diff --git a/README.md b/README.md
index c8cbf9a1..aa359ee1 100644
--- a/README.md
+++ b/README.md
@@ -40,9 +40,12 @@ need these features.
 Install `@nuxtjs/robots` dependency to your project:
 
 ```bash
-npx nuxi@latest module add robots
+npx nuxt module add robots
 ```
 
+> [!TIP]
+> Using an AI agent? Get the @nuxtjs/robots Skill on [skilld.dev/gh/nuxt-modules/robots](https://skilld.dev/gh/nuxt-modules/robots).
+
 💡 Configured your robots.txt? Verify what crawlers and AI bots can access with the free [Robots.txt Validator](https://nuxtseo.com/tools/robots-txt-validator), or monitor crawl & indexing status with [Nuxt SEO Pro](https://nuxtseo.com/pro).
 
 ## Documentation
diff --git a/docs/content/1.getting-started/1.installation.md b/docs/content/1.getting-started/1.installation.md
index 56b990de..10778d16 100644
--- a/docs/content/1.getting-started/1.installation.md
+++ b/docs/content/1.getting-started/1.installation.md
@@ -20,6 +20,16 @@ To get started with Nuxt Robots, you need to install the dependency and add it t
 
 :ModuleInstall{name="robots" included-in="@nuxtjs/seo"}
 
+Use Nuxt 4.6 or Nuxt 5.
+Use Node 22.22.3+, Node 24.15.0+, or Node 26+.
+
+Import app functions from `#robots/app`. Import server functions from `#robots/server`.
+If you use Nuxt Content, use version 3.
+
+::tip
+Using an AI agent? Get the @nuxtjs/robots Skill on [skilld.dev/gh/nuxt-modules/robots](https://skilld.dev/gh/nuxt-modules/robots).
+::
+
 ## Verifying Installation
 
 To ensure the module is behaving as expected, you should first check [`/robots.txt`](http://localhost:3000/robots.txt) is being generated.
@@ -31,6 +41,10 @@ However, we want to see what a production environment would look like.
 
 For this, it's recommended to use the Nuxt DevTools Robots tab to see the current configuration and how it's being applied.
 
+::note
+The DevTools panel needs the `nuxtseo-devtools-host` package. Install it as a dev dependency with `pnpm add -D nuxtseo-devtools-host`, then restart the dev server. You do not add it to `modules`.
+::
+
 The DevTools will show you that in production we're just serving a minimal robots.txt file.
 
 ```robots-txt [robots.txt]
diff --git a/docs/content/1.getting-started/3.troubleshooting.md b/docs/content/1.getting-started/3.troubleshooting.md
index 80f8bb50..99473f25 100644
--- a/docs/content/1.getting-started/3.troubleshooting.md
+++ b/docs/content/1.getting-started/3.troubleshooting.md
@@ -15,6 +15,7 @@ relatedPages:
 ## Debugging
 
 **Nuxt DevTools.** The best tool for debugging is the Nuxt DevTools integration with Nuxt Robots. This will show you the current robot rules and your robots.txt file.
+The panel needs the `nuxtseo-devtools-host` package as a dev dependency. Install it with `pnpm add -D nuxtseo-devtools-host`, then restart the dev server.
 
 **Debug config.** You can enable the [debug](/docs/robots/api/config#debug) option which will give you more granular output. This is enabled by default in development mode.
 
diff --git a/docs/content/2.guides/1.robots-txt.md b/docs/content/2.guides/1.robots-txt.md
index b1bddce0..1bf30d13 100644
--- a/docs/content/2.guides/1.robots-txt.md
+++ b/docs/content/2.guides/1.robots-txt.md
@@ -31,21 +31,23 @@ Additionally, the following paths are supported by default:
 ```bash [Example File Structure]
 # root directory
 robots.txt
-# asset folders
-assets/
-├── robots.txt
-# pages folder
-pages/
-├── robots.txt
-├── _dir/
-│   └── robots.txt
 # public folder
 public/
 ├── _robots.txt
 ├── _dir/
 │   └── robots.txt
+# source directory (app/ on Nuxt 4)
+app/
+├── assets/
+│   └── robots.txt
+└── pages/
+    ├── robots.txt
+    └── _dir/
+        └── robots.txt
 ```
 
+The module reads the `assets/` and `pages/` paths from your `srcDir`. On Nuxt 4, the default `srcDir` is `app/`.
+
 **Custom paths.** If you find this too restrictive,
 you can use the `mergeWithRobotsTxtPath` config to load your `robots.txt` file from any path.
 
diff --git a/docs/content/2.guides/2.how-it-works.md b/docs/content/2.guides/2.how-it-works.md
index 4dd4e9ab..8be4b7e9 100644
--- a/docs/content/2.guides/2.how-it-works.md
+++ b/docs/content/2.guides/2.how-it-works.md
@@ -27,7 +27,7 @@ public
 ```
 
