LLMs.txt
NotForm publishes documentation in formats designed for AI assistants. Use the index first, then fetch only the pages relevant to the task.
Usage
For most requests, point the assistant at llms.txt. It lists the NotForm documentation sets and links to the raw Markdown for each page.
https://notformdocs.vercel.app/llms.txt
Ask the assistant to follow the links for the APIs it needs. For example, a dynamic array form needs <NotArrayField> and useNotForm.
Raw pages use this URL pattern:
https://notformdocs.vercel.app/raw/docs/<section>/<page>.md
Focused pages are best when the request concerns one component, composable, or workflow. They keep the assistant's context relevant and make it easier to identify the authoritative API.
Full Documentation Bundle
Use llms-full.txt when the assistant cannot follow links or the task spans installation, composables, components, and Nuxt integration.
llms-full.txt is large. Prefer llms.txt plus focused raw pages for a single form or component so the assistant does not process unrelated documentation.Prompt examples
- Focused component task
Read https://notformdocs.vercel.app/llms.txt first. Then fetch the NotArrayField and useNotForm raw pages before implementing a dynamic list of email addresses. Use only the documented APIs and explain any assumption.
- Focused composable task
Read https://notformdocs.vercel.app/raw/docs/composables/use-not-form.md before changing this form's validation or reset behavior. Match the existing API and do not invent options.
- Broad task
Read https://notformdocs.vercel.app/llms-full.txt before configuring a Nuxt form with server errors, dynamic arrays, and custom field components.
Recommended context flow
Give the assistant the application context after it has read the relevant NotForm pages. Include the schema, framework, package versions, current code, expected behavior, and exact error. Ask it to cite the NotForm page it used and distinguish documented behavior from assumptions.
Do not include credentials, private form submissions, production data, or private endpoints in a prompt or reproduction.