Skip to content

Use with AI agents

Coding agents such as Claude Code, Cursor or Codex write better fframes-kit code when they read the reference instead of guessing props. The site gives them the same docs you read, in forms they handle well.

Replace the trailing slash of a docs URL with .md:

https://fframes-kit.pages.dev/docs/math/function-plot/ the page
https://fframes-kit.pages.dev/docs/math/function-plot.md the same page as Markdown

The Markdown has the explanation, a link to the rendered gallery clip, the example code and the API table of every field. The menu at the top of each page copies it, opens it, or starts a chat about it in Claude or ChatGPT.

File What it holds
/llms.txt An index in the llms.txt format: every page with a one-line description and a link to its Markdown.
/llms-full.txt Every page in one file, in sidebar order. Large: give an agent a page or two instead when you can.
/llms-small.txt Every component in a few lines: what it is, its cargo feature, its page, and each field as name: Type = default with its description.
/components.json The same catalog as JSON: name, summary, feature, page and Markdown URLs, gallery scenes and clips, source location, fields and methods.

Point an agent at llms.txt first; it finds the pages it needs from there.

The repository has an Agent Skill for making videos with the kit: the component model, timing, theming, picking a component, the explainer kits, formulas, and checking a video without watching it.

Terminal window
npx skills add khanhthanhdev/fframes-kit

Or copy skills/fframes-kit from the repository into your project’s .claude/skills/ (or .agents/skills/). Agents that change the kit itself read AGENTS.md at the repository root.

Make a 20 second fframes video with fframes-kit that explains the derivative of x².
Read https://fframes-kit.pages.dev/llms.txt first, then the Markdown of the pages you use.
Use only components and fields that the docs list. Check the result with the video's
`strip` and `inspect` commands before you finish.