 This file is generated differently depending on the environment:
-- When deploying using `nuxi generate` or the `nitro.prerender.routes` rule, this is a static file.
+- When deploying using `nuxt generate` or the `nitro.prerender.routes` rule, this is a static file.
 - Otherwise, it's handled by the server and generated at runtime when requested.
 
 When indexing is disabled a `robots.txt` will be generated with the following content:
diff --git a/docs/content/2.guides/5.disable-indexing.md b/docs/content/2.guides/5.disable-indexing.md
index bf8d9c15..e8695924 100644
--- a/docs/content/2.guides/5.disable-indexing.md
+++ b/docs/content/2.guides/5.disable-indexing.md
@@ -96,10 +96,12 @@ to whatever makes sense for your project.
 If you deploy the same app to multiple domains (e.g. `mydomain.com` and `test.temp-domain.com`), you can disable indexing for specific hosts at runtime using a Nitro plugin.
 
 ```ts [server/plugins/robots-domain.ts]
+import { getNitroOrigin } from '#site-config/server'
+
 export default defineNitroPlugin((nitroApp) => {
   nitroApp.hooks.hook('site-config:init', ({ event, siteConfig }) => {
-    const host = event.headers.get('host')
-    if (host?.includes('staging') || host?.includes('test')) {
+    const host = new URL(getNitroOrigin(event)).hostname
+    if (host.includes('staging') || host.includes('test')) {
       siteConfig.push({ indexable: false, _context: 'host-indexing' })
     }
   })
diff --git a/docs/content/2.guides/8.bot-detection.md b/docs/content/2.guides/8.bot-detection.md
index 5f063425..763421e1 100644
--- a/docs/content/2.guides/8.bot-detection.md
+++ b/docs/content/2.guides/8.bot-detection.md
@@ -95,7 +95,7 @@ Since server-side detection only uses HTTP headers, detection can only work for
 You can detect bots inside a Nitro route, middleware, or API handler.
 
 ```ts
-import { getBotDetection } from '#robots/server/composables/getBotDetection'
+import { getBotDetection } from '#robots/server'
 
 export default defineEventHandler((e) => {
   const detection = getBotDetection(e)
@@ -116,7 +116,7 @@ When using bot detection in Nuxt, it will use the `User-Agent` header by default
 
 ```vue
 <script setup lang="ts">
-import { useBotDetection } from '#robots/app/composables/useBotDetection'
+import { useBotDetection } from '#robots/app'
 
 const { isBot, botName, botCategory, trusted } = useBotDetection({
   fingerprint: true, // detects using botd
@@ -152,7 +152,7 @@ localStorage.getItem('__nuxt_robots:botd') // returns the cached bot detection r
 The properties returned from the composable are all `ref`s. It's important to watch these for changes if you're using fingerprinting, as the results will not be immediately available when the composable is called.
 
