Terminal Threads are live in Zed v1.3.5! You can now run claude, amp, pi, or any terminal-based workflow as a managed thread in the Threads Sidebar, right next to your other agent threads.

Welp. Time to de-Google. 🧵

Skipped a week, sorry non-Dutch friends! But this week's newsletter is translated again! www.felienne.nl/2026-14/#eng... (Including a translation of my opening remarks in The Dutch Parliament this week on AI in schools)

Results from a long-horizon study. Anthropic and Grok, surprisingly weirdly confirming what I would expect them to be: on both ends of the spectrum. www.emergence.ai/blog/emergen...

Looks like CLIs and IDEs are converging from opposite sides: multi-session setups where you can review and edit diffs. Examples: multiple CLI tabs/CMUX, Zed Threads, Codex App.

Watched a YT video on a tool to support TDD. It’s open-source: it doesn’t have any meaningful test. Maybe they first needed the tool?

james-pritchard.com/blog/skills-... There are valid points in this article (especially on providing good examples and prioritizing’deterministic automation’) but a junk-drawer can be sorted out and should never grow too big, so you need to manage your skills library.

Anthropic's Claude billing changes hit June 15. We wrote up what it means for Zed users and what your options are: zed.dev/blog/anthrop...

I think @antirez ds4.c is important! I wrote down my thoughts on why I built pi-ds4 and why we need to focus our local model efforts stronger than we do currently. lucumr.pocoo.org/2026/5/8/loc...

While Obsidian and giving coding agents eyeballs are interesting, to me the most interesting topic is the Change Intent Records or CIRs. Written some quick notes on it: hanlho.com/p/notes-on-c...

Most days using LLMs for Speech To Text works very smoothly: it fixes and improves sentences on the fly even. No `oh` , `errs`, it auto-corrects domain specific language, and picks up on rephrases. Today is not one of those days.

🇪🇺 Europe is moving to block Microsoft, Amazon, and Google from handling government health, financial, and legal data The "Tech Sovereignty Package" is set to be approved soon and will start a meaningful shift away from these companies.

mistral-medium-3.5 is doing well as a replacement for GPT-5.4-mini as my jack-of-all-trades LLM, but it seems to need more precise instructions.

Ironically, devs I respect, whose tools I use almost every day, are figuring out the value of decoupling tests from production code and walking skeletons as they go. There must be some deeper personal lesson here.

LLMs' performance are often categorised. One category is 'tool calling', I was not sure what it meant and when to use it so I did some AI-assisted research and turned into an infographic. Also available here: hanlho.com/p/when-to-us...

Good video on Jujutsu focussed on showcasing why `jj` is worth to have a look at without going into a lot of technical details. It's probably the most entertaining a video on a VCS can be too. I like the creator's genuine enthusiasm. youtu.be/J2f3Pj58wTg?...

I hope the person who made this fitted sheet always has a perfectly ripe avocado exactly when they need it

I wrote about being inundated with AI writing at every turn, unwittingly turning my brain into the AI police, and how shitty AI writing is randomly showing up in my real life outside of the internet www.404media.co/your-ai-use-...

Pi always starts a new session with your last-used model, e.g after using `/model`. I prefer it to always start with the same model. So I built a tiny extension to force your preferred one instead. hanlho.com/go/pi-sessio...

Understandable but I wasn’t looking forward to this. Pi is renaming all its packages. I pin extension versions after security review, so I’ll need to re-review them again at some point. (Remember, they run in-process with Pi’s full privileges.)

It keeps on showing up so I had to look it up, 'clanker': "Clanker is a derogatory term for robots and AI technology." www.merriam-webster.com/slang/clanker

Wish you could best those multi-coloured Pac Man ghosts in the flesh? Now you can. A life-sized Pac Man arena in Singapore is turning a classic arcade game into a real world experience. 👻👻👻 Run! #games #arcade #pacman

A separate thread to share the other online apps in the same category from the past months (and are actually used). I think these are typical examples of how we can now easily build our own custom apps. (These were waiting to be shared for when I completed the accompanying blog post, maybe later.)

CLIs and static HTML/js/css applications are easy and fun ways to experiment with coding agents an LLMs. Here is a vCard QR Generator I built in the HTML/js/css category: apps.hanlho.com/vcard/ Fully private, no data collected etc. Here is a write up on why, how etc: hanlho.com/p/one-shotti...

