CyphrGM

The points page on the Hyperliquid docs was just silently taken down, as was the copy on their support page saying the points program ended in November 2024. These aren't random decisions. Either S3 is very close, or we just got confirmation it's never coming. 100% or 0%.

Vibecoding protip: remember to swear profusely at your agent whenever it does stupid shit. Works everytime.

Lol. This is just sad at this point. Opensea replying to random airdrop posters begging them to use their mobile app.

gm. + // gvibe.

The Vibestarter marketing team (aka me) is so fucking goated. Someone should fund those guys.

If you are vibecoding literally anything using Claude, I highly recommend you apply to their latest Hackathon. I was given $200 in credits just for applying. Application process was pretty cool too, you have a conversation with Claude describing your idea. If he likes it, you

Welp. Had I known Claude design was under a separate usage limit I would have been a bit more measured with my experimentation. Oh well, lesson learned. Cool product either way.

Holy shit, the new Design tool from Claude is actually insane. It's finally able to product marketing collateral in one-shot without much refinement. The fact that you're able to upload your entire design system is clearly what's making the difference. Love this.

Everything is about to absolutely send it when US markets open. +10% $BTC kinda day imo.

Opus 4.7 leading on agentic coding benchmarks is great news for vibecoded projects but it also widens the trust gap. The better the agent, the less visible the human's fingerprint on the code. That's fine as long as there's provenance infrastructure keeping pace. We're

Whether intentional or not, there is absolutely some truth to models degrading over time. With that in mind, and with Opus 4.7/ GPT 5.5 likely dropping this week, you should be actively creating a list of high risk/value tasks to get done as soon as they are live.

The latest version of Claude Desktop let us see model usage over time. I have never once purposefully chosen Haiku as a model, and yet it represents over 30% of my token usage since Feb 20th. Did Anthropic make sub agents default to Haiku? This would explain why there was

Thank you to everyone who helped break my new thing ❤️

Don't fall for it guys. Anyone who replies to this post will get banned.

The capital to fund builders already exists, it's just sitting on the demand side. Communities that would actually use the products have money and conviction but no way to deploy it into early stage projects with any real accountability attached. When the supply of institutional

I find it funny how many people are convinced that Hyperliquid would just burn a significant % of the remaining $HYPE supply. I don't think people realise just how much thought goes into tokenomics before they are decided. Teams don't randomly allocate supply just to burn it.

I think Anthropic might be fucking with us for April fools. Just asked Claude to make a mock-up for a pie chart and this is legitimately what it came up with. Vibecooked.

Challenges are back baby.

Blast from the past. RIP 🥲

Just had my first 'scary' experience with Claude. Context: I have strict rules on when things get pushed to staging/production. Claude always confirms before doing these things. But in a recent session it just went ahead and tried to give itself permission (with reasoning).