AI assistants (Claude Code, GitHub Copilot, Cursor, ChatGPT…) don't need to read the pages like a person: the menu and styles only use up their context. That's why all this documentation is also published in Markdown, in the llms.txt format: clean text, the tables, the steps and the real code of each integration (the same file you download), in English and Spanish.
That way, when you ask your assistant to connect your website to Jazui, use the API or read your blog, it checks the documentation instead of making up parameters.
Using Claude? Install the Jazui Skill: it carries this same documentation and the procedure to connect your website.
The files
| File | What it has |
|---|---|
/llms.txt | Index in English: what Jazui is, the rules an assistant must follow and the link to every page. |
/llms-es.txt | The same index in Spanish. |
/llms-full.txt | All the documentation in one file, in English (to paste it whole into a chat). |
/llms-full-es.txt | All the documentation in Spanish. |
/llms/en/<page>.txt/llms/es/<page>.txt | Each page on its own, for example /llms/en/integrations-aspnetcore.txt. |
The plan table isn't copied into these files because it changes: they tell the assistant to read it live from GET https://api.jazui.com/api/PublicPlans.
Use it in your project
Add these lines to your assistant's instructions file (Claude Code reads CLAUDE.md, GitHub Copilot .github/copilot-instructions.md and Cursor .cursor/rules):
<!-- Claude Code: CLAUDE.md · GitHub Copilot: .github/copilot-instructions.md · Cursor: .cursor/rules/jazui.mdc -->
## Jazui
This website uses Jazui (prerender and SEO for SPAs). Before writing integration, API or blog code, read
https://docs.jazui.com/llms.txt and open the page of each topic (https://docs.jazui.com/llms/en/<page>.txt).
Never put the Jazui token in client-side code: read it from configuration or environment variables.In a chat you can also paste the llms-full.txt link or the link of the page you need, or use the Copy for AI button and paste the text.
From every page
Under each page title you have Copy for AI, which copies the page in Markdown in the language you're reading it in, and Markdown, which opens the file. Pages also announce it in their <head> with <link rel="alternate" type="text/plain">.
Always up to date
The files are generated from the texts and code of these pages and of the integrations, so nothing is written twice. After changing the documentation they are regenerated with:
python Jazui/Solution/0_Clients/Jazui.Client.Document/tools/docs_ai.py # llms.txt, llms-full.txt y llms/{en,es}/*.txt
python Jazui/Solution/0_Clients/Jazui.Client.Document/tools/docs_search_index.py # buscador y sitemap