2024/SD/site-building-tools

Site Building Tools was a session at IndieWebCamp San Diego 2024.

• Archived from: https://etherpad.indieweb.org/site-building-tools

TODO: add video once archived

---

IndieWebCamp San Diego 2024
Session: Site Building Tools
When: 2024-12-07 14:50

Participants

• Peter Kaminski
• Al Abut
• Template:herebox
• ilovecomputers
• Angelo Gladding

Topics Overview

• Facilitator: Template:herebox
• also interested: Peter Kaminski, Anthony Ciccarello, benji, Angelo Gladding, Al Abut, ilovecomputers

• Notebooks like Jupyter, "live" pages
• Static site builders
• Claude and other AI tools for programming

• What are "Notebooks" and how do they differ from wiki / blog / static sites. Examples are ragt.ag, Jupyter, Observable, MATLAB, Hypercard, etc.

• James says hello from Scotland! Have a wonderful IWC!
  • Thank you, James! - pete

Personal Stories

Al Abut

• Static site with RSS, webmentions, Astro framework, https://astro.build/
• Anthony POSSE into static from bridgy fed

Template:herebox

Notebooks / "live" pages, personal notes at https://herebox.org/notebooks

Related tooling

• observablehq and examples
• OpenAI "canvas" - https://openai.com/index/introducing-canvas/

• https://www.joshwcomeau.com/javascript/modulo-operator/
• https://explorabl.es/
• https://brilliant.org

Peter Kaminski

• Example of Massive Wiki static site -> https://ai101.peterkaminski.wiki
  • Use Obsidian + Obsidian git plugin as an editor for Markdown
  • "Obsidian is a really simple IDE for Markdown files."
• Netlify toml file will publish massive wiki as static site

Using LLMs to write code and make websites

• "Jankifiers" -> https://ai101.peterkaminski.wiki/jankifiers
• Claude works well for developing front-end with React and Tailwind

• Work with o1 model to write specs
• Ask for feedback on the specs
• Ask for LLM to write a new spec based on improvements / conversations
• Ask for LLM to build based on the spec
• Tip: Big input context, small output context ( 128k vs 8k )
  • Go a head and send non-human sized input for review
  • Ask for additional detail on output ( e.g. "Make complete output files" )
  • Ask for it to "continue" if it hits a limit on output
• Tip: LLMs are stateless and will re-read all of the input on each new request

Tools

• Massive Wiki Builder - build & deploy static HTML site from markdown wiki files
  • https://github.com/Massive-Wiki/massivewikibuilder
• Obsidian
• Claude
• Netlify, Vercel
• Glitch

See Also

• 2024/SD/Schedule