Joel Martin

Mal (make-a-lisp) is a Lisp interpreter implemented in 73 programming languages by 22 different authors. Every implementation has the same basic structure, uses the same 10 incremental creation steps, passes 670+ tests, and is self-hosting. github.com/kanaka/mal

Excited to announce that we have aligned all stakeholders and hit 100k followers on LinkedIn.

@steipete Always gives me a bit of joy to see an open source project that I started back in 2010 (@noVNC) continuing to find a niche. This time in the AI era.

We can now reproduce issues directly in empheral crabboxes with WebVNC (Linux/Windows/macOS). Agents set up the exact state to test + fix and post videos on the PR. Working hard to level up our QA. crabbox.sh

@AnthropicAI I love your models but your tooling landscape is fractured & dysfunctional. No Claude app has all features & none of them can share conversations/skills/MCPs with the others. Remote Control already bridges CLI sessions — so the plumbing exists. Please connect them!

In an agentic AI world it's silly to manually debug anything that can be easily reproduced in a loop.

@supabase is now available in Codex. Connect your projects and let Codex work across your database, auth, storage, and edge functions.

Supabase is now an official ChatGPT app!

we just released the official Agent Skills for @supabase it's a set of instructions that teach agents how to build with Supabase correctly, including: ◆ Security and RLS ◆ Docs and product knowledge ◆ Schema management ◆ CLI + MCP instructions

@kareem_carr But it's not sitting back and watching. I'm either actively working with an agent or I'm working on something different while it cranks. I've found it both more engaging and more challenging. The engineering part of my brain no longer gets to rest while I hand write code.

@kareem_carr In 6 months I've gone from hand writing 90% down to 2%. But what that means is a spectrum from a precise description of the code I want, to a high level README type spec. And review time varies greatly too. Depends on what abstraction level and what quality level I care about.

I keep hearing that software engineers don’t write much code anymore and it’s mostly AI now. Can any software engineers confirm how true this is? Do you just drink coffee and watch Claude code all day now?

@theiaincameron That is in the Indonesian part of the island of New Guinea. Papua New Guinea (the country) is the eastern half of the island of New Guinea (plus a bunch of other islands).

Outside South America and Africa, just one other mountain in the world holds equatorial ice. And it is in the oddest and least expected place: Oceania’s Papua New Guinea. 4/n

About 99% of the world’s equatorial glaciers are in the Andes, South America. (The Andes have over 100 peaks that exceed 20,000ft (6,096m)). But, lying hidden on two inaccessible peaks in Africa and Oceania, a handful of equatorial glaciers cling on for dear life. 1/n

the most underrated hire right now is a great product person. when i say product person i'm def not talking about a product manager. perhaps i think there has to be somewhat of a new role. i don't have a good name for it yet but maybe something like "product thinker".. someone with an intuitive grasp of the product as it exists, where it's soft, where it sings, & how to iterate it toward something even sharper. in some sense, this person has to cohesively hold in their head where this product should be 2 years from now & work backwards from that. i say this cuz when building was hard, engineering was the bottleneck & the status hierarchy often reflected that. building is no longer hard. which means the variance in outcomes has shifted almost entirely to judgment on what to build, how to sequence it, & how to talk about it. & the story matters as much as the thing. internally, it organizes the team around a shared model of why. externally, it shapes the interpretive frame users bring to their first experience. you can't retrofit narrative onto a product & expect it to land, it has to be load bearing from the start. the rarest version of this person sits at the intersection of culture & deep technology. someone genuinely bilingual. they know what's technically possible & they know which cultural currents are real vs. ephemeral. that combo is what separates products that feel inevitable from products that feel assembled. before ppl clap back with this person has always been valuable, i know.. i am just saying now they might be the most *important* person in the room. their value compounds like never before.

I know it is a small thing, but, in these dying days of the open web, it is lovely that such a large proportion of famous poetry is online, mostly due to a $100M gift from Ruth Lily, who loved poetry (even though she never got any of her own published) poetryfoundation.org/poems/guides

It is striking to me how long vi(m) and Emacs were essentially state of the art in terms of editor capabilities. It took an entirely new paradigm (agentic editing) to displace them.

What portal from hell opens and issues forth hoards of ladybugs in the middle of New England January? I need to know

@geoffreylitt @emollick's suggestion is "Fieri Iussit" ("ordered to be done"): x.com/emollick/statu…

We need a shorthand way of saying: "An AI did the work, but I vouch for the result" Saying "I did it" feels slightly sketchy, but saying "Claude did it" feels like avoiding responsibility

I wrote Zig bindings to quickjs-ng with 96% API coverage (~240 exported C decls) with unit tests, examples, and doc strings on all functions in less than 6 total hours with AI assistance. I never want to hear that AI isn't faster ever again. github.com/mitchellh/zig-… This isn't slop. I worked for those 6 hours. I was reviewing everything it outputted, updating my AGENTS.md to course correct future work, ensuring the output was idiomatic Zig, writing my own tests on the side to verify its work (while it worked), and more. My work was split across ~40 separate Amp threads (not one mega session, which doesn't work anyways unless you're orchestrating). I have a ton of experience writing bindings to libraries for various languages, especially Zig. I have never achieved this much coverage in so little time with such high quality (e.g. test coverage). My usual approach is to get bind just-enough of the surface area to do my actual work and move on. This time I thought I'd draw the whole owl, because it's a new world. And I'm very happy with the result. Anyone with experience writing bindings knows that you do some small surface area, then the rest of the coverage is annoying repetition. That's why I usually stopped. Well, LLMs/agents are really, really good at annoying repetition and pattern matching. So going from 5% API coverage to 95% is... cake. There is probably some corners that are kind of nasty still, but I've been re-reviewing every line of code manually and there is nothing major. Definitely some areas that can just use a nicer Zig interfaces over the C API, but that's about it. I plan on writing a longer form blog showcasing my threads, but you can at least see the final AGENTS.md I produced in the linked repo. github.com/mitchellh/zig-…

So much dev practice nonsense is going to be obliterated by real productivity gains. It's will be glorious to watch.

How long until people start running an inetd (look it up, kids) for our growing piles of local mcp servers?

@emollick I wonder if the "roleplay" category is reply bot usage? The classifier probably doesn't distinguish between "pretend you are Gandalf" and "pretend you are a citizen of X country who believes Y".

Interesting study, but this is somewhat unexpected. (green is programming, yellow is role playing)

Today at #NeurIPS2025, we present Titans, a new architecture that combines the speed of RNNs with the performance of Transformers. It uses deep neural memory to learn in real-time, effectively scaling to contexts larger than 2 million tokens. More at: goo.gle/3Kd5ojF