Basic Usage Guide

This guide will walk you through a common scenario for using the Docusaurus LLMs Builder plugin, assuming you have already completed the Installation and Basic Configuration steps.

Our goal is to configure the plugin to process your main Docusaurus documentation and your blog posts, preparing them for potential use with Large Language Models (LLMs).

Scenario: Processing Docs and Blog Content

Let's say your Docusaurus site has:

1. Standard documentation in the docs folder.
2. A blog in the blog folder.

You want to use the LLMs Builder plugin to create consolidated text files from these sources, which might then be used for tasks like:

• Feeding into an LLM for question-answering over your documentation.
• Generating summaries.
• Performing content analysis.

Configuration Example

Here’s how you might configure docusaurus-plugin-llms-builder in your docusaurus.config.js for this scenario:

// docusaurus.config.js
module.exports = {
  // ... other configurations
  plugins: [
    [
      "docusaurus-plugin-llms-builder",
      {
        version: "2.0.0",
        llmConfigs: [
          {
            title: "Website Content for LLM",
            sessions: [
              {
                type: "docs",
                docsDir: "docs", // Points to your main docs folder
                sessionName: "Main Documentation",
                sitemap: "sitemap.xml", // Assumes sitemap.xml exists at the site root
                patterns: {
                  // Optional: exclude specific folders or files
                  // ignorePatterns: ["docs/archive/**"],
                },
              },
              {
                type: "blog",
                docsDir: "blog", // Points to your blog folder
                sessionName: "Blog Posts",
                rss: "atom.xml", // Assumes atom.xml exists for your blog
              },
            ],
            generateLLMsTxt: true, // Generate a concatenated llms.txt
            generateLLMsFullTxt: false, // Optionally, generate a more detailed version
            extraSession: {
              sessionName: "Key Project Links",
              extraLinks: [
                { title: "Project GitHub", link: "https://github.com/your-repo" },
                { title: "Community Chat", link: "https://discord.gg/your-server" },
              ],
            },
          },
        ],
      },
    ],
    // ... other plugins
  ],
};

Understanding the Configuration

• title: "Website Content for LLM": A descriptive name for this particular setup.
• Sessions:
  • The first session is of type: "docs". It targets the /docs directory, names this part "Main Documentation", and uses sitemap.xml to find all the document pages.
  • The second session is of type: "blog". It targets the /blog directory, names this part "Blog Posts", and uses atom.xml to find all the blog entries.
• generateLLMsTxt: true: This is a key option. When enabled, the plugin will process the content from the defined sessions (Main Documentation and Blog Posts) and generate a file named llms.txt (the exact location and content format are specific to the plugin's implementation). This file is typically a consolidation of the text from your documents and blog posts, structured for easier processing by an LLM.
• extraSession: This adds a section to your LLM-ready content that includes important external links, providing context or further resources.

What Happens Next?

After configuring the plugin and running your Docusaurus build process (e.g., npm run build), the docusaurus-plugin-llms-builder will:

1. Scan the content in your docs and blog directories based on your sitemap and RSS feed.
2. Apply any patterns you've defined to include or exclude specific files.
3. Generate the llms.txt file (and llms_full.txt if enabled) containing the processed text from your documentation and blog, plus the extra links.

You can then use these generated text files as input for your chosen LLM applications.

Using the llms CLI Command

The Docusaurus LLMs Builder plugin provides a CLI command that allows you to generate the llms.txt and llms-full.txt files manually, without needing to run a full Docusaurus build. This can be useful for quick updates or when you only need to regenerate the LLM-specific files.

To use the command, you can run it via npx:

npx docusaurus llms

This command will execute the LLMs Builder plugin based on your existing docusaurus.config.js configuration. The generated files (e.g., llms.txt, llms_full.txt) will be placed in the root directory of your Docusaurus project.

Further Customization

This is a basic example. The plugin offers many more options for fine-tuning how your content is processed. For a complete list of all configuration parameters and their detailed explanations, please refer to the Plugin Options API Reference.

You might want to:

• Specify exact ordering of documents using orderPatterns.
• Ignore certain sub-folders or specific documents using ignorePatterns.
• Configure multiple llmConfigs blocks if you have very different sets of content for different LLM purposes.

By tailoring the configuration, you can ensure that the generated text files are perfectly suited for your LLM integration needs.