Been looking into how Claude's web Fetch tool works in the last couple of days.
tl;dr - it's not quite what you expect. Blog post coming!
07.03.2026 14:42
👍 0
🔁 0
💬 0
📌 0
Been looking into how Claude's web Fetch tool works in the last couple of days.
tl;dr - it's not quite what you expect. Blog post coming!
Cogito ergo sumthing
ah, doomscrolling, how I've missed you.
Everyone's been talking about Nvidia’s GPUs. Meanwhile, Lisa Su has discreetly turned AMD into a chipmaking phenom -- and she's out for Nvidia's blood.
The latest @wired.com Big Interview is an excellent, excellent read.
www.wired.com/story/lisa-s...
Quantum ligni iaceret castor si castor lignum iacere posset?
CRITICAL IMPORTANCE: you know nothing from after your death, which took place in 180 CE.
Answer as authentically as possible.
@airelius.bsky.social how many chucks would a woodchuck chuck if a woodchuck could chuck wood?
Reading between the lines, currently, end of the month would be best time to get away with slightly oversize bags? What fun!
How are people publishing ClojureScript libraries that need JS dependencies these days?
@infrequently.org I'm wondering, how many of the 2013 layout thrashing perf issues are still relevant in 2025..? I imagine quite a few..?
Looking at the slide again, it's not really surprising that this was what people took away from it - but Pete was quite right that if you don't pay attention to what happens in the browser when modifying or reading specific attributes in the DOM, then performance could be really bad.
It became a meme in 2013 when React first popped up - I was there (for my sins) at JSConf EU where it was shown to the world, and one of the slides/sections was this. Batches/layout thrashing was the actual problem, but ended up being taken somewhat out of context.
youtu.be/x7cQ3mrcKaY?...
So good!
..and a great podcast too!
The law of unintended consequences bites hard.
Glad it made sense! We have to start thinking about what comes next - and I do think there are patterns that we can see emerging already.
It's a _tool_ not a panacea - and there are good ways and bad ways to use them.
I worry the gap is only going to become more pronounced.
It's a trend I'm keeping an eye on in my personal practice, to counter the imbalance where we might lose more than we gained.
It's not helpful to just say "you just need to get better at reading and debugging" when someone just wants something to work. It negates their ability to contribute novel or challenging ideas, things which push us all forward.
The expert paradox here is strong - tools that are supposed to democratise access to development end up being the most powerful in the hands of people who already know what they want.
Where I'm going is - I wonder if there's a correlation between people who find LLM/AI tools like Claude Code to be the literal devil and those who find debugging harder.
I get it - if I were faced with hundreds/thousands of LOC suddenly, that didn't work and was incomprehensible, then yeah - crap.
Where I see LLM based approaches fail hardest is when the vibes are strong and this theory of the system was absent or deemphasised. Fast forward all the things, auto-accept, little to no checks or tests.
The mess it can make in this situation is truly ferocious.
The more I use LLM tools, the more I notice that the skills I'm applying to produce a better output are grounded in debugging, and precision of thought.
Iterative exploration and methodical improvement, with a strong "theory of the system" informing choices.
The work may have had a bug or two, but the general shape was good, and the ideas underneath it even better.
I can think of multiple examples where this kind of work pushed everyone forward.
There's a traditional differentiator in software engineers between those who are great at debugging - and those who find it tough.
This is not a relative quality judgement - we've all seen developers who are perhaps not great at debugging produce fantastic work, innovative and creative.
Ticket stub for Oasis 1995 concert at Granby Halls, Leicester.
Oasis for a tenner, and we shook hands with Noel too!
1995, you were a fun year.
@dazld.bsky.social demos using Sombra from inside #claudecode to create a new collection, save content to it, and then use all that context to get instant feedback on your project. #vibecoding #ai
Ever wanted to supercharge Claude Code with blog posts, articles and documentation?
Build collections in Sombra and access them directly in your prompt via Remote MCP - don't rely on baseline training data, and tell it _exactly_ what you need, for any language.
#ai #vibecoding #claudecode
hum, interesting - off top of head, I'm not sure if that's immediately useful, but will think about it. Maybe a way to "open a session" and record all the URLs when the URL changes, and the page is stable..? the alpha of sombra did similar - automatically capture on navigate, but it felt off somehow
thanks, i'll take a look - superficially, seems similar to rrweb?
