@ontosdk/next
v1.6.7
Published
Extracts semantic Markdown from React/Next.js pages for AI Agents
Maintainers
founder-ontoReadme
@ontosdk/next
Optimize your Next.js site for AI agents in 2 minutes
The Onto SDK automatically converts your Next.js pages into AI-friendly Markdown and intelligently routes bot traffic to optimized endpoints. Reduce token usage by 95% and improve AI agent comprehension.
Features
- 🤖 AI Bot Detection: Recognizes 21+ AI crawlers (GPTBot, ClaudeBot, Perplexity, etc.)
- 📄 Automatic Extraction: Converts React/HTML to clean, semantic Markdown
- 🚀 Edge Middleware: Sub-10ms routing on Vercel/Cloudflare Edge
- 🔍 AI Discovery: Auto-generates
llms.txtfrom config - 📊 Analytics: Optional Control Plane for bot tracking (premium)
- ⚡ Zero Config: Works out-of-the-box with sensible defaults
Quick Start
1. Install
npm install @ontosdk/next2. Initialize
Run the initialization command to automatically generate your onto.config.ts and middleware.ts:
npx onto-next init3. Configure Build
Update your package.json:
{
"scripts": {
"build": "next build && onto-next"
}
}That's it! Your site is now optimized for AI agents.
How It Works
Build Time
The onto-next CLI scans your .next/server/app/ directory and converts rendered HTML pages to clean Markdown files in public/.onto/.
Runtime
The Edge middleware:
- Detects AI bots via User-Agent or Accept headers
- Rewrites requests to serve pre-generated
.onto/*.mdfiles - Returns optimized content in <10ms
AI Discovery
When AI agents request /llms.txt, the middleware dynamically generates it from your onto.config.ts file—no static file needed.
Components
OntoHead
Auto-discovery component for AI agents. Injects link tags into your page head.
import { OntoHead } from '@ontosdk/next/components';
export default function RootLayout({ children }) {
return (
<html>
<head>
<OntoHead />
</head>
<body>{children}</body>
</html>
);
}OntoProvider
Wraps your app and automatically injects AI discovery tags with dynamic paths.
import { OntoProvider } from '@ontosdk/next/provider';
export default function RootLayout({ children }) {
return (
<OntoProvider baseUrl="https://example.com">
<html>
<head />
<body>{children}</body>
</html>
</OntoProvider>
);
}Configuration
Environment Variables
# Optional: Enable Control Plane analytics
ONTO_API_KEY=onto_live_xxxx
ONTO_DASHBOARD_URL=https://app.buildonto.devonto.config.ts Schema
interface OntoConfig {
name: string; // Site name (required)
summary: string; // Brief description (required)
baseUrl: string; // Base URL (required)
routes?: { // Key routes
path: string;
description: string;
}[];
externalLinks?: { // External resources
title: string;
url: string;
description?: string;
}[];
sections?: { // Custom markdown sections
heading: string;
content: string;
}[];
}See LLMS_TXT_GUIDE.md for complete configuration documentation.
Detected Bots
| Company | Bots | | :--- | :--- | | OpenAI | GPTBot, ChatGPT-User, OAI-SearchBot | | Anthropic | ClaudeBot, Claude-User, anthropic-ai | | Google | Googlebot, Google-CloudVertexBot, Google-Extended | | Perplexity | PerplexityBot, Perplexity-User | | Meta | Meta-ExternalAgent, Meta-ExternalFetcher | | Others | CCBot, Bytespider, Applebot-Extended, cohere-ai, YouBot |
Benefits
For Your Site
- Reduced Load: AI bots get static Markdown instead of triggering server renders
- Better Indexing: AI agents comprehend your content with 95% fewer tokens
- Edge Performance: Middleware runs in <10ms on the edge
- SEO+: Following emerging "Agent SEO" best practices
For AI Agents
- Clean Data: No React boilerplate, Tailwind classes, or layout noise
- Structured: Markdown with proper headings, lists, and links
- Fast: Instant responses from edge cache
- Discoverable: Standard
llms.txtmanifest for site navigation
Control Plane (Premium)
Unlock analytics and dynamic features:
- Bot Analytics: Track which AI agents visit your site
- Context Injection: Append dynamic "hidden prompts" to specific routes
- Route Monitoring: See which pages AI agents access most
Examples
- Demo App - Full working example with config
- onto.config.example.ts - Configuration template
- LLMS_TXT_GUIDE.md - Complete llms.txt guide
API Reference
Middleware
import { ontoMiddleware } from '@ontosdk/next/middleware';
// Export as default middleware
export { ontoMiddleware as middleware };Config Types
import type { OntoConfig, OntoRoute } from '@ontosdk/next';Utility Functions
import { generateLlmsTxt } from '@ontosdk/next';
// Generate llms.txt content
const llmsTxt = generateLlmsTxt(config);TypeScript Support
Full TypeScript support with comprehensive type definitions for all exports.
Requirements
- Next.js ≥14.0.0
- React ^18.0.0 || ^19.0.0
License
MIT
Support
Made with ❤️ by the Onto team