Vibe Coding
give in to the vibes and share stuff you vibe build
Andrew Jiang pfp

@ok

If you’re /vibecoding and want to join the channel, reply or message me!
60 replies
3 recasts
35 reactions

Daikie.eth pfp

@daikie

After having installed openclaw on my 4th device I have come to the conclusion these LLM's make setup procedure include a random bug to remind their user what it's like to pain takingly copy past code fixes back and forth to get the thing to boot up properly 😆
0 reply
0 recast
1 reaction

Papajams pfp

@papa

got round to setting up a clawdbot note that its super slow using just KIMI x NVIDIA testing with opencode next hoping for FREE + SPEEDY
3 replies
0 recast
11 reactions

Daikie.eth pfp

@daikie

My experience with C̶l̶a̶w̶d̶b̶o̶t̶ ̶M̶o̶l̶t̶b̶o̶t̶ OpenClaw so far
1 reply
0 recast
3 reactions

Aaron Elijah Mars pfp

@aaronjmars.eth

for all the new vibecoders, here are some repos you can play with & have fun very quickly 👇
0 reply
0 recast
2 reactions

GEAUX pfp

@geaux.eth

If you're into vibecoding apps and you don't know what a topology map is, congratulations- you're me literally 2 minutes ago. Let me save you HOURS of repeated prompting buy copying the following text: "Moving forward I want you to stop treating each feature-related request as isolated. Instead, I want you to understand the full topology of how features flow through the app. When I say "fix this feature" you should already know that means: The current functions of the feature. The current functions of the feature across the various apps it's experienced on. How these app experiences demand different things from our code based on the unique environments of the app(s) where the feature is deployed. Currently, you don't have that map in your head. You're discovering it reactively as I remind you of these things. The solution: You should create a feature topology document that maps out: All feature entry points - every way content enters the system All platform contexts - every environment the app runs in All content types - documents, audio, video, images, etc. that are relevant to the feature and the app The topology map now becomes a checklist you consult whenever I ask for a change. Instead of me telling you "don't forget to check how that's implemented there as well" - you should check the map and see it yourself. This topology check needs to be an automated feature moving forward." Use this next time you're working with @claudeai Grok or ChatGPT or any other LLM. Happy coding!
2 replies
2 recasts
8 reactions

llbxdll pfp

@mrbriandesign

I spent 1/2 of 2024 vibecoding and deployed several web3 apps. Also taught myself pixel art and 3D design not bad for a 50 year old screw-up from upstate new york, living in japan https://pixel-world-marketplace-git-main-mrbdesigns-projects.vercel.app/
6 replies
5 recasts
34 reactions

Aaron Elijah Mars pfp

@aaronjmars.eth

why is everyone suddenly excited about vibe coding?
2 replies
0 recast
2 reactions

allyourbase pfp

@allyourbase

Nice read from Chris Messina: "Code as commodity" https://medium.com/chris-messina/code-as-commodity-b9b7492dc4eb
0 reply
0 recast
0 reaction

na pfp

@na

K-clouding my dataset rn
1 reply
0 recast
2 reactions

Sol Su 🌞 pfp

@sirsu.eth

The one thing I would leave for vibe coders to consider is that you’ll run into walls the more complex your ideas are. And you don’t realize the scale of technical debt you’re incurring because there are fundamental programming concepts you just don’t know that you don’t know. Don’t let it discourage you, but if you allow yourself to learn some of the concepts, it will make your approach to vibe coding all the more exciting. Your prompts are better, you become more discerning, and you don’t just accept what the tool spits out. Competency is still vital to making good products and experiences.
3 replies
1 recast
9 reactions

na pfp

@na

Occurred to me that coding agent training, no matter what experts employed for new data, might be more incentivized to return successful results than failure. This is a mismatch with real conditions, where tokens are limited per reply, and operations should fail. Doing stuff should fail given insufficient resources. Perhaps allocation of resources and acknowledging failure (both current agents are bad at) are reflections of meaning-making.
1 reply
0 recast
0 reaction

Andrew Jiang pfp

@ok

0 reply
1 recast
10 reactions

ting.⌐◨-◨ pfp

@ting

@ok (and vibe coders) would love your thoughts 💜 can you recommend a release flow for @shira?
2 replies
0 recast
3 reactions

na pfp

@na

The correct prompt
2 replies
0 recast
2 reactions