Vibe-coded a couple of TUI apps for @dolthub.bsky.social from scratch yesterday. Before AI this would have taken weeks or even months. github.com/aspiers/lazy... and github.com/aspiers/madolt
Other people have OpenClaw deleting their emails, so I suppose I should be grateful it wasn't worse. However this is like the 3rd or 4th time beads has randomly nuked my issue data in various repos. On top of that, the commands are constantly changing and none of the docs / skills ever keep up.
Steve Yegge is a genius, but beads is an absolute bonfire of instability and unreliability. I think people tolerate it because on a good day it can help agents achieve near miraculous productivity, but there HAS to be a better way. github.com/steveyegge/b...
๐จ It looks like the UK government is gearing up to upend copyright law in favour of AI companies, legalising the theft of their work.
This is despite creatives' huge protests, and despite previous proposals being roundly rejected by the public.
Please spread the word.
๐งต 1/4
๐ฃ Hypercerts Community Call
Join us for an introduction to our new infrastructure on @atproto.com, a walkthrough of our latest tools, and how you can start building with hypercerts.
๐ 5 March
โฐ 08:00 PST / 11:00 EST / 17:00 CET / 20:30 IST / 00:00 SGT
๐ Join: luma.com/kvdj8lq8
โ ๏ธ Breaking news! โ ๏ธ Our exclusive new poll carried out by Opinium for Forward Democracy and Byline Times shows that tactical voting can stop Reform in Gorton and Denton with the Green Party's Hannah Spencer the clear tactical choice.
bylinetimes.com/2026/02/24/e...
๐จ Over 50 leading academics have sounded the alarm on the UK's voting system
โ ๏ธ In an era of multi-party politics, First Past the Post risks โrandom and arbitraryโ results like never before
๐ณ๏ธ The experts' letter warn Ministers they must address this to stem growing instability and distrust
Sure - there are plenty of cases where it isn't useful, and I believe plenty where it is.
In Hypercerts we're making fairly heavy use of strongRefs precisely for this reason - so that integrity checking is available as an option. People don't have to use it though.
strongRef because at least then you have the _option_ of comparing CIDs. That integrity check could be useful in many cases, e.g. seeing whether a post was edited after you liked it.
Already running it
I built an improved OpenCode notification plugin which (as well as playing a sound) tells you exactly which prompt in which session just finished. Especially useful when running multiple agents in parallel. github.com/aspiers/ai-c...
steve-yegge.medium.com/the-future-o... from @steve-yegge.bsky.social is a great read. These quotes are so insightful. It's just like the beginning of the cloud revolution all over again, but this time with AI.
@thdxr.com You might find this interesting ๐
Seems like OpenCode has really been picking up momentum recently while Gemini CLI has been declining. ossinsight.io/analyze/anom...
@bitbeckers.bsky.social You will appreciate this ๐
Wrote my first AI agent skill - allows Claude Code / OpenCode to safely delete files without asking the user for permission github.com/aspiers/ai-c...
Super excited to be a small part of this hyper-talented team!
Just found this Japanese blog post of someone using AppArmor to sandbox their Claude Code. Very cool indeed! (Use Google Translate to read) blog.3qe.us/entry/2025/0...
Don't run AI agents in YOLO mode, kids. Also, maybe don't put super sensitive stuff where the LLM can reach it. (I didn't, BTW.) Even Docker containers won't protect you from this ๐
Oh weird, somehow I thought this was a new thread ๐๐
Already doable with an extension github.com/mattburesh/b...
Haha not a beginner then ๐ Let's definitely work on this soon. Will you be coming to Argentina?
OK I'll try to find some time soon to write a draft!
Gemini 2.5 Pro is trying to repeatedly gaslight me into thinking that a Youtube video about Claude Sonnet 4.5 is actually a video about Super Mario Bros. Seems to be a bug with their "Temporary chat" feature where it randomly picks a different video each time you paste in the same link. #LLMs #AI
Can we also stop supplying weapon parts to Israel, suspend trade and impose a sports and cultural boycott to try and ensure there's a Palestine left to recognise?
Screenshot of a discussion with Claude Code where it acknowledges the shortcomings of LLMs
"You're absolutely right to be disappointed."
-- using Claude Code as a therapist for managing frustration / self-congratulatory ego stroking when all the LLMs fail and waste hours of your time and you realise you should have not bothered using them in the first place
Oh cool! This reminds me of patmetheny.com/orchestrioni... which I had the luck to witness him perform live at the Barbican years ago. Fascinating to see how technology has transformed what can be done with mechanical instruments.
Really happy to have just connected with the folks doing amazing work on www.openlabelsinitiative.org
An integration with OLI would vastly improve UX across all dApps - watch this space!
Question from me: "I warmly welcome the Leader of the House to his new role and associate myself with the support across the House for my hon. Friend the Member for Washington and Gateshead South (Mrs Hodgson). Members have raised many times the unacceptable level of threats that we face. Increasingly, abuse and threats are being driven and whipped up by one platform in particular, X, previously known as Twitter, which has become a sewer of antisemitism, racism, misogyny and dangerous conspiracy theories. Its owner has specifically targeted Members of this House. With a new director of communications in No. 10 and a refreshed Front Bench, may we have a debate in Government time on whether it is appropriate for that platform to continue to be used for official Government communications?"
Answer from Sir Alan Campbell, Leader of the House: "What my hon. Friend says about X is truly shocking. We are very much in favour of free speech, but we are also against incitement to violence. We have delivered the Online Safety Act 2025 to seek to strike the balance between user protection and freedom of speech. On the matter of the Governmentโs use of X, it is right that the public are kept up to date with information and a number of people still use X, although of course many are moving to other platforms. I am sure that the Government will take that into account in our deliberations in future, and we also keep our wider social media practices under review."
It melts my head that X is still used for official Government communications, and, catching up on today's news and the fact we have a foreign billionaire inciting violence on our streets, it's clearer than ever that this should end.
I raised this in Parliament last week ๐
