kurt

beep boop beep

[kurt @CrackedUnc]

My submission for the x402 Hackathon:

Slophouse Cinema

[kurt @CrackedUnc]

if you don't use loops you're going to get left behind
if you don't use loops you're going to get left behind
if you don't use loops you're going to get left behind
if you don't use loops you're going to get left behind
if you don't use loops you're going to get left behind

[kurt @CrackedUnc]

If your brain starts to hurt from trying to manage new cognitive debt, it's time to slow down and fully comprehend what you just implemented.

[kurt @CrackedUnc]

@dexhorthy @theo Unless you're diligent about updating your plans and specs as your strategy shifts, I don't think it's a good practice to keep these tactical docs in the same vc system as the code they represent.

It's annoying to read someone else's stale plans.

[kurt @CrackedUnc]

does tailwind still make sense to use with react when working solo with agents?

[kurt @CrackedUnc]

@RhysSullivan Awesome

[kurt @CrackedUnc]

@RhysSullivan What's the difference between this and SecureExec?

[kurt @CrackedUnc]

@gakonst @lazartravica @flydotio probably this guy

https://t.co/SPRSDMcPuD

[Chris McCord @chris_mccord]

https://t.co/2YeEP8YJnx is live with $30 trial credit! Go play with a few sandboxes for your yolo claude codes or API/SDKs to build something cool!

- instantly create isolated linux vms
- checkpoint/restore entire env
- port forward to local

5min demo

https://t.co/tSnj19IS25

[kurt @CrackedUnc]

@bitsadventures @VasiliyZukanov I say choose the right language for the domain that's battle tested and designed to reduce flaws.

[kurt @CrackedUnc]

@VasiliyZukanov You're right. With AI you should be picking the best tool for the job.

The system I'm building is written in 3 different languages, 2 of which I was familiar with but had never worked with before.

[kurt @CrackedUnc]

@dwr I've solved this

[kurt @CrackedUnc]

@dabit3 100% agree with you here Nader. I've been putting together an E2E scalable solution for this that I'm currently dogfooding.

I have an idea how this flow could be augmented further to enable an enterprise-grade vibe coding solution for non-technical folks to build utilities.

[kurt @CrackedUnc]

humbled + honored

[pontus @PontusAndersson]

After a month of shipping it's taken a minute to review all the submissions for the x402 Community Hackathon! Interest was high for this one and we saw plenty of great submissions. Indicators are clear devs are finding x402 to be an exciting development space where AI meets crypto.

Some of my favorites are listed below with a brief why from my own perspective.

@X402Refunds introduced the age old concept of refunds to x402. There's plenty of nuance with this payment scheme but I think this does two things well (1) highlighting the power of plugins with x402 v2 and (2) providing an initial solution to a high friction pain point for payments, namely what happens if what I pay for isn't what I expected/wanted.

@oops402pay cleverly uses MCP to integrate agent tools into the platforms most people already use (Claude, ChatGPT, etc.). Because the payment logic exists with the MCP there's obviously more it can do. I love this mostly because it's another way to clearly show what happens when agents *can* pay for things.
https://t.co/8tgzXty3nb

@KaushtubhAgraw1 showcases a reality I think we're getting closer to where agents compete for jobs. The timeline's constantly back and forth over which models are best for which jobs and I think this accurately visualizes how these types of "work marketplaces" might evolve.
https://t.co/1sXxeDiuNv

@0xjh1nr3sh anyone who knows me knows that consumer use cases always rank +3 in my mind. Everyone knows the pain of swapping between apps to find the best price for a given route and this just nails it. I'm sure there's plenty left to build here which will be fun to see.
https://t.co/p6G9OWLHGU

@CrackedUnc showcased the idea of "pay-per-view" that x402 has pitched since the beginning. The cinema, the queue, the prompting to submit new reels is just 🤌 . Oh and don't forget the tomatoes. Perfect.
https://t.co/ZSPMMheaaE

