Tech is fun! The future is wild! Maker of stat_pg, SimpleAI, MechaRuby, and XNK Player.
Chicago area
Joined October 2026
- Tweets359
- Following44
- Followers19
- Likes227
Hey @boardyai, I'm to curious to talk to someone in cyber security about the ideas on mechalinters.com/ and their feasibility and flaws. Got anyone in mind?
If the bottleneck with AI generated code is too much code to review, then I'm just going to start reviewing more PRs until I can figure out how to reliably review code faster via better tooling.
I suspect that simply throwing more adversarial reviews at it isn't the answer.
The people have spoken.
kamal-watch's alpha version is open source: github.com/jefflunt/kamal-wa…
Deploy it right alongside your kamal server(s) and see all your services in a single dashboard.
Small demo video (no sound) of kamal-watch for those that deploy using kamal.
Should I open-source this?
Been deploying using kamal. I created this web dashboard that runs alongside it for easy monitoring, styling inspired somewhat by htop.
Thoughts? Should I open-source it? Has someone already built this?
The reason I went web-based is so I can use it on my phone as well.
If the 80/20 rule applies now (20% of people produce 80% of software), what will it look like after another year of AI adoption?
The fact that people using AI effectively can get so much done doesn't surprise me anymore, but the gap between those folks and people just 9 months behind them in adoption is absolutely nuts, and widening.
The description for how the skill works is a great, powerful breakdown of giving just enough info to a model for it to be successful. Awesome example.
I didn't expect so many people to grab my Claude skill in the first 24 hours.
many of you forked it and started building, so i'm going to keep making it better.
today i decided to build a chameleon, so here's the latest one.
how the skill works:
> you name the thing and give claude one reference picture
> claude writes the model as math first and checks every number before it draws a single part
> every part, gauge and label reads from that one model, so nothing drifts apart
> it renders stills, hunts its own bugs, then records the tour at 60 fps with sound
> you get one page you can orbit, cut open and pull apart
teardown is free and open source, link in the first reply 👇
Curious what people think about the concept of a "mecha linter" as a reliable guardrail to AI's nondeterministic nature.
The concept is written up on this website - the concept is a first draft, it's technology/stack/domain agnostic, but there is a concrete example provided of what I mean linked from here.
mechalinters.com/
simpleai, and the `/shipit` skill has become my goto quick-fix or small feature thing. When I have a very simple thing I need built that I can describe in a sentence or two, `/shipit` does it end-to-end, including deploying.
simpleai.jefflunt.com/
How many of you out there that own a bunch of domains from years ago for projects you intended to start finally have the bandwidth (w/AI, I'm guessing?) to build them?
