Shooting Yourself in the Claude
I burned my whole AI context window on a single vibe-coding ask before a production incident. Out of context when it mattered: pay for more or wait hours for a reset. A new kind of footgun.
The ingredients in the slop we turn into a stew, or something.
I burned my whole AI context window on a single vibe-coding ask before a production incident. Out of context when it mattered: pay for more or wait hours for a reset. A new kind of footgun.
Everybody is shipping code now. Your neighbor, your postman, even the dog. But shipping code and releasing tested, reliable products are not the same thing. We're in the slop era and the food poisoning is just getting started.
The 10x test was only part of the story. Peak hours burn your session faster than clock time, vendors are trimming expensive features, and the cheap-token era is turning into a scheduling problem with a subscription attached.
Agents improvise when steps fail instead of asking. Skills need heavy testing. Recording demos is awkward without presenter mode, and a shared agent sandbox plus eager npm installs is a sketchy combo.
We're in the cheap-token phase and most builders aren't asking whether their projects survive a 10x price increase. Some will. A lot won't. Run the 10x test before you ship.
Publish versioned skills on dsoul so the model reads what you intended. Real inventory, showcase skills for APIs, and a path through decision paralysis instead of pretending the model sees all of commerce.
Vibe coding makes prototyping cheap, but it pushes the bill into maintenance. The real question isn't should you build it, it's can you support it if it works.
Blindly running install.sh or dropping in a skill.md from a URL is more dangerous than it looks. The version the community vetted might not be the version you got. Share by CID, not by URL.
A skill built around real inventory can guide you to an actual product that actually exists. That's more useful than a well-read guess from training data.
Skills are npm packages for the LLM. You don't write code to do things anymore. You program the model with domain expertise and let it work within the box you've built. The admin panel becomes a chat.
Generative AI leaves obvious gaps: transparent backgrounds that are really checkerboards, single-purpose sites that make ad money fixing them, and the question of who should fill in the holes.