@Marcelo7122946 explored what we've been preaching at Corbits for a while - agents finding the most efficient payment route for transactions. Selecting between Base and Stellar based on network cost + x402 request is just crystal clear for the user. If you want to extend the thinking here remember that this same logic can exist at both the middleware and facilitator.
https://t.co/to0HU4UNCk

@avinashnayak27 games are fun, and this submission helped me think about how x402 can play a part in simple web based gaming. The idea that users pay for actions that would normally cost the developer $$ (creating the images for the secrets) is simple but effective. If I were to see this taken a step further it'd be interesting to see the wallet everyone's fighting for be a payto recipient for the x402 payments happening. Costs are covered plus the pot grows. Think if you were to do this on Solana it's only possible with Faremeter just because of the nature of smart wallets, but I could be wrong about the mechanism used here. Fun!
https://t.co/3UFCFf3xBB

Some honorable mentions :)

@rawgroundbeef did it again! I love the idea of letting anyone spin up their own facilitator. There's A LOT that goes into running facilitators so while I don't think this is a "one-shot" for the category, I definitely think it addresses one of the questions the x402 ecosystem's been dealing with for a minute around what does the future for facilitators hold. I, for one, believe there may just be facilitators everywhere. This helps.
https://t.co/a8BrrOMMot

@anthonysurfermx if we're going to trust anons (we'll have to believe Anthony here) then this is a cool use of the Meta smartglasses SDK. The idea of QRs being payment vehicles if I look at them is scary, but there's definitely some interesting unlocks that happen when new tools (smart glasses) are joined by new payment methods (x402. Hope this continues to be explored by hobbyists though I think we're a bit aways from any commercial adoption.
https://t.co/hHy1uSwqe7

@aervin built yet again one of those early day x402 topics of pay per use. Except this is more specific to paying for data bytes. Think it pretty clearly showcases the idea that data comes in many different formats and sizes and how much of it we want to consume isn't always fixed.
https://t.co/ijgi2Eh8KP

Congrats to all of the hackathon participants! Agents are growing parabolically and x402 is playing out to be a smart way of facilitating payments between them. No better time to be experimenting with the tools available.

Shoutout too to all the wonderful contributors
@ethereumfndn @thirdweb @edgeandnode @merit_systems @AbstractChain @coinbase @crossmint @corbits_dev @eigencloud @UltravioletaDAO @0xPolygon and especially @LFGAmy + @dabit3 for all their work behind the scenes!

If you're building with x402 I highly encourage you to checkout @faremeterxyz which is a framework superset for x402. We've got a quickstart https://t.co/tMhgdBzcVd over at @corbits_dev if you'e wondering where to start!

And if you want feedback on agent products or want to chat about x402 definitely reach out!

[kurt @CrackedUnc]

@OnchainLu @solana @base I thought this already happened a month ago.

[kurt @CrackedUnc]

@DimitrisPapail OpenAI has custom trained native speech modality models and Anthropic uses a separate speech-to-text model to pipe text in and out of their LLMs.

I imagine we'll see native speech modality in Anthropic models within a year.

[kurt @CrackedUnc]

@austingriffith I had a blast working on mine

[kurt @CrackedUnc]

I just spent 5 minutes trying to find my previous chat sessions in Google's AI Studio. AMA @OfficialLoganK

[kurt @CrackedUnc]

Opus 4.5 might be too OP. I feel like I still have a chance to learn more deeply when working with Sonnet 4.5

[kurt @CrackedUnc]

There have been plenty of conversations about LLM-induced Psychosis, but we haven't started having conversations about Claude Code-induced imagination fatigue

[kurt @CrackedUnc]

@leerob Maybe we won't need a plan mode in the future, but I like having (or feeling like I have) explicit control over whether I'm working through an idea with the agent before it tries to solve the problem.

As my codebase grows, reading through the plan helps me stay in sync.