> ## Documentation Index
> Fetch the complete documentation index at: https://docs.livepeer.org/llms.txt
> Use this file to discover all available pages before exploring further.

# Features and AI Integrations

> The Livepeer documentation includes several powerful features designed to enhance your experience and make information more accessible.

## Search

### Built-in Search

The search functionality is integrated into the header of every page:

* **Semantic Search** - Understands context and meaning, beyond exact keyword matches
* **Instant Results** - See suggestions and results as you type
* **Page Previews** - Quick preview of matching content before clicking
* **Keyword Enhancement** - Pages include metadata keywords to improve discoverability

### How to Use Search

1. Click the search bar in the header (or press `/` to focus it)
2. Type your question or search term
3. Browse results or select a suggestion
4. Use the AI Assistant option for conversational queries

<Tip>
  Search results are ranked by relevance, with pages containing your keywords in metadata prioritised.
</Tip>

## AI Assistant

The AI Assistant is integrated with the search functionality and provides intelligent answers to your questions.

### Capabilities

* **Answer Questions** - Get answers about Livepeer Protocol, products, and ecosystem
* **Explain Concepts** - Understand complex topics with clear explanations
* **Find Documentation** - Discover relevant pages and guides
* **Code Guidance** - Get help with code examples and implementation

### How to Use the AI Assistant

1. Click the search bar in the header
2. Type your question in natural language
3. Review the AI-generated response
4. Follow links to relevant documentation for more details

<Tip>
  The AI Assistant uses the documentation content as its knowledge base to provide accurate, up-to-date information. For the most current details, always refer to the official documentation pages.
</Tip>

### AI Integrations

The documentation is designed to be AI-friendly and integrates with:

* **OpenAI** - For AI Assistant functionality
* **Claude** - Alternative AI integration support
* **Other LLMs** - Structured content optimised for AI parsing

### Contributor Research Workflow

Docs maintainers also have an experimental research workflow for factual review.

It is designed for situations where a page may contain:

* stale thresholds or limits
* contradictory setup guidance
* time-sensitive business or support claims
* repeated facts that appear across multiple related pages

Use it when you need evidence-backed review instead of structural lint:

* [Research and Fact Checking](./research-and-fact-checking)

### AI-Optimised Content

The documentation is structured to be easily consumed by AI systems:

* **Semantic Headings** - Clear, descriptive headings that convey meaning
* **Structured Metadata** - Comprehensive metadata for better AI understanding
* **Machine-Readable References** - OpenAPI specs, JSON examples, and structured data
* **Clear Explanations** - Concise summaries and explanations optimised for LLM parsing

## Navigation Features

### Tab Navigation

Content is organised into tabs based on user roles:

* **Home** - Overview and getting started
* **About** - Core concepts and protocol details
* **Community** - Community resources
* **Developers** - Building on Livepeer
* **Gateways** - Gateway operations
* **Orchestrators** - Orchestrator setup and management
* **Delegators** - Staking and delegation
* **Resources** - Reference hub

### Sidebar Navigation

* **Collapsible** - Maximise reading space by collapsing the sidebar
* **Hierarchical** - Organised groups and sections
* **Anchors** - Quick access to Quickstarts and Reference Hub
* **Breadcrumbs** - Always know where you are

### Version Switching

Access both current (v2) and legacy (v1) documentation:

* **Version Selector** - In the header, switch between versions
* **Preserved Content** - All v1 content is maintained for reference
* **Clear Indication** - Current version is marked

## Interactive Elements

### Tabs

Within pages, tabs separate content by context:

* Different operating systems (Linux, macOS, Windows)
* Different workflows (AI vs Video, on-chain vs off-chain)
* Different user types or use cases

### Views

Custom views show different content based on:

* Operating system
* User path or journey
* Configuration type

### Steps

Sequential instructions use the Steps component for:

