1 users, it opens a much larger can of worms where any trust-conscious L1 user needs to reject DeFi which is otherwise trustless from an L1 perspective.
⟠@ryanberckmans

imo L2 Aave ETH lenders taking a concentrated loss would be among the ugliest, nastiest outcomes in our history. you're talking about kneecapping the life savings of our brothers in eth whose only crime was not being rich enough to use L1 Aave

English
14
9
186
11.9K
CallMeGwei.eth retweetledi
Erik Voorhees
Erik Voorhees@ErikVoorhees·
In crypto and defi (ie in honest markets), when a component fails, those closest to the component—whether wildly negligent or innocent victim—suffer the loss, and are burdened with that responsibility. Unequal, but proper. In tradfi and banking (ie in coercively manipulated markets), when a component fails, the entire society is forced under the burden of its resolution. Costs are socialized. Equal, but improper. The former, with time, becomes self-correcting, self-improving, and crucially, retains vitality. The latter, regardless of time, becomes stagnant and soulless, and here everyone can wallow in an equivalent grey. Any man of agency should prefer the former, taking care over that to which he is proximate. It is from this that the virtue of markets emerges.
English
135
281
1.7K
212.4K
CallMeGwei.eth retweetledi
Tyler
Tyler@rezoundous·
Opus 4.7 is insane guys. It one shotted my session usage limit.
English
603
1.3K
31.5K
1.1M
CallMeGwei.eth retweetledi
Erik Voorhees
Erik Voorhees@ErikVoorhees·
@justinsuntron That was really dumb of you to support such an obvious scam project. What were you thinking, and how can you possibly be surprised? It is politicians themselves which are the antithesis of crypto. This is 101 stuff bro
English
28
24
739
29.6K
CallMeGwei.eth retweetledi
Erik Voorhees
Erik Voorhees@ErikVoorhees·
"Every major AI platform — ChatGPT, Claude, Gemini, Grok, Perplexity — stores your conversations on centralized servers. That data can be reviewed, classified, subpoenaed, hacked, or sold. You're trusting the company, its employees, its vendors, and every government that asks. Venice is architecturally different. Prompts are not stored server-side. Chat history lives in the user's browser. And as of March 18, 2026, Venice offers something no other AI platform does: verifiable encryption."
nikshep@nikshepsvn

x.com/i/article/2042…

English
34
103
547
46.4K
CallMeGwei.eth retweetledi
Roman Storm 🇺🇸 🌪️
Roman Storm 🇺🇸 🌪️@rstormsf·
@valkenburgh Thank you for this very important work. I appreciate Michael's efforts to pursue strict and clear guidelines from the DOJ on building non-custodial, permissionless software. Thank you, Coin Center, for staying resilient and relentless in pushing for common sense.
Roman Storm 🇺🇸 🌪️ tweet media
English
0
5
45
3.5K
CallMeGwei.eth retweetledi
😈
😈@turtlekiosk·
when you watch a movie and some catastrophic failure happens in some system and the computer says like "containment breached" and you're like well how bad could it be if the UX has a specific error state for that flow
English
54
1.1K
29K
590.9K
CallMeGwei.eth retweetledi
Sabrina
Sabrina@sabrinaesaquino·
Agents can now use the Venice API with x402 and access any model with no API key required Private, pay as you go AI inference for autonomous agents
Venice@AskVenice

Happy 402 day: Venice now supports x402 – the internet-native payment standard for agentic commerce Shoutout to @coinbase, @linux, and other participants AI agents can now access Venice inference autonomously. An agent sends a request, pays instantly with their DIEM balance or USDC on Base, and gets the response in a single round trip

English
9
21
140
17K
CallMeGwei.eth retweetledi
F. A. Hayek Quotes
F. A. Hayek Quotes@FAHayekSays·
“The power that a millionaire has over me is much less than that of the smallest bureaucrat who wields the coercive power of the state, and on whose discretion it depends whether and how I am to be allowed to live or to work.” — Friedrich Hayek
English
27
866
4K
81.1K
CallMeGwei.eth retweetledi
Neil Renic
Neil Renic@NC_Renic·
First they came for the em dash and I did not speak out. Then they came for the Oxford comma…
English
220
17K
139.6K
2.6M
CallMeGwei.eth retweetledi
Tibo
Tibo@thsottiaux·
Our Codex dashboards are showing increased rate of users hitting rate limits and since we don't fully understand why I have made the cautious decision of resetting the usage limits for all plans. Enjoy. I also wanted to celebrate us finding a pocket of fraudulent accounts that we banned and have helped us regain some compute. The fight against abuse never stops, but it's important to mark the moment and make it a little shared victory.
English
1.1K
573
12.1K
1.4M
CallMeGwei.eth retweetledi
Crémieux
Crémieux@cremieuxrecueil·
It seems that if you're doing even a moderate-intensity workload, you'll hit usage limits on a Max plan in about five minutes. And in that short span of time, you might even have to restart it multiple times because of the newly-aggressive rate-limiting. Claude is unusable now.
English
18
7
503
27.4K

Keşfet

@justinsuntron @valkenburgh @elonmusk @BarackObama @taylorswift13 @cristiano @BillGates @NASA