llms.txt

What is llms.txt?

llms.txt is an open standard that lets projects provide structured, machine-readable documentation that LLMs can consume directly. Think of it like robots.txt, but for AI assistants, it tells them exactly what your project offers and how to use it correctly.

When an AI assistant has access to an llms.txt file, it can generate more accurate code, suggest correct API usage, and avoid hallucinating component names or properties that don't exist.

ng-oat's llms.txt

The ng-oat documentation site serves two LLM-focused endpoints. You can point any AI tool at these URLs to give it full knowledge of the library:

What's included

The llms-full.txt file contains the complete ng-oat knowledge base:

How to use it

There are several ways to give your AI assistant access to ng-oat's documentation:

Option 1: MCP Server (recommended)

The ng-oat MCP server exposes the same data as queryable tools, so your assistant can look up exactly what it needs on demand rather than loading everything at once. This is the most efficient approach.

// .vscode/mcp.json, recommended approach
{
  "servers": {
    "ng-oat": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@letsprogram/ng-oat-mcp"]
    }
  }
}

Option 2: Direct URL in prompts

Paste the URL into your AI chat to give it the full library context. Most AI assistants can fetch and parse the content:

Use the ng-oat documentation at https://ng-oat.letsprogram.in/llms-full.txt
as reference for all ng-oat components, tokens, and utilities.

Option 3: Custom instructions

Add the URL to your IDE's custom instructions or system prompt so it's always available. For example, in VS Code's .github/copilot-instructions.md:

## ng-oat Reference

When working with @letsprogram/ng-oat, refer to the full API docs:
https://ng-oat.letsprogram.in/llms-full.txt

This contains all 37 components, 87 design tokens, 300+ utility classes,
and 12 page recipes for the ng-oat Angular component library.

Option 4: Download locally

Fetch the file and keep it in your project for offline access:

curl -o llms-full.txt https://ng-oat.letsprogram.in/llms-full.txt

Some AI tools support loading local reference files, placing llms-full.txt in your project root can help.

llms.txt file structure

Here's a preview of how the llms.txt file is structured:

# ng-oat v0.2.4, Angular Component Library
> Built on Oat CSS. 37 components, 87 tokens, 300+ utilities.

## Components

### NgOatButton
- Selector: `ng-oat-button`
- Import: `import { NgOatButton } from '@letsprogram/ng-oat'`
- Inputs: variant, size, disabled, loading, type
- Example: `<ng-oat-button variant="primary">Save</ng-oat-button>`

### NgOatCard
...

## Design Tokens
| Token          | Category | Light       | Dark        |
|----------------|----------|-------------|-------------|
| --primary      | color    | #574747     | #fafafa     |
| --background   | color    | #fff        | #09090b     |
...

## Utility Classes
- Layout: hstack, vstack, center, flex, grid
- Spacing: m-0..m-12, p-0..p-12, gap-0..gap-12
- Responsive: sm:flex, md:grid, lg:hidden
...

## Recipes
- Login page, Dashboard, Pricing, Blog, Settings...

## Theming
- provideNgOatTheme({ tokens: { '--oat-primary': '#6366f1' } })
- NgOatThemeRef for runtime switching
- CSS light-dark() for automatic dark mode

MCP Server vs llms.txt

Both approaches give AI assistants ng-oat knowledge, but they work differently:

Recommendation: Use the MCP server for daily development work. Keep llms.txt as a fallback for tools that don't support MCP yet.