 ```ts
-import { useBotDetection } from '#robots/app/composables/useBotDetection'
+import { useBotDetection } from '#robots/app'
 import { watch } from 'vue'
 
 const { isBot } = useBotDetection({
@@ -169,7 +169,7 @@ watch(isBot, (detected) => {
 Alternatively you can use the `onFingerprintResult` callback to handle the result when fingerprinting completes.
 
 ```ts
-import { useBotDetection } from '#robots/app/composables/useBotDetection'
+import { useBotDetection } from '#robots/app'
 
 const botd = useBotDetection({
   fingerprint: true,
diff --git a/docs/content/4.api/3.use-bot-detection.md b/docs/content/4.api/3.use-bot-detection.md
index a23eb1cc..39f9d2dd 100644
--- a/docs/content/4.api/3.use-bot-detection.md
+++ b/docs/content/4.api/3.use-bot-detection.md
@@ -27,7 +27,7 @@ The composable provides reactive access to bot detection results with automatic
 **Basic detection:**
 
 ```ts
-import { useBotDetection } from '#robots/app/composables/useBotDetection'
+import { useBotDetection } from '#robots/app'
 
 const { isBot, botName, botCategory, trusted } = useBotDetection()
 // isBot: ComputedRef<boolean>
@@ -39,7 +39,7 @@ const { isBot, botName, botCategory, trusted } = useBotDetection()
 **With fingerprinting:**
 
 ```ts
-import { useBotDetection } from '#robots/app/composables/useBotDetection'
+import { useBotDetection } from '#robots/app'
 
 const { isBot, botName, botCategory, trusted, reset } = useBotDetection({
   fingerprint: true,
@@ -55,7 +55,7 @@ const { isBot, botName, botCategory, trusted, reset } = useBotDetection({
 **Watching for changes:**
 
 ```ts
-import { useBotDetection } from '#robots/app/composables/useBotDetection'
+import { useBotDetection } from '#robots/app'
 import { watch } from 'vue'
 
 const { isBot, botName, botCategory } = useBotDetection()
@@ -112,7 +112,7 @@ interface UseBotDetectionReturn {
 On the server, bot detection runs when you use the composable:
 
 ```ts
-import { useBotDetection } from '#robots/app/composables/useBotDetection'
+import { useBotDetection } from '#robots/app'
 
 // Only runs when composable is used
 const { isBot, botName, botCategory } = useBotDetection()
@@ -130,7 +130,7 @@ If the server-side detection is unavailable (e.g. SPA mode), it will fallback to
 Client-side fingerprinting is automatic when enabled:
 
 ```ts
-import { useBotDetection } from '#robots/app/composables/useBotDetection'
+import { useBotDetection } from '#robots/app'
 
 const { isBot, botName, botCategory } = useBotDetection({
   fingerprint: true,
diff --git a/docs/content/5.nitro-api/2.get-bot-detection.md b/docs/content/5.nitro-api/2.get-bot-detection.md
index a1401646..443b4c90 100644
--- a/docs/content/5.nitro-api/2.get-bot-detection.md
+++ b/docs/content/5.nitro-api/2.get-bot-detection.md
@@ -70,10 +70,10 @@ export default defineEventHandler(async (event) => {
   const detection = getBotDetection(event)
 
   // Block untrusted bots from API routes
-  if (event.node.req.url?.startsWith('/api/') && detection.isBot && !detection.trusted) {
+  if (event.url.pathname.startsWith('/api/') && detection.isBot && !detection.trusted) {
     throw createError({
-      statusCode: 403,
-      statusMessage: 'Bot access denied'
+      status: 403,
+      statusText: 'Bot access denied'
     })
   }
 })
diff --git a/docs/content/5.nitro-api/3.nitro-hooks.md b/docs/content/5.nitro-api/3.nitro-hooks.md
index 3b931612..3dbf9896 100644
--- a/docs/content/5.nitro-api/3.nitro-hooks.md
+++ b/docs/content/5.nitro-api/3.nitro-hooks.md
@@ -49,7 +49,7 @@ For host-based page indexing, use [Site Config hooks](/docs/robots/guides/disabl
 ```ts [server/plugins/robots-domain.ts]
 export default defineNitroPlugin((nitroApp) => {
   nitroApp.hooks.hook('robots:config', (ctx) => {
-    const host = ctx.event?.headers.get('host')
+    const host = ctx.event?.req.headers.get('host')
     if (host?.includes('staging') || host?.includes('test')) {
       ctx.groups[0].disallow = ['/']
     }
@@ -72,7 +72,7 @@ This hook allows you to modify the robots.txt content before it is sent to the c
 