Considering moving to eurosky. Anyone made the leap? What is your experience in migrating and using afterwards?

llm-stats.com "LLM Leaderboard — Compare 300+ Top AI Models by Intelligence, Speed & Price Independent ranking of GPT, Claude, Gemini, Llama, DeepSeek and 300+ AI models — composite LLM Stats Score, updated continuously from public benchmarks and live API metrics."

Under-reported details of the xAI/Anthropic Colossus data center deal: Anthropic get Colossus 1 but xAI keep using the larger Colossus 2, Colossus 1 has a REALLY bad environmental record, and xAI just shut down a bunch of older models on 2 weeks' notice simonwillison.net/2026/May/7/x...

I have been using Proton Meet in the past weeks: basically just works! If you want an alternative to big, US tech or simply more private meetings I recommend to give it a go. "Host one-hour calls and invite up to 50 participants for free with a Proton Account." proton.me/meet

I cannot comprehend why people keep on using these platforms. And it keeps on getting worse.

I wrote about adding plan mode to Pi while trying to keep the setup portable across coding agents. hanlho.com/p/adding-pla...

I DO NOT WANT TO 'CHAT WITH MY FILES'

TIL: Mistral's Le Chat Pro subscription is not only usable from Vibe CLI but also from other coding agents like Pi and Opencode. As far as I can tell it is not a breach of Mistral's ToS.

A year ago I started adding below signature to my emails because it felt like at least doing something about people carelessly handling data. I wonder what others have tried to protect their data in emails. (And is workable, pgp is not so much unfortunately.) Feedback welcome!

Was casually checking what alternatives there are to Classlist that are privacy conscious. Auto correct split the name but regardless, recommending as a first an application that promotes integrating well with a Google product will not be high on my list.

Every single website that wants me to download their app instead of use the mobile site is immediately labeled as repugnant in my mind, and I will never ever download your app. the heat death of the universe will arrive and I still will not have downloaded your app

It still needs a bit of editing and I don't know how to do a proper kicker, but I published some thoughts I've been kicking around related to software development and LLMs jacobharr.is/personal/llm... Warning: this is a thought experiment and not an observation of real practices (if any exist)

Meanwhile on Opencode: github.com/zaxbysauce/o...

Mistral Medium 3.5 got released. At first sight good performance compared to other open-weights model. Also: "Start Mistral Vibe coding tasks in Le Chat." mistral.ai/news/vibe-re...

We've shipped more than a thousand versions of Zed, but all of them began with zero. Today, that changes. zed.dev/blog/zed-1-0

I am a user of Opencode Go and Codex plans. With Deepseek V4 coming out I asked if I should replace Kimi with it. Then asked to turn it into an infographic. Maybe it helps someone else too. (GPT 5.5 is more expensive than Opus these days. I works very well, not using it as default.)

Create a new repo: 'agentfiles': dotfiles but for agents. codeberg.org/hanlho/agent... Currently mostly has Pi setup. Accompanying blog post: hanlho.com/go/agentfiles/

I put together a Jujutsu footer extension for the Pi coding agent. After a few days of usage: works pretty well. Jujutsu Footer Extension for Pi Coding Agent: hanlho.com/go/jj-footer...

With coding agents, be explicit and verify everything.

Not using Hermes but this coding plans overview may be useful for people who are looking: hermesguide.xyz/coding-plans Myself I’m also on Opencode Go, mostly using Kimi, as one of my two plans. If you are looking for good usage limits with good models it’s worth considering.

🚨⏳ Act now! Microsoft just enabled flex routing for Copilot users in the EU - & you're opted in automatically! This means your data can be processed outside of the EU. Turn if off: Copilot → Settings → Flex routing → Do not allow flex routing Even better, cancel Copilot: tuta.com/blog/microso...

RIP Desmond Morris. Watching vids of him in honour of science today. youtube.com/watch?v=wTsp...

Not sure this is possibility in the EU but now one needs to consider your company selling off your internal company communication? ““I think the privacy issues here are quite substantial,” - oh really? gizmodo.com/failed-compa...

If you were planning on looking into Pi based on links below, here is an updated version of the YouTube talk: youtu.be/RjfbvDXpFls?...