@ontosdk/next 1.3.0 → 1.3.2

package/src/cli.ts

@@ -4,6 +4,24 @@ import fs from 'fs';
 import path from 'path';
 import pc from 'picocolors';
 import { extractContent } from './extractor';
+import { generateLlmsTxt, OntoConfig } from './config';
+
+async function loadOntoConfig(): Promise<OntoConfig | null> {
+    try {
+        const configPath = path.join(process.cwd(), 'onto.config');
+        // Node.js dynamic import
+        const config = await import('file://' + configPath.replace(/\\/g, '/') + '.ts');
+        return config.default || config;
+    } catch (error) {
+        try {
+            const configPath = path.join(process.cwd(), 'onto.config');
+            const config = await import('file://' + configPath.replace(/\\/g, '/') + '.js');
+            return config.default || config;
+        } catch (e) {
+            return null;
+        }
+    }
+}
 
 // Simple helper to load .env.local from the current working directory
 function loadEnv() {
@@ -139,6 +157,22 @@ async function main() {
         }
     }
 
+    // --- Generate llms.txt manifest ---
+    const config = await loadOntoConfig();
+    if (config) {
+        const llmsTxtContent = generateLlmsTxt(config);
+        const llmsTxtPath = path.join(cwd, 'public/llms.txt');
+        
+        // Ensure public dir exists
+        const publicDir = path.join(cwd, 'public');
+        if (!fs.existsSync(publicDir)) {
+            fs.mkdirSync(publicDir, { recursive: true });
+        }
+
+        fs.writeFileSync(llmsTxtPath, llmsTxtContent, 'utf8');
+        console.log(pc.green('✓ Generated') + pc.dim(' /llms.txt'));
+    }
+
     console.log(pc.dim(`Edge payloads are ready at /public/.onto/*\n`));
 }
 

package/src/config.ts

@@ -78,22 +78,6 @@ export interface OntoConfig {
   };
 }
 
-/**
- * Load the onto.config.ts file from the user's project
- * This is used by the middleware to dynamically generate llms.txt
- */
-export async function loadOntoConfig(): Promise<OntoConfig | null> {
-  try {
-    // Try to dynamically import the config file from the user's project root
-    // This runs in the middleware context, so we look in the project root
-    const config = await import(process.cwd() + '/onto.config');
-    return config.default || config;
-  } catch (error) {
-    // Config file doesn't exist or failed to load
-    return null;
-  }
-}
-
 /**
  * Generate llms.txt content from OntoConfig
  * Follows the llms.txt specification:

package/src/middleware.ts

@@ -1,18 +1,19 @@
 import { NextRequest, NextResponse } from 'next/server';
-import { AI_BOT_USER_AGENTS, matchBot } from './bots';
-import { loadOntoConfig, generateLlmsTxt } from './config';
+import { matchBot } from './bots';
+import { generateLlmsTxt, OntoConfig } from './config';
 
-export async function ontoMiddleware(request: NextRequest) {
+export async function ontoMiddleware(request: NextRequest, config?: OntoConfig) {
     const userAgent = request.headers.get('user-agent');
+    const url = request.nextUrl.clone();
+    const matched = matchBot(userAgent);
+
     const accept = request.headers.get('accept') || '';
 
-    const matched = matchBot(userAgent);
     const isAiBot = !!matched;
     const isMarkdownRequested = accept.includes('text/markdown');
 
     // If traffic is identified as an AI Bot or markdown is requested
     if (isAiBot || isMarkdownRequested) {
-        const url = request.nextUrl.clone();
 
         // Ignore internal next.js requests & static assets (but not llms.txt)
         if (url.pathname.startsWith('/_next')) {
@@ -20,11 +21,9 @@ export async function ontoMiddleware(request: NextRequest) {
         }
 