 ```ts [server/plugins/robots-remove-comments.ts]
 export default defineNitroPlugin((nitroApp) => {
-  if (!process.dev) {
+  if (!import.meta.dev) {
     nitroApp.hooks.hook('robots:robots-txt', async (ctx) => {
       // remove comments from robotsTxt in production
       ctx.robotsTxt = ctx.robotsTxt.replace(/^#.*$/gm, '').trim()
diff --git a/docs/content/6.releases/2.v7.md b/docs/content/6.releases/2.v7.md
new file mode 100644
index 00000000..7f1a51f9
--- /dev/null
+++ b/docs/content/6.releases/2.v7.md
@@ -0,0 +1,196 @@
+---
+title: v7.0.0
+description: Release notes for Nuxt Robots v7.0.0.
+ogImage:
+  component: Release
+  props:
+    colorMode: dark
+    title: "Release: ✨ Nuxt Robots v7"
+    description: Nuxt 4.6 and Nuxt 5 support, AI training opt out, and exact page metadata matching.
+relatedPages:
+  - path: /docs/nuxt-seo/releases/v6
+    title: Nuxt SEO v6
+  - path: /docs/nuxt-seo/migration-guide/v5-to-v6
+    title: Nuxt SEO v5 to v6
+  - path: /docs/site-config/releases/v5
+    title: Nuxt Site Config v5.0.0
+  - path: /docs/robots/releases/v6
+    title: v6.0.0
+---
+
+## Introduction
+
+Nuxt Robots v7 moves the module to the portable server APIs of Nuxt 4.6 and Nuxt 5.
+It ships in [Nuxt SEO v6](/docs/nuxt-seo/releases/v6).
+If you use `@nuxtjs/seo`, follow the [Nuxt SEO v5 to v6 migration guide](/docs/nuxt-seo/migration-guide/v5-to-v6).
+
+It also adds an AI training opt out and matches page metadata with your app router.
+
+Check the [breaking changes](#breaking-changes) first. The [upgrade steps](#upgrading) list what to change.
+
+## ⚠️ Breaking Changes {#breaking-changes}
+
+### Nuxt 4.6 and Node 22
+
+Nuxt Robots v7 requires Nuxt `^4.6.0 || ^5.0.0`. Nuxt 3 support is removed. ([#333](https://github.com/nuxt-modules/robots/pull/333))
+
+It also requires Node `^22.22.3 || ^24.15.0 || >=26.0.0`.
+
+### Site Config v5
+
+Nuxt Robots loads Nuxt Site Config `>=5.0.0` as a module dependency. ([#345](https://github.com/nuxt-modules/robots/pull/345))
+
+Site Config v5 changes the priority of build time config. Check the [Site Config v5 release notes](/docs/site-config/releases/v5).
+
+### Nuxt Content v2 Removed
+
+The Nuxt Content integration now needs Nuxt Content v3. ([#350](https://github.com/nuxt-modules/robots/pull/350))
+
+If your project has Nuxt Content v2, Nuxt stops at startup with a version error. Upgrade to [Nuxt Content v3](https://content.nuxt.com) and add [`defineRobotsSchema()`](/docs/robots/advanced/content) to your collection.
+
+### Public Server and App Imports
+
+Import server functions from `#robots/server`. Import app composables from `#robots/app`. ([#333](https://github.com/nuxt-modules/robots/pull/333))
+
+The `#internal/nuxt-robots` and `#internal/nuxt-simple-robots` aliases are removed.
+
+```diff
+- import { getBotDetection } from '#robots/server/composables/getBotDetection'
+- import { defineEventHandler } from 'h3'
++ import { getBotDetection } from '#robots/server'
++ import { defineEventHandler } from 'nuxt/server'
+```
+
+Auto imports still work. If you disable the module, both aliases export typed mocks.
+If you disable bot detection, only the bot detection functions are mocks.
+
+### Server Hooks Use `RequestEvent`
+
+Server functions and Nitro hook contexts use the `RequestEvent` type from `nuxt/server`. ([#333](https://github.com/nuxt-modules/robots/pull/333))
