Paul

[Paul @paulmanley]

@hancore_linux excellent. i like the approach.

[Paul @paulmanley]

@DrAlexTatem @hubermanlab I can personally confirm my first week at .75 my HR went up 8 bpm for about 48 hrs then back to normal. Now up to 3.5 mg and no recurring side effects.

[Paul @paulmanley]

@dhh I'm still working through larger omarchy glass project (omaistro) which will use your Quickshell/QML style to manage AI development workflow (ADW). It will launch grouped glass panes and follow the work. These panels are built to track burn rate ($$) and usage. Grok was added.

[Paul @paulmanley]

@theo transitioning from a front to back-end dev.

[Paul @paulmanley]

@theo Ditto. Even with explicit indructruons. It’s baked into the model to always end with friction to keep u going.

[Paul @paulmanley]

@dhh Playing with the alpha and not as intimidating as feared. Have LLM setting snapshots and going to town. It is a fantastic addition. The clean shells and design are chefs kiss.

[Paul @paulmanley]

@dhh I jumped the shark and pulled down alpha and loving every minute of it. Never been patient.

[Paul @paulmanley]

@elonmusk Band is getting back together

[Paul @paulmanley]

@natebjones i run all model families side x side. Opus 5 needs heavy hand, blows by specs/logs, and jumps to conclusions too quickly. Fable/4.8 is better IMO. I'm using more Fable w/ Grok 4.5 daily

[Paul @paulmanley]

@theo It's , meh, it bypasses explicit instructions in repos. Ignored the spec and blasted thru repetitive jobs already in logs. Honestly, feel like high strung junior operator with severe ADHD.

[Paul @paulmanley]

@mattpocockuk Ditto thus far

[Paul @paulmanley]

@dhh Bought my Ferrari F12 18 months ago and has more than doubled. Never intend to sell but less inclined to drive. Haha

[Paul @paulmanley]

@mattpocockuk Hypland on Linux can't be beat

[Paul @paulmanley]

Bob, my question is how often do you test? I’ve grown from just vibe coding to reading the code, but I admit I’m not very good at it. Like you, I rely on LLMs to assist with testing through reviews and validators. The real question is, how often should this be done? I feel that at least half the time, tests are run via ladders, and each session can’t end without passing the XY test. It really kills the momentum.

[Paul @paulmanley]

@dhh Crazy question but is it possible to embed Quickshell inside hyprland pane?

[DHH @dhh]

And of course, you can just add your own panels, plugins, or even entire menu bars. Quickshell is supremely agent friendly too. So easy to get great QML from the latest models.

[Paul @paulmanley]

@Houstonomics Must have counted in August.

[Elon Musk @elonmusk]

@huaijiangzhu It is pretentious and poisonous for for-profit companies to call themselves “labs” when they are actually corporations and to create a lords & peasants culture between (looksmaxxing) “researchers” & engineers when they are both, on their better days, actually engineers

[Paul @paulmanley]

@dhh I’m almost done developing “omaglass” that is a multi agent hyprland compositor that automates top tier agents working directly with each other in glass (panes). Think of TMUX/CMUX but instead using the Omarchy hyprland configs. Uses local only subs so no API calls and can bolt on open LLMs. Unfortunately I spent quite a bit of time integrating waybar and walker to set up a notification rail for UX and agent notifications. Now I’m stuck until quckshell drops, so hurry. I can wire the base omaglass to 4.0 shell and off to races. 💪

[Paul @paulmanley]

@claudeai Heck yeah. She (fable) still likes me, i know she does.