This site is also available as markdown for AI agents. For a complete index of all pages, see the site index at /llms.txt

AI side project2026–presentEN

toonstrip, Live Comic Strips Drawn from Data

Solo builder (side project)

View in 한국어 →

An open-source library, extracted from Grues in Comic, that renders a comic strip (characters, poses, emotions, speech balloons, backdrops) from data you supply, as a live canvas widget for Astro, React, Vue, or plain HTML.

While building Grues in Comic, the part that drew the comic panels turned out to be useful on its own: it didn’t care about Zork at all, only about who’s in a panel, what they say, and how they feel. toonstrip is that half, pulled out and generalized into a library anyone can drop into a page.

You describe a strip as data (panels, characters, an emotion for each line, the dialogue), and toonstrip draws it on real <canvas> elements that re-flow as the page resizes, rather than as a flat image. It deliberately doesn’t generate anything. Writing the dialogue is up to you, by hand, by script, or by asking an LLM yourself. toonstrip only renders what it’s given.

  • Framework-friendly. A framework-free custom element sits at the center, with thin wrappers for Astro, React and Vue, and a no-build HTML example.
  • Asset packs. The original 15-character Comic Chat roster ships as the default pack, and a packager CLI turns your own character art into a new pack.
  • Agent-ready. An included agent skill teaches an AI coding agent to write and validate a strip document.

Like Grues in Comic, it was built through AI-agent pair-programming. Eleven packages are on npm, and the comic strip in this site’s Grues in Comic post runs on it.

← Back to portfolio