docusaurus-plugin-llms-txt

"SEO for AI Agents"

Automatically generate llms.txt and llms-full.txt files from your Docusaurus documentation. This allows Large Language Models (like ChatGPT, Claude, Cursor, and local LLMs) to ingest your documentation context accurately without hallucinating.

📦 Installation

npm install docusaurus-plugin-llms-txt
# or
pnpm add docusaurus-plugin-llms-txt

## Usage

Add the plugin to your docusaurus.config.ts (or .js):

```ts
import type { Config } from '@docusaurus/types';
import llmsTxt from 'docusaurus-plugin-llms-txt'; // Import the plugin

const config: Config = {
  // ...
  plugins: [
    [
      llmsTxt,
      {
        // Optional: Exclude specific pages or folders
        exclude: ['**/secret/**', '**/drafts/**'],
        // Optional: Output formats (defaults to ['txt'])
        format: ['txt', 'jsonl'],
      },
    ],
  ],
  // ...
};

export default config;

Advanced Usage

Frontmatter Exclusion

You can exclude specific pages from llms.txt by adding llms: false to the document frontmatter:

---
title: Secret Page
llms: false
---

JSONL Output

If you enable jsonl in the format option, the plugin will generate llms.jsonl in your build folder. This is useful for fine-tuning LLMs with your documentation data.

Output

When you run npm run build, this plugin will generate two files in your build/ folder:

1. llms.txt: A standardized Markdown list of all your documentation pages with descriptions.
2. llms-full.txt: The full text corpus of your documentation, optimized for RAG (Retrieval-Augmented Generation) and context windows.

Contributing

We welcome contributions! Please feel free to submit a Pull Request.