Vibe coded pull requests that basically rewrite every line of code in your repo. Yep, I'm not gonna merge that.
Vibe coded pull requests that basically rewrite every line of code in your repo. Yep, I'm not gonna merge that.
Not so sure about this liquid glass UI.
A bit too over done for my liking. More style than substance.
Had my first case of a prospective client get in touch about an app they βvibe codedβ (their words) and needing help getting it working.
Nah. Fuck that shit. Not touching it with a barge pole.
Ushered Terry Nutkins to his seat at the cinema in west swindon.
Funnily enough, itβs probably βtokenize itβ.
So what are you saying? That product makers will stop putting ai in their products, or theyβll come up with a more marketable name for it?
"I recommend a hybrid approach..."
Claude, just make a frickin decision please.
Question for vibe coders:
How do you create/manage your stack-specific context files (.cursorrules etc)?
- do you hand-roll them for every project?
- or get LLMs to write them?
- do you keep a context library for different stacks?
- or copy from a website?
- is this a PITA that needs a solution?
It's a beast. I live in a hilly part of the world so the 500w battery is an essential.
First ride of the year that wasnβt face-achingly cold. Even a bit of Blueskyβ’οΈ on show.
π§ Just published Anthropix v0.6. Now you can see Claude β¨thinkβ¨ about exactly how many R's are in strawberry in unnecessarily impressive depth!
github.com/lebrunel/ant...
#ai #claude #elixir
I don't think coding is dead, but I do think the role "developer" has already changed and will continue to trend towards more of an AI wrangler than a code writer. The tech will move faster than humans/teams can adapt to it... but eventually, it'll all be vibes.
bsky.app/profile/aaro...
I wrote a long and slightly rambling post about the hot new way to code that all the cool kids are doing.
2point0.ai/posts/vibe-c...
#ai #coding #vibe-coding
I wrote a long and slightly rambling post about the hot new way to code that all the cool kids are doing.
2point0.ai/posts/vibe-c...
#ai #coding #vibe-coding
Cool thats good to hear. Do you have any insight with any of the Svelte based component frameworks (eg Skeleton), and whether AI models grok them well?
Everyone I know who codes using LLMs all seem to stick to next/react/shadcn etc.
I'm wondering what people's experiences are with other frameworks, eg Vue or Svelte, or even non-JS stacks. Are the models capable enough to veer off the beaten path or are we not there yet?
Had a slightly disappointing week with a work opportunity that, well, didn't work out.
Need to do a flex build πͺπ» to get my mojo back. What should I build?
It lowers the barrier to entry, which results in a much more competitive landscape.
Soβ¦ both? π€·ββοΈ
As a brit watching reports of the LA fires, I have one question:
What is so wrong with the word "burgle" that you had to invent "burglarize"?
Nice. Can probably leave it like that for another 5 years. ππΌ
My impression from evaluating both is that langchain, being a port of the python lib, brings with it design choices from a different language and time.
AIsdk is a fresher take, cleaner api, typescript first, and built by vercel who already have a stellar reputation in js open source land.
Just a daily note - it all goes in there. Itβs an unholy mess, but Iβve tried pkms and they never really work for me.
What is success? Getting stuff built, or growing social metrics?
10 days well spent πππ
Prompt fragments in Agentflow
Squeezed in a quick Agentflow release before the weekend! π
New in this release:
- Prompt fragments - reusable, composable prompt engineering components
- Role attribute - fine-tune agent personas and behaviours
Check out the docs: agentflow.2point0.ai/guide/workfl...
I like it. Having a shit day, just bump the version.
This was good fun. I don't even know python (hence first result π³), but seems a nice way to learn some basics.
Leetle #8 2/6 6:38
[8 lines]
β οΈβ οΈβ οΈβ οΈβ οΈβ οΈ
π©π©π©π©π©π©
leetle.app
Generate structured outputs
Preload and unload models from memory
Released a nice little update to Ollama-ex. What's new?
1. Finally, generate structured outputs by passing a JSON schema map in a request (requires ollama >= 0.5)
2. Preload and unload models from memory
Code: github.com/lebrunel/oll...
Docs: hexdocs.pm/ollama
#elixir #ollama #buildinpublic
So there you have it. Be careful out there. Oh, and this seems very sensible: bsky.app/profile/bsky...
I was unaware that any of this had happened until people found my real X profile and began accusing me of being a βruggerβ. It took a while for me to figure out WTF was happening. It sucks that people lost money, but I think people buying these kinds of tokens know the game theyβre playing.