* Installation guides
* Configuration processes
* Setup workflows

### Card Groups

Visual groupings for:

* Portal pages with key entry points
* Resource hubs
* Related content collections

### Callouts

Important information highlighted with:

* **Info** - General information and tips
* **Tip** - Helpful suggestions
* **Warning** - Important cautions
* **Danger** - Critical warnings
* **Note** - Additional context

## Feedback Mechanisms

### Page Feedback

Provide feedback on any page:

* **Thumbs Up/Down** - Quick feedback on page helpfulness
* **Comments** - Share specific feedback or suggestions
* **Issue Reporting** - Report errors or outdated information

<Tip>
  Feedback mechanisms may vary based on Mintlify features. Check the bottom of pages for available feedback options.
</Tip>

## Automations

The documentation includes multiple automation pipelines to ensure accuracy and reduce manual overhead:

### Data Fetching

* **External Documentation** - Automatically fetch and embed specs from GitHub
* **API Documentation** - Generate API docs from OpenAPI specifications
* **Dynamic Data** - Fetch and display GitHub releases, forum posts, blog posts
* **Code Information** - Automatically update version numbers and code examples

### Content Generation

* **SEO Metadata** - Automatically generate SEO tags for all pages
* **API References** - Generate API documentation from OpenAPI specs
* **Component Examples** - Maintain up-to-date component examples

### Future Automations

Planned automations include:

* **Language Translation** - Automatic translation into multiple languages
* **AI Prompt Pages** - Generate quickstart guides from prompts
* **Feedback Loops** - Integration with Discord and GitHub for community feedback

<Note>
  The research workflow is advisory-first. It complements the existing MDX, style, and link pipelines instead of replacing them.
</Note>

## Docs Structure Features

### Automatic Page Index Generation

The repository includes an internal docs feature that keeps `v2/` index files synchronised with the current folder and markdown file structure.

What it generates:

* `v2/<top-level-section>/index.mdx` for each direct child folder under `v2/`
* `v2/index.mdx` as the root aggregate index

Format rules:

* Root-level markdown links appear first at the top of each top-level folder index.
* Folder and subfolder names are rendered as headings.
* Links are rendered in markdown link-list format (`- [Title](/v2/home/mission-control)`).
* Links include a warning marker when a page is missing from `docs.json` navigation:
  `- [⚠️ Title](/v2/home/mission-control)`.

Validation and cleanup rules:

* `docs.json` is treated as a read-only route allowlist for warning markers.
* Matching is strict after normalisation (remove `.md`/`.mdx`, remove trailing `/index`, trim trailing `/`).
* Nested `index.mdx` / `index.md` files under top-level sections are automatically removed in `--write` mode.
* In verify mode (no `--write`), nested indexes cause a failure until removed.

Automation behaviour:

* Runs in pre-commit when staged files include `v2/` changes:
  `node operations/scripts/generate-pages-index.js --staged --write --stage`
* Supports manual verification/rebuild:
  * `node operations/scripts/generate-pages-index.js`
  * `node operations/scripts/generate-pages-index.js --write --rebuild-indexes`

<Tip>
  This is a repository documentation feature (internal indexing), not an external automation workflow.
</Tip>

## Accessibility

The documentation is designed with accessibility in mind:

* **Keyboard Navigation** - Full keyboard support
* **Screen Reader Support** - Semantic HTML and ARIA labels
* **High Contrast** - Readable in both light and dark themes
* **Responsive Design** - Works on all device sizes

## Downloadable Documentation

The documentation is structured to be:

* **AI-Friendly** - Optimised for AI parsing and integration
* **Machine-Readable** - Structured formats for programmatic access
* **Exportable** - Content available in formats suitable for AI training and integration

## Next Steps

* Learn how to [Use the Documentation](./documentation-guide) well
* Discover how to [Contribute](./contribute-to-the-docs) and provide feedback
* Explore the [Component Library](./component-library) for developers