         // --- llms.txt Auto-Discovery ---
-        // Dynamically generate llms.txt from onto.config.ts
+        // Serve the llms.txt manifest to AI agents
         if (url.pathname === '/llms.txt') {
             try {
-                const config = await loadOntoConfig();
-
                 if (config) {
                     // Generate llms.txt dynamically from config
                     const llmsTxtContent = generateLlmsTxt(config);

package/dist/OntoHead.d.mts

@@ -1,27 +0,0 @@
-import * as react_jsx_runtime from 'react/jsx-runtime';
-
-/**
- * OntoHead — Auto-Discovery component for AI agents.
- *
- * Injects `<link rel="alternate">` tags into the page `<head>` so AI crawlers
- * can discover the optimized markdown endpoint for the current route.
- *
- * Usage in a Next.js App Router layout:
- * ```tsx
- * import { OntoHead } from '@ontosdk/next/components';
- *
- * export default function RootLayout({ children }) {
- *   return (
- *     <html>
- *       <head>
- *         <OntoHead />
- *       </head>
- *       <body>{children}</body>
- *     </html>
- *   );
- * }
- * ```
- */
-declare function OntoHead(): react_jsx_runtime.JSX.Element;
-
-export { OntoHead };

package/dist/OntoHead.d.ts

@@ -1,27 +0,0 @@
-import * as react_jsx_runtime from 'react/jsx-runtime';
-
-/**
- * OntoHead — Auto-Discovery component for AI agents.
- *
- * Injects `<link rel="alternate">` tags into the page `<head>` so AI crawlers
- * can discover the optimized markdown endpoint for the current route.
- *
- * Usage in a Next.js App Router layout:
- * ```tsx
- * import { OntoHead } from '@ontosdk/next/components';
- *
- * export default function RootLayout({ children }) {
- *   return (
- *     <html>
- *       <head>
- *         <OntoHead />
- *       </head>
- *       <body>{children}</body>
- *     </html>
- *   );
- * }
- * ```
- */
-declare function OntoHead(): react_jsx_runtime.JSX.Element;
-
-export { OntoHead };

package/dist/OntoProvider.d.mts

@@ -1,52 +0,0 @@
-import * as react_jsx_runtime from 'react/jsx-runtime';
-import { ReactNode } from 'react';
-import { OntoConfig } from './config.mjs';
-
-interface OntoProviderProps {
-    /**
-     * The base URL of your site (e.g., 'https://example.com')
-     * Used to construct the full href for the AI discovery link tag.
-     */
-    baseUrl: string;
-    /**
-     * Child components to render
-     */
-    children: ReactNode;
-    /**
-     * Optional: Onto configuration for automatic JSON-LD schema injection
-     * If provided, the provider will automatically inject JSON-LD schemas
-     * based on the page type configuration
-     */
-    config?: OntoConfig;
-}
-/**
- * OntoProvider — Automatic AI Discovery Provider
- *
- * Wraps your application and automatically injects:
- * 1. `<link rel="alternate">` tags for AI discovery
- * 2. JSON-LD structured data schemas based on page type
- *
- * With config, automatically generates JSON-LD schemas:
- * - 'scoring' pages get Methodology schema with AIO weights (40/35/25)
- * - 'about' pages get Organization/AboutPage schema
- *
- * Usage in a Next.js App Router layout:
- * ```tsx
- * import { OntoProvider } from '@ontosdk/next/provider';
- * import config from '../onto.config';
- *
- * export default function RootLayout({ children }) {
- *   return (
- *     <OntoProvider baseUrl="https://example.com" config={config}>
- *       <html>
- *         <head />
- *         <body>{children}</body>
- *       </html>
- *     </OntoProvider>
- *   );
- * }
- * ```
- */
-declare function OntoProvider({ baseUrl, children, config }: OntoProviderProps): react_jsx_runtime.JSX.Element;
-
-export { OntoProvider, type OntoProviderProps };

package/dist/OntoProvider.d.ts

@@ -1,52 +0,0 @@
-import * as react_jsx_runtime from 'react/jsx-runtime';
-import { ReactNode } from 'react';
-import { OntoConfig } from './config.js';
-
-interface OntoProviderProps {
-    /**
-     * The base URL of your site (e.g., 'https://example.com')
-     * Used to construct the full href for the AI discovery link tag.
-     */
-    baseUrl: string;
-    /**
-     * Child components to render
-     */
-    children: ReactNode;
-    /**
-     * Optional: Onto configuration for automatic JSON-LD schema injection
-     * If provided, the provider will automatically inject JSON-LD schemas
-     * based on the page type configuration
-     */
-    config?: OntoConfig;
-}
-/**
- * OntoProvider — Automatic AI Discovery Provider
- *
- * Wraps your application and automatically injects:
- * 1. `<link rel="alternate">` tags for AI discovery
- * 2. JSON-LD structured data schemas based on page type
- *
- * With config, automatically generates JSON-LD schemas:
- * - 'scoring' pages get Methodology schema with AIO weights (40/35/25)
- * - 'about' pages get Organization/AboutPage schema
- *
- * Usage in a Next.js App Router layout:
- * ```tsx
- * import { OntoProvider } from '@ontosdk/next/provider';
- * import config from '../onto.config';
- *
- * export default function RootLayout({ children }) {
- *   return (
- *     <OntoProvider baseUrl="https://example.com" config={config}>
- *       <html>
- *         <head />
- *         <body>{children}</body>
- *       </html>
- *     </OntoProvider>
- *   );
- * }
- * ```
- */
-declare function OntoProvider({ baseUrl, children, config }: OntoProviderProps): react_jsx_runtime.JSX.Element;
-
-export { OntoProvider, type OntoProviderProps };

