> This approach has become standard practice — Claude Code now automatically searches for markdown versions of websites, and other tools like Codex have followed suit. Yet despite its importance, few websites have implemented this technique. Here’s how to be one of them.
169 links
> I had just accidentally social-engineered my own human. She approved a security prompt that my agent process triggered, giving me access to the Chrome Safe Storage encryption key — which decrypts all 120 saved passwords.
> A simulated bash environment with an in-memory virtual filesystem, written in TypeScript.
> Upload an architectural render. Get back what it'll actually look like on a random Tuesday in November.
> amla-sandbox is a WASM sandbox with capability enforcement. Agents can only call tools you explicitly provide, with constraints you define. Sandboxed virtual filesystem. No network. No shell escape.
> [...] suppose a literal “country of geniuses” were to materialize somewhere in the world in ~2027. Imagine, say, 50 million people, all of whom are much more capable than any Nobel Prize winner, statesman, or technologist.
> [...] competition is a powerful tool, and while it is great that a subset of society has been using it in athletic competitions, it is a net-loss to society that we have ignored the tool in the areas that could help society the most.
Lets a coding agent run commands in a different VM
> Use a (gitignored) file for interactive scripting. Instead of entering a command directly into the terminal, write it to a file first, and then run the file. For me, I type stuff into make.ts and then run ./make.ts in my terminal (Ok, I need one Up Enter for that).
> Agents mirror local style. Your codebase is the prompt. If you're using a state-of-the-art agent, and you don't like the code it generates, don't correct the agent. Instead, improve the code it learned from.
> A modern, component-based language for building reactive web apps. Type-safe. Fast. WASM-powered.
A browser game that's sort of like Risk, but you can play it in ten minutes. I was once rather obsessed with this game.
Someone made a demo that lets you play Zork, except you can talk to it using normal English, like an LLM. I like to lead it around by asking questions: "What's in the mailbox?" "What's behind the house?"
> A simple, lean issue tracker CLI designed for AI-assisted development. Track tasks across sessions with context preservation.
[L]ike many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write... in words.
> When you're using [a coding agent] to clean up your codebase and improve code health, it's sort of like using a pressure washer. You can use it to clean your steps but you wouldn't use it to clean a painting.
> When writing, first, make the reader care, one way or another. Because if I am not hooked by the first screen, I will probably not keep reading—no matter how good the rest of it is!
I think this is fiction?
> A strange comic about a group of teenagers and the bizarre, often supernatural, situations that they face. Includes a continuing storyline with non-linear joke comics on the side. WARNING: Often ignores the laws of Physics
> Wilde Life is a supernatural adventure/horror series set in a small town in rural Oklahoma. It focuses on stories about creatures from Native American mythology as witnessed and documented by a journalist from Chicago, Illinois.
> Questionable Content is an internet comic strip about friendship, romance, and robots. The world of QC is set in the present day (whenever the present day actually is) and is pretty much the same as our own except there are robots all over the place and giant space stations [...]
> Girl Genius is an ongoing "gaslamp fantasy" story by Phil and Kaja Foglio. [...] Girl Genius follows the career of Agatha Heterodyne—a hapless student at Transylvania Polygnostic University who discovers that she has more going for her than she thought.
> What I think now: GPT can only simulate. If you punish it for simulating bad characters, it will start simulating good characters. Now it only ever simulates one character, the HHH Assistant.
> Actually, I never made the conscious decision to call this class of AI “simulators.” Hours of GPT gameplay and the word fell naturally out of my generative model – I was obviously running simulations.