+
+The change covers `getSiteRobotConfig()`, `getPathRobotConfig()`, `getBotDetection()`, `isBot()`, and `getBotInfo()`.
+It also covers the `robots:config` and `robots:robots-txt` hooks. Read request data from the web standard `Request`:
+
+```diff
+ nitroApp.hooks.hook('robots:config', (ctx) => {
+-  const host = ctx.event?.headers.get('host')
++  const host = ctx.event?.req.headers.get('host')
+ })
+```
+
+### `robots:config` Edits Stay in Their Request
+
+A `robots:config` hook that uses `ctx.event` now changes only the `robots.txt` response for that request. ([#340](https://github.com/nuxt-modules/robots/pull/340))
+
+In v6, these edits leaked into the indexing rules of later requests.
+To disable indexing for a host, use the `site-config:init` hook:
+
+```ts [server/plugins/robots-domain.ts]
+import { getNitroOrigin } from '#site-config/server'
+
+export default defineNitroPlugin((nitroApp) => {
+  nitroApp.hooks.hook('site-config:init', ({ event, siteConfig }) => {
+    if (new URL(getNitroOrigin(event)).hostname.startsWith('staging.'))
+      siteConfig.push({ indexable: false, _context: 'host-indexing' })
+  })
+})
+```
+
+See [Handling Multiple Domains](/docs/robots/guides/disable-indexing#handling-multiple-domains).
+
+### `useRobotsRule(true)` Respects Site Indexing
+
+`useRobotsRule(true)` now sends `robotsDisabledValue` when Site Config disables indexing. ([#340](https://github.com/nuxt-modules/robots/pull/340))
+
+In v6, it sent `robotsEnabledValue`, so a staging page could become indexable.
+`null` now keeps the existing rule, as `undefined` already did.
+
+### Route Rules Take Precedence Over Page Metadata
+
+When a route rule and `definePageMeta({ robots })` both match a path, the route rule now wins. ([#340](https://github.com/nuxt-modules/robots/pull/340))
+
+### Custom `robots.txt` Paths Use `srcDir`
+
+The module reads `assets/robots.txt`, `pages/robots.txt`, and `pages/_dir/robots.txt` from `srcDir`. ([#340](https://github.com/nuxt-modules/robots/pull/340))
+
+On Nuxt 4, the default `srcDir` is `app/`. If you keep that default, move these files to `app/assets/` or `app/pages/`.
+The `public/` and root directory paths do not change.
+
+### Zod 4 for `defineRobotsSchema()`
+
+The optional `zod` peer dependency is now `^4.6.5`. If you use `defineRobotsSchema()`, install Zod 4.6.5 or a later 4.x release. ([#333](https://github.com/nuxt-modules/robots/pull/333))
+
+### DevTools Needs `nuxtseo-devtools-host`
+
+The Robots DevTools panel needs the `nuxtseo-devtools-host` package as a dev dependency. ([#333](https://github.com/nuxt-modules/robots/pull/333))
+
+```bash
+pnpm add -D nuxtseo-devtools-host
+```
+
+Restart the dev server after you install it. You do not add it to `modules`.
+
+## 🚀 Features
+
+### Block AI Training Crawlers
+
+Set `blockAiBots: 'training'` to opt out of AI training. Dedicated AI search and answer agents stay unblocked. ([#349](https://github.com/nuxt-modules/robots/pull/349))
+
+```ts [nuxt.config.ts]
+export default defineNuxtConfig({
+  robots: {
+    blockAiBots: 'training',
+  },
+})
+```
+
+This mode blocks `GPTBot`, `ClaudeBot`, `Google-Extended`, `meta-externalagent`, `Applebot-Extended`, and `CCBot`.
+`Google-Extended` also controls Gemini grounding, so this mode opts out of grounding too.
+`blockAiBots: true` still blocks every token in the `AiBots` list.