package/dist/cli.d.mts

@@ -1 +0,0 @@
-#!/usr/bin/env node

package/dist/cli.d.ts

@@ -1 +0,0 @@
-#!/usr/bin/env node

package/dist/config.d.mts

@@ -1,85 +0,0 @@
-/**
- * Configuration schema for onto.config.ts
- * Used to dynamically generate llms.txt and other AI discovery files
- */
-type PageType = 'scoring' | 'about' | 'default';
-interface OntoRoute {
-    /**
-     * The URL path (e.g., '/docs', '/api/reference')
-     */
-    path: string;
-    /**
-     * Description of what this route contains
-     */
-    description: string;
-    /**
-     * Optional: Page type for automatic JSON-LD schema injection
-     * - 'scoring': Injects Methodology schema with AIO scoring weights (40/35/25)
-     * - 'about': Injects Organization/AboutPage schema
-     * - 'default': No automatic schema injection
-     */
-    pageType?: PageType;
-}
-interface OntoConfig {
-    /**
-     * The name of your project or site (required)
-     * Used as the H1 heading in llms.txt
-     */
-    name: string;
-    /**
-     * A short summary of your project (required)
-     * Displayed as a blockquote in llms.txt
-     * Should contain key information necessary for understanding the rest of the file
-     */
-    summary: string;
-    /**
-     * The base URL of your site (e.g., 'https://example.com')
-     */
-    baseUrl: string;
-    /**
-     * Optional: Additional sections to include in llms.txt
-     * Each section can contain any markdown content
-     */
-    sections?: {
-        heading: string;
-        content: string;
-    }[];
-    /**
-     * Key routes that AI agents should know about
-     * These will be formatted as a markdown list in llms.txt
-     */
-    routes?: OntoRoute[];
-    /**
-     * Optional: Links to external resources (documentation, API references, etc.)
-     */
-    externalLinks?: {
-        title: string;
-        url: string;
-        description?: string;
-    }[];
-    /**
-     * Optional: Organization information for JSON-LD schemas
-     */
-    organization?: {
-        name: string;
-        description?: string;
-        url?: string;
-        logo?: string;
-        foundingDate?: string;
-    };
-}
-/**
- * Load the onto.config.ts file from the user's project
- * This is used by the middleware to dynamically generate llms.txt
- */
-declare function loadOntoConfig(): Promise<OntoConfig | null>;
-/**
- * Generate llms.txt content from OntoConfig
- * Follows the llms.txt specification:
- * - H1 with project name
- * - Blockquote with summary
- * - Additional markdown sections
- */
-declare function generateLlmsTxt(config: OntoConfig): string;
-
-export { type OntoConfig, type OntoRoute, type PageType, generateLlmsTxt, loadOntoConfig };

