prestruct 1.0.0 → 1.1.0

package/init/package.json

@@ -1,16 +1,21 @@
 {
   "name": "prestruct",
-  "version": "1.0.0",
+  "version": "1.1.0",
   "description": "Build-time SEO prerender layer for Vite + React + Cloudflare Pages",
   "type": "module",
   "scripts": {
     "build": "vite build && node scripts/inject-brand.js && node scripts/prerender.js",
     "test": "vitest",
     "test:run": "vitest run",
-    "test:ui": "vitest --ui",
     "lint": "eslint . --ext js,mjs",
     "lint:fix": "eslint . --ext js,mjs --fix"
   },
+  "overrides": {
+    "tinypool": "^2.2.0"
+  },
+  "engines": {
+    "node": ">=20"
+  },
   "peerDependencies": {
     "react": ">=18",
     "react-dom": ">=18",
@@ -20,8 +25,8 @@
   "devDependencies": {
     "@vitejs/plugin-react": "^4.3.1",
     "eslint": "^9.0.0",
-    "vite": "^5.4.0",
-    "vitest": "^2.0.0",
+    "vite": "^6.4.5",
+    "vitest": "^5.0.5",
     "react": "^18.3.1",
     "react-dom": "^18.3.1",
     "react-router-dom": "^6.30.1"

package/init/scripts/prerender.js

@@ -272,6 +272,30 @@ export function cleanCache({ cacheFile = enginePaths().cacheFile } = {}) {
 
 // -- Sitemap -------------------------------------------------------------------
 
+//
+// AI-crawler policy preset (aeo option):
+// Search-visibility bots get explicit Allow groups; TRAINING-crawl preference
+// is a separate choice via config.robots.aiBots ('allow' | 'block', default
+// 'block'). Bot names are the token each vendor documents for its own robots
+// product (2026: OpenAI OAI-SearchBot/GPTBot, Anthropic
+// Claude-SearchBot/ClaudeBot, PerplexityBot, Apple Applebot/Applebot-Extended,
+// Common Crawl CCBot, Google-Extended)
+const AEO_SEARCH_BOTS = [
+  'OAI-SearchBot',      // OpenAI: surfaces sites in ChatGPT search
+  'PerplexityBot',      // Perplexity: search index, not model training
+  'Claude-SearchBot',   // Anthropic: Claude's search
+  'Applebot',           // Apple: Siri/Spotlight/Safari search
+  'Googlebot',          // Google Search incl. AI Overviews eligibility
+  'Bingbot',            // Bing / Copilot discovery
+]
+const AEO_TRAINING_BOTS = [
+  'GPTBot',             // OpenAI: model training crawl
+  'ClaudeBot',          // Anthropic: model development crawl
+  'Google-Extended',    // Gemini/Vertex grounding (product token, no HTTP hits)
+  'Applebot-Extended',  // Apple: foundation model training
+  'CCBot',              // Common Crawl: open corpus crawlers
+]
+
 /**
  * Generate robots.txt when config.robots is opted in; otherwise null so the
  * author's own public/robots.txt is left untouched. The Sitemap: line points
@@ -290,16 +314,25 @@ export function generateRobots(config) {
   const cfgBody    = typeof cfg === 'string' ? cfg : (cfg.body || '')
   const cfgSitemap = cfgBool || (cfg && typeof cfg === 'object' ? cfg.sitemap !== false : true)
   const siteUrl    = (config.siteUrl || '').replace(/\/$/, '')
-
-  if (cfgBool) {
-    const lines = [
-      'User-agent: *',
-      'Allow: /',
-    ]
-    if (cfgSitemap && siteUrl) {
-      lines.push('', `Sitemap: ${siteUrl}/sitemap.xml`)
+  const sitemapLine = cfgSitemap && siteUrl ? ['', `Sitemap: ${siteUrl}/sitemap.xml`] : []
+
+  // { aeo: true | {} } -> search bots allowed, training crawls blocked by
+  // default. { aeo: { aiBots: 'allow' } } flips training to Allow. aeo with
+  // an explicit body: the body is appended after the preset groups (letting
+  // owners add extra paths) only via body; the preset itself ignores body.
+  const aeo       = !cfgBool && cfg && cfg.aeo
+  const aeoCli    = aeo ? (aeo.aiBots === 'allow' ? 'allow' : 'block') : null
+
+  if (cfgBool || aeo) {
+    const lines = []
+    if (cfgBool) {
+      lines.push('User-agent: *', 'Allow: /')
+    } else {
+      for (const bot of AEO_SEARCH_BOTS) lines.push(`User-agent: ${bot}`, 'Allow: /', '')
+      const verdict = aeoCli === 'allow' ? 'Allow: /' : 'Disallow: /'
+      for (const bot of AEO_TRAINING_BOTS) lines.push(`User-agent: ${bot}`, verdict, '')
     }
-    return lines.join('\n') + '\n'
+    return lines.concat(sitemapLine).join('\n').replace(/\n+$/, '\n')
   }
 
   // Custom body; sitemap pointer appended unless { sitemap: false }
@@ -649,19 +682,21 @@ export async function prerender(cfg = config, { force = false, root } = {}) {
       console.log('[prerender] ok /robots.txt')
     }
 
