Hi! I’ve got some gaps in my availability (a few hours/week) for:. 🔴 Small WordPress projects and maintenance.🔵 Plugin development.🟠 General web dev consultancy. Give me a shout if you need a second pair of hands on any tasks.

RT @jeremylovesyall:

Building & testing apps with LLMs is a bit of a chicken-and-egg situation. It's so easy to pivot functionality using AI that tests immediately become out-dated. But tests are so easily written they can almost be an afterthought to "pin" functionality in place. Thoughts?. I.

RT @hdevalence: oh, i get it now. the reason for docker is because it's literally easier to bundle an entire operating system than a single….

RT @forgebitz: because LLMs are trained on tutorials, they seem really happy to just reset your database

RT @svpino: Those who are foolish enough to trust the hype and quit learning programming deserve the future that awaits them. https://t.co/….

I discovered a beautiful little 20 minute remix album of the Severance soundtrack by Odesza - Music to Refine To (stretched out here to an 8 hour workday mix 😂). It's very chill. It makes my work feel mysterious and important.

RT @DanielXShmaniel: @hexaglab Because having an answer for every single little question makes a world uninteresting.

I’ve been getting into Diablo IV finally and oh that’s a lot of fun isn’t it. Surprisingly casual for a game like that, easy to just drop in and click click smash smash click click. Really satisfies an itch.

One neat thing about CSS is how it lends itself to becoming almost a markup language for LLMs. So easy to describe a layout to LLM using grid notation:

Granted I'm still at the very low end of the Scoville scale. But I am now immune to all the hot sauces Tesco sells and I need more.

I'm getting back into hot sauces again. Word of advice: don't get into hot sauces. You'll want them with everything. Sandwich? Hot sauce. Coffee? I could have hot sauce in that. Ice cream? Yep, get the bottle out.

My current personal project is trying to link up AI stuff in such a way that I can get better visibility on "me" from the outside and tell myself to stop getting so caught up in things that ultimately have no significant long-term impact on anything much at all. I know these.

I wish LLMs had direct access to e.g. X, LinkedIn, Facebook accounts. Not so they can post for me, but so they can see what I'm up to and then send me a message "What are you doing, man? stop taking it all so seriously. It's just a website omg". Genuinely think the #1 use case.

When you read things like this, just. ignore the noise. Find a process that works for you, implement any AI into it. Doesn't matter which, they all do the same thing. (provide semi-useful answers that you'll inevitably end up frustrated with if you rely on them too much).

Also on the subject, Claude Code (and similar) any good memory MCP servers? Currently I find that e.g. CLAUDE\.md is parsed instantly but any memory solutions take ages reading/processing and using up tokens. Seems CLAUDE\.md (the main "memory" file) is processed on the fly.

Big issue with Claude Code is that the terminal isn't exactly the best place for uploading, dragging/dropping files. Anyone found an easy way around that?.

After using Claude Code for a few weeks, the idea of using an LLM that *doesn't* have access to my filesystem feels lame and weird.

Unforeseen drawback of utilising AI more for automating tasks - the gaps while waiting for agents to complete their li’l jobs means more time for me to make too much coffee.