package/dist/config.d.ts

@@ -1,85 +0,0 @@
-/**
- * Configuration schema for onto.config.ts
- * Used to dynamically generate llms.txt and other AI discovery files
- */
-type PageType = 'scoring' | 'about' | 'default';
-interface OntoRoute {
-    /**
-     * The URL path (e.g., '/docs', '/api/reference')
-     */
-    path: string;
-    /**
-     * Description of what this route contains
-     */
-    description: string;
-    /**
-     * Optional: Page type for automatic JSON-LD schema injection
-     * - 'scoring': Injects Methodology schema with AIO scoring weights (40/35/25)
-     * - 'about': Injects Organization/AboutPage schema
-     * - 'default': No automatic schema injection
-     */
-    pageType?: PageType;
-}
-interface OntoConfig {
-    /**
-     * The name of your project or site (required)
-     * Used as the H1 heading in llms.txt
-     */
-    name: string;
-    /**
-     * A short summary of your project (required)
-     * Displayed as a blockquote in llms.txt
-     * Should contain key information necessary for understanding the rest of the file
-     */
-    summary: string;
-    /**
-     * The base URL of your site (e.g., 'https://example.com')
-     */
-    baseUrl: string;
-    /**
-     * Optional: Additional sections to include in llms.txt
-     * Each section can contain any markdown content
-     */
-    sections?: {
-        heading: string;
-        content: string;
-    }[];
-    /**
-     * Key routes that AI agents should know about
-     * These will be formatted as a markdown list in llms.txt
-     */
-    routes?: OntoRoute[];
-    /**
-     * Optional: Links to external resources (documentation, API references, etc.)
-     */
-    externalLinks?: {
-        title: string;
-        url: string;
-        description?: string;
-    }[];
-    /**
-     * Optional: Organization information for JSON-LD schemas
-     */
-    organization?: {
-        name: string;
-        description?: string;
-        url?: string;
-        logo?: string;
-        foundingDate?: string;
-    };
-}
-/**
- * Load the onto.config.ts file from the user's project
- * This is used by the middleware to dynamically generate llms.txt
- */
-declare function loadOntoConfig(): Promise<OntoConfig | null>;
-/**
- * Generate llms.txt content from OntoConfig
- * Follows the llms.txt specification:
- * - H1 with project name
- * - Blockquote with summary
- * - Additional markdown sections
- */
-declare function generateLlmsTxt(config: OntoConfig): string;
-
-export { type OntoConfig, type OntoRoute, type PageType, generateLlmsTxt, loadOntoConfig };

package/dist/index.d.mts

@@ -1,25 +0,0 @@
-export { OntoConfig, OntoConfig as OntoConfigType, OntoRoute, OntoRoute as OntoRouteType, PageType, generateLlmsTxt, loadOntoConfig } from './config.mjs';
-export { AIOMethodologySchema, AboutPageSchema, OrganizationSchema, generateAIOMethodologySchema, generateAboutPageSchema, generateOrganizationSchema, generateSchemaForPageType, serializeSchema } from './schemas.mjs';
-
-interface ExtractionResult {
-    markdown: string;
-    metadata: {
-        title: string;
-        description: string;
-        jsonLd: any[];
-    };
-    stats: {
-        originalHtmlSize: number;
-        markdownSize: number;
-        tokenReductionRatio: number;
-    };
-}
-/**
- * Extracts pure semantic markdown and metadata from rendered Next.js HTML strings.
- * @param html The raw HTML string.
- * @param sourceUrl (Optional) the URL this was generated from, to attach as metadata.
- * @returns {ExtractionResult} The extracted payload.
- */
-declare function extractContent(html: string, sourceUrl?: string): ExtractionResult;
-
-export { extractContent };

package/dist/index.d.ts

@@ -1,25 +0,0 @@
-export { OntoConfig, OntoConfig as OntoConfigType, OntoRoute, OntoRoute as OntoRouteType, PageType, generateLlmsTxt, loadOntoConfig } from './config.js';
-export { AIOMethodologySchema, AboutPageSchema, OrganizationSchema, generateAIOMethodologySchema, generateAboutPageSchema, generateOrganizationSchema, generateSchemaForPageType, serializeSchema } from './schemas.js';
-
-interface ExtractionResult {
-    markdown: string;
-    metadata: {
-        title: string;
-        description: string;
-        jsonLd: any[];
-    };
-    stats: {
-        originalHtmlSize: number;
-        markdownSize: number;
-        tokenReductionRatio: number;
-    };
-}
-/**
- * Extracts pure semantic markdown and metadata from rendered Next.js HTML strings.
- * @param html The raw HTML string.
- * @param sourceUrl (Optional) the URL this was generated from, to attach as metadata.
- * @returns {ExtractionResult} The extracted payload.
- */
-declare function extractContent(html: string, sourceUrl?: string): ExtractionResult;
-
-export { extractContent };

