I want to share an interesting pattern - get calude code on the web (or any other autonomous agent + sandbox) to do R&D on a coding question. Based on these blog posts - https://simonwillison.net/2026/Feb/6/pydantic-monty/ and https://simonwillison.net/2025/Nov/6/async-code-research/. I wanted to learn about WASM and Pyodide as I don't know much about these technologies. I asked Claude code on the web to do the following "I want to learn about WASM and pyodide, read this great blog post and then create a /tutorial dir that contains all of the resources needed for serving a pyodide-wasm-tutorial.html files, it will be served by GitHub Pages so they can load the WASM and wheel from a relative path since the .html files will be served from the same folder as the wheel and WASM file. I know a lot about python but little about these tools.", then I pointed it at the first of those blog posts. After a few minutes it created a PR into one of my repos in Github, and after turning on github pages I am able to now learn about those technologies in this mini tutorial https://ian-bmj.github.io/bmj-ai-rapid-prototype/tutorial/pyodide-wasm-tutorial.html.
Ian Mulvany
February 7, 2026
quick LLM sandboxing - thank you Simon!
About Ian Mulvany
Hi, I'm Ian - I work on academic publishing systems. You can find out more about me at mulvany.net. I'm always interested in engaging with folk on these topics, if you have made your way here don't hesitate to reach out if there is anything you want to share, discuss, or ask for help with!