-    // Cache stats -- static-sites-safe monitor: a JSON snapshot beside dist/
-    // (Intentionally not a runtime endpoint; static hosts serve files, not
-    // processes. Follows the PRD "static-sites-safe option preferred" note.)
+    // Cache stats -- static-sites-safe monitor: a JSON snapshot in the
+    // project's .prestruct/ dir, NOT dist/. Build telemetry must not become a
+    // deployed URL (issue 23): dist/ is the public artifact on every static
+    // host, and the old wall-clock generatedAt made two byte-identical
+    // builds differ. Same monitor, invisible to deploys, no opt-out needed
+    // because there is nothing to opt out of shipping.
     const stats = {
-      generatedAt: new Date().toISOString(),
       routeCount: ROUTES.length,
       succeeded,
       cached: Object.keys(cache).length,
       force,
       routes: ROUTES.map(r => ({ path: r.path, cached: Boolean(cache[r.path]) })),
     }
-    fs.mkdirSync(path.join(p.dist, '.prestruct'), { recursive: true })
-    fs.writeFileSync(path.join(p.dist, '.prestruct', 'status.json'), JSON.stringify(stats, null, 2), 'utf-8')
+    fs.mkdirSync(path.join(p.cacheDir, '..'), { recursive: true })
+    fs.writeFileSync(path.join(p.root, '.prestruct', 'status.json'), JSON.stringify(stats, null, 2), 'utf-8')
 
     console.log(`[prerender] Done. ${succeeded}/${ROUTES.length} pages rendered.\n`)
 

package/init/src/App.jsx

@@ -9,7 +9,13 @@
  * homepage with no error message. See AGENTS.md for the full explanation.
  */
 
-import { BrowserRouter } from 'react-router-dom'
+import * as rrdNS from 'react-router-dom'
+// Neither plain named imports nor a bare namespace destructure survive
+// both majors (see AppLayout.jsx). Destructure from whichever side of
+// the interop carries the exports object.
+const rrd = rrdNS.default ?? rrdNS
+const { BrowserRouter } = rrd
+
 import AppLayout from './AppLayout.jsx'
 
 export default function App() {

package/init/src/AppLayout.jsx

@@ -20,7 +20,14 @@
  * runs in Node where these globals don't exist.
  */
 
-import { Routes, Route, useLocation } from 'react-router-dom'
+import * as rrdNS from 'react-router-dom'
+// Neither plain named imports nor a bare namespace destructure survive
+// both majors: v7 ships CJS (named imports rejected under vite SSR; the
+// node interop namespace only reliably carries `default`), v6 is ESM
+// with possibly no default. Destructure from whichever has the exports.
+const rrd = rrdNS.default ?? rrdNS
+const { Routes, Route, useLocation } = rrd
+
 import { useEffect } from 'react'
 import { mountIslands } from './islands.js'
 import { islands }      from './AppIslands.jsx'

package/init/src/hooks/usePageMeta.js

@@ -43,7 +43,14 @@
  */
 
 import { useEffect } from 'react'
-import { useLocation } from 'react-router-dom'
+import * as rrdNS from 'react-router-dom'
+// Neither plain named imports nor a bare namespace destructure survive
+// both majors: v7 ships CJS (vite SSR rejects named imports; node interop
+// gives a namespace whose only reliable export is `default` -- the exports
+// object), while v6 is ESM with possibly no default. Destructure from
+// whichever carries the exports.
+const rrd = rrdNS.default ?? rrdNS
+const { useLocation } = rrd
 
 /** Update the SPA head tags on every route change (see the module head doc). */
 export default function usePageMeta({ title, description, path, siteUrl }) {

package/init/ssr.config.example.js

@@ -46,6 +46,14 @@ export default {
   // robots: true,              // minimal block (all agents + Sitemap line)
   // robots: "User-agent: *\nDisallow: /private",  // custom body; sitemap auto-appended
   // robots: { body: "...", sitemap: false },      // custom body, no sitemap line
+  //
+  // AEO preset -- per-bot crawl policy from the 2026 vendor docs:
+  // robots: { aeo: true },                  // search AI bots Allow, training crawls Disallow
+  // robots: { aeo: { aiBots: 'allow' } },   // same, but also allow training crawls
+  // Search group: OAI-SearchBot, PerplexityBot, Claude-SearchBot, Applebot,
+  // Googlebot, Bingbot. Training group: GPTBot, ClaudeBot, Google-Extended,
+  // Applebot-Extended, CCBot. Search access and training/data-use are separate
+  // choices; the Sitemap line is appended as usual.
 
   // === Routes ===
   // Each route gets prerendered to static HTML at build time.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "prestruct",
-  "version": "1.0.0",
+  "version": "1.1.0",
   "description": "AI-first SEO prerender for Vite + React apps. Drop in, config, build.",
   "type": "module",
   "license": "SEE LICENSE IN LICENSE",
@@ -14,6 +14,9 @@
     "./prerender": "./engine/scripts/prerender.js",
     "./inject-brand": "./engine/scripts/inject-brand.js",
     "./validate-config": "./engine/scripts/validate-config.js",
+    "./hooks": "./engine/src/hooks/usePageMeta.js",
+    "./islands": "./engine/src/islands.js",
+    "./engine/*": "./engine/*",
     "./package.json": "./package.json"
   },
   "files": [
@@ -34,7 +37,10 @@
     "e2e": "node tests/e2e-fixture.mjs"
   },
   "engines": {
-    "node": ">=18"
+    "node": ">=20"
+  },
+  "overrides": {
+    "tinypool": "^2.2.0"
   },
   "peerDependencies": {
     "react": ">=18",
@@ -45,12 +51,11 @@
   "devDependencies": {
     "express": "^4.22.3",
     "happy-dom": "^21.0.4",
-    "puppeteer": "^22.15.0",
     "react": "^18.3.1",
     "react-dom": "^18.3.1",
     "react-router-dom": "^6.30.1",
-    "vite": "^5.4.0",
-    "vitest": "^2.1.0"
+    "vite": "^6.4.5",
+    "vitest": "^5.0.5"
   },
   "repository": {
     "type": "git",