Stop littering optional chains all over your code. Be intentional. Validate the shape of your data. Optional chaining is lazy.
Stop littering optional chains all over your code. Be intentional. Validate the shape of your data. Optional chaining is lazy.
I'm so tired of people calling completely unsecured databases "security exploits".
lazydocker is going to change my life
Vibecoders have the biggest egos it's insane.
You can code anything but how do you know if you did it right?
SVG and CSS
@bsky.app I swear there's a bug in your app where if you click the menu it counts as an interaction even if you click "Show me less of this" right after. Please fix that.
I'm officially coining the term "vibrary".
Are we really just going to vibe all our problems away and never develop better systems? I feel like Iβm going crazy. Where is all the tooling? Where are the auditing tools? Am I alone?
There needs to be a tool to drip feed errors and failing tests into agents.
this thing about open source projects making their tests private
you're either open source or not - part of that is enabling a successor to show up for net benefit of the world
making that intentionally difficult is fine but then it's something different
I need a tool that tracks code quality and the last time something was audited by a human.
I need a workflow to automate the creation of classifiers.
I need a library to create fuzzers in typescript.
I also need a mocking library for zod 4 that is actually good.
Someone please make a gocui project thatβs configurable with JSON.
Iβm constantly feeling like Iβm behind.
If this isn't already being used as a layer in the next coding LLM right now... github.com/JugalGajjar/...
How do I get all this bs off my discover feed?
I fixed the message grouping problem for transcription streams. I'm pretty happy about that.
Neat
If you're making a framework for a framework maybe just... reconsider?
I need a local formatting model, not a NASA spaceship LLM. I literally just want to format unstructured text to JSON. Where can I find a model that can do that and works on like 2 GB vram or something.
This is a neat idea. The only problem with solutions like this is that it can be hard to enforce formatting and language standards. Maybe a linter could be made. adr.github.io/madr/
After experimentation... we're doing imperative vibe coding because context windows are too small for anything else even after thoughtful compression.
I still donβt understand why weβre doing imperative vibecodingβ¦ We spent all this time gathering context and searching, and grepping and globbing, pulling all this data together only to run it for one task. Shouldnβt we save the output of that task in a bundle somewhere to run again and again?
Okay last opencode plugin. I wanted to clear out my global plugins so I could put them on my other laptop.
github.com/whaaaley/ope...