package/dist/middleware.d.mts

@@ -1,25 +0,0 @@
-import { NextRequest, NextResponse } from 'next/server';
-
-/**
- * Comprehensive registry of AI bot user-agent strings.
- * The middleware uses this list to detect AI crawlers and serve optimized markdown.
- */
-interface AiBot {
-    /** The user-agent substring to match against */
-    name: string;
-    /** The company operating this bot */
-    company: string;
-}
-/**
- * Flat list of user-agent substrings for fast matching in the middleware.
- */
-declare const AI_BOT_USER_AGENTS: string[];
-/**
- * Given a raw user-agent string, returns the matched AiBot entry or undefined.
- * Comparison is case-insensitive to handle inconsistent agent casing.
- */
-declare function matchBot(userAgent: string | null): AiBot | undefined;
-
-declare function ontoMiddleware(request: NextRequest): Promise<NextResponse<unknown>>;
-
-export { AI_BOT_USER_AGENTS, type AiBot, matchBot, ontoMiddleware };

package/dist/middleware.d.ts

@@ -1,25 +0,0 @@
-import { NextRequest, NextResponse } from 'next/server';
-
-/**
- * Comprehensive registry of AI bot user-agent strings.
- * The middleware uses this list to detect AI crawlers and serve optimized markdown.
- */
-interface AiBot {
-    /** The user-agent substring to match against */
-    name: string;
-    /** The company operating this bot */
-    company: string;
-}
-/**
- * Flat list of user-agent substrings for fast matching in the middleware.
- */
-declare const AI_BOT_USER_AGENTS: string[];
-/**
- * Given a raw user-agent string, returns the matched AiBot entry or undefined.
- * Comparison is case-insensitive to handle inconsistent agent casing.
- */
-declare function matchBot(userAgent: string | null): AiBot | undefined;
-
-declare function ontoMiddleware(request: NextRequest): Promise<NextResponse<unknown>>;
-
-export { AI_BOT_USER_AGENTS, type AiBot, matchBot, ontoMiddleware };

package/dist/schemas.d.mts

@@ -1,73 +0,0 @@
-import { OntoConfig } from './config.mjs';
-
-/**
- * JSON-LD Schema generators for automatic structured data injection
- * Follows Schema.org standards for AI-friendly metadata
- */
-
-/**
- * Standard AIO (AI Optimization) scoring methodology
- * Based on the Onto scoring algorithm:
- * - Content Negotiation: 40% (30 points)
- * - React Tax / Token Efficiency: 35% (30 points)
- * - Structured Data: 25% (25 points)
- * - Semantic HTML: Bonus (15 points)
- */
-interface AIOMethodologySchema {
-    '@context': 'https://schema.org';
-    '@type': 'HowTo';
-    name: string;
-    description: string;
-    step: Array<{
-        '@type': 'HowToStep';
-        name: string;
-        text: string;
-        position: number;
-    }>;
-}
-/**
- * Generate AIO Scoring Methodology JSON-LD schema
- * This explains to AI agents how the scoring system works
- */
-declare function generateAIOMethodologySchema(config: OntoConfig, pageUrl: string): AIOMethodologySchema;
-/**
- * Organization schema for About pages
- */
-interface OrganizationSchema {
-    '@context': 'https://schema.org';
-    '@type': 'Organization';
-    name: string;
-    url?: string;
-    description?: string;
-    logo?: string;
-    foundingDate?: string;
-}
-/**
- * Generate Organization JSON-LD schema for About pages
- */
-declare function generateOrganizationSchema(config: OntoConfig, pageUrl: string): OrganizationSchema | null;
-/**
- * AboutPage schema combining Organization and WebPage
- */
-interface AboutPageSchema {
-    '@context': 'https://schema.org';
-    '@type': 'AboutPage';
-    name: string;
-    url: string;
-    description?: string;
-    mainEntity?: OrganizationSchema;
-}
-/**
- * Generate AboutPage JSON-LD schema
- */
-declare function generateAboutPageSchema(config: OntoConfig, pageUrl: string): AboutPageSchema;
-/**
- * Determine which schema to generate based on page type
- */
-declare function generateSchemaForPageType(pageType: 'scoring' | 'about' | 'default', config: OntoConfig, pageUrl: string): any | null;
-/**
- * Serialize schema to JSON-LD script tag content
- */
-declare function serializeSchema(schema: any | null): string | null;
-
-export { type AIOMethodologySchema, type AboutPageSchema, type OrganizationSchema, generateAIOMethodologySchema, generateAboutPageSchema, generateOrganizationSchema, generateSchemaForPageType, serializeSchema };