@austincsgo_ Sounds like they're trying to solve your "recovery question" or whatever lol
English
Manu.ts
5.6K posts
Manu.ts
@Neolectron
Gilet jaune des devs. Typesafe obsessive compulsive disorder guy. https://t.co/mMl1s8gHt1 @Codinglab_io
Toulouse, France Katılım Nisan 2012
689 Takip Edilen12.8K Takipçiler
@anomalyco Iphone only, sounds like someone at Anomaly is lacking useless knowledge to build cross platform mobile apps.
English
@Console_buche Somehow supermaven still works so I never stopped using it, and yes I was thinking of the LLM access like codex / opencode / Claude subs etc
English
@TrustedSheriff Je refais pas la roue là, je reimplem pas ma propre auth, je prend juste des libraires pas usine a gaz et vachement + lowscope car je peux très facilement les composer.
Better-auth c'était un pain-point récurrent dans ma codebase et une GROSSE limitation sur pas mal de trucs.
Français
@Neolectron Oui il peut y avoir des soucis, mais de la a cracher sur la lib, je te trouve un peu too much sur ce coup.
Mais parcontre félicitations pour refaire a ta propre sauce 👍🏼
Français
C'est marrant j'étais en train de faire sauter cette lib bloated de merde aujourd'hui mdrr.
Devinez par quoi je l'ai remplacée ?
(un service Effect, et il marche mieux)
Better Auth@better_auth
Better Auth is joining Vercel. Vercel shares our commitment to keeping auth open source, framework and platform agnostic. Joining forces gives us more resources and room to focus on building the framework that much of the community relies on. Together, we’ll also be able to invest more deeply in where auth is heading: a future where agents act on users’ behalf and need secure, scoped, and revocable access.
Français
Manu.ts retweetledi
I’m convinced that a large % of programmers don’t actually like computers.
As a side effect, are also perfectly happy to throw away their reasoning to a model as soon as they can.
I don’t get it, at ALL. Don’t you *LIKE* understanding the magic of the machine?
You do realize hand-programming (I hate that I even have to specify hand now) is fun…right?
English
Manu.ts retweetledi
A brief history of neural networks, excerpt from my video "I Built an LLM From Scratch"
English
La narrative de Anthropic avec Fable strikes again.
Ça fait 3x qu'il faut absolument vite vite prendre un abo pour utiliser Fable avant qu'il disparaisse, et qu'au final ils repoussent la limite.
Tkt pas mon gars le modèle il va rester si il est renta.
FOMO strategy.
Jean P.D. Meijer ― 🇪🇺 eu/acc@initjean
tokenmaxxers are almost working themselves to death now we've created a unhealthy environment in tech and a massive fear of missing out, building doesn't mean anything if it kills you
Français
@soubiran_ @CloudflareDev Pour moi les gens dorment trop dessus, codemode c'est op jtrouve
Français
this week, it's time to talk about @CloudflareDev code mode and how insanely powerful it can be
an article is coming in the next day, and a complete series before the end of the week 🎆
English
@cnakazawa Automated prompt I would say.
There no difference between what you call a skill and just typing the content in the chatbox yourself.
Apart maybe the fact that you let your llm decide if it's worthy to inject now based on some keyword.
But it's just saved prompts...
English
Most people don’t realize that skills are just… functions. Non-deterministic markdown text, but still just functions.
English
100% agree. This is the current sweet spot.
All the skills and setups have big “I optimized my text editor” energy.
Lucas@rvcas
I don’t use loops, skills, fancy orchestration, plugins or anything like that. I raw dog the agent with its default harness, I read the code, and I’m shipping more than ever to production for real projects used by real people. Sometimes if it’s a low risk part of the application I might skip reading the code but only sometimes. I think most are either lying about the amount of value they are creating or just having fun while understanding they aren’t creating value but failing to distinguish that to their audience.
English
Manu.ts retweetledi
Corollary: top devs with the most boring AI coding setups will outperform mid devs with fancy multi-agent hundred-skill etc. setups
Operator matters more than the tools
Michael Arnaldi@MichaelArnaldi
This will piss you off but it's true
English
Je vais vous prouver que vos LLM puent la merde, que ça soit GPT 5.6 Sol nanani Fable mes couilles.
Ensuite on part sur du Effect, et je vous partage ma vision sur ce que c'est du Typescript robuste et stylé a la fois.
twitch.tv/neolectron
Français
@majidmanzarpour @MOGLRD @threejs The technology you're speaking about is creativity through constraints.
Nowadays you have almost 0 constraints, therefore you don't need to be creative to circumvent them.
English
AI game dev challenge: build a surfing wave like this in @threejs
Xbox Nostalgia@Xbox_Nostalgia
Kelly Slater's Pro Surfer (2002 Video Game)
English
@raphhaixai @iFeyz2 J'ai envie d'être de mauvaise foi et te dire que c'est pour économiser des lettres mais même pas....
Je suis rempli d'anglissismes, je prompt en anglais, j'ai tout en anglais, etc du coup ouais.. forcément
Français
@Neolectron @iFeyz2 Le terme decisionmaking me donne envie de te frapper très fort. Même si pour le reste je suis d'accord
Français
beaucoup de gens ici
null_tips ⚡@null_tips
Moi par exemple j'utilise que des IA plus intelligente que moi
Français
Manu.ts retweetledi
Today Aperture Science has officially stopped selling digital science, and we’ll only be providing physical science by hand, in exchange for cold hard physical cash.
PAY UP PEOPLE.
GIF
English
@FabriceSimonet @AlexandreChsDev Moi jvous ai trouvé drole les 2 c'est carré
Français
@AlexandreChsDev @Neolectron Nan ça va je t'ai pas insulté ça passe 😂
Français
Alors merci pour les modèles, mais désolé, c'est totalement incompréhensible pour un utilisateur lambda... Sol, Luna, Terra... A quel moment tu choisis lequel...
Français
@iFeyz2 Bof jtrouve.
Surtout que ya 0 decisionmaking contextetuel dans les training data (par definition, on print pas notre reflexion dans du text les humains en ft), et qu'elles sont même plutot bourré en volume par des trucs random de github qui ont + d'incidence sur leur decisions.
Français