+
+### Page Metadata Matches Your Router
+
+`definePageMeta({ robots: false })` now matches paths with the app router. ([#340](https://github.com/nuxt-modules/robots/pull/340), [#341](https://github.com/nuxt-modules/robots/pull/341), [#342](https://github.com/nuxt-modules/robots/pull/342))
+
+Matching respects `router.options` `sensitive`, `strict`, and `end`, plus aliases and nested routes.
+The module compiles the rules after page route transforms, such as Nuxt I18n locale routes.
+
+### Nuxt Content v3 in Any Module Order
+
+Nuxt Robots and Nuxt Content v3 now work in either module order. ([#340](https://github.com/nuxt-modules/robots/pull/340))
+
+Call `useRobotsRule()` with the frontmatter value to send the same rule in the `X-Robots-Tag` header:
+
+```ts
+useSeoMeta(page.value?.seo || {})
+useRobotsRule(page.value?.robots ?? undefined)
+```
+
+### Agent Skill
+
+The package ships an Agent Skill at `skills/nuxtjs-robots/SKILL.md`. ([#339](https://github.com/nuxt-modules/robots/pull/339))
+
+Coding agents read it from `node_modules/@nuxtjs/robots/skills/`.
+
+## 🔧 Bug Fixes
+
+- An explicit `allow` route rule now beats a catch-all `noindex` header. ([#337](https://github.com/nuxt-modules/robots/pull/337))
+- `disableNuxtContentIntegration` now turns off the Nuxt Content integration. ([#340](https://github.com/nuxt-modules/robots/pull/340))
+- A string `comment` in a `groups` entry now prints one comment line. ([#340](https://github.com/nuxt-modules/robots/pull/340))
+- `disallowNonIndexableRoutes` no longer writes `Disallow` lines for the build assets directory. ([#340](https://github.com/nuxt-modules/robots/pull/340))
+- `getBotDetection()` from `@nuxtjs/robots/util` now accepts a `Headers` instance. ([#340](https://github.com/nuxt-modules/robots/pull/340))
+- `useBotDetection()` builds without a separate `@vueuse/core` install. ([#333](https://github.com/nuxt-modules/robots/pull/333))
+- On Nitro 3, boolean `robots` route rules keep their meaning. ([#333](https://github.com/nuxt-modules/robots/pull/333))
+- On Nitro 3, the module sets `X-Robots-Tag` after the middleware header merge. ([#331](https://github.com/nuxt-modules/robots/pull/331))
+- App imports work on both Nuxt 4 and Nuxt 5. ([#331](https://github.com/nuxt-modules/robots/pull/331))
+
+## Upgrading
+
+1. Upgrade to Nuxt 4.6 or Nuxt 5. Use Node 22.22.3+, 24.15.0+, or 26+.
+2. If you use `@nuxtjs/seo`, follow the [Nuxt SEO v5 to v6 migration guide](/docs/nuxt-seo/migration-guide/v5-to-v6). Otherwise, install `@nuxtjs/robots@^7`.
+3. Replace `#internal/nuxt-robots` and deep `#robots/...` imports with `#robots/server` or `#robots/app`.
+4. Replace `event.headers` and `event.node.req` in robots hooks with `event.req.headers`.
+5. Move per host indexing from `robots:config` to the `site-config:init` hook.
+6. Move `assets/robots.txt` and `pages/robots.txt` into `srcDir`.
+7. If you use Nuxt Content v2, upgrade to Nuxt Content v3.
+8. To keep the DevTools panel, run `pnpm add -D nuxtseo-devtools-host`.
diff --git a/skills/nuxtjs-robots/SKILL.md b/skills/nuxtjs-robots/SKILL.md
index ee5b9371..9a0f35d2 100644
--- a/skills/nuxtjs-robots/SKILL.md
+++ b/skills/nuxtjs-robots/SKILL.md
@@ -7,7 +7,7 @@ compatibility: "Requires Node.js ^22.22.3 || ^24.15.0 || >=26.0.0. Requires Nuxt
 
 # @nuxtjs/robots
 
-Tested with `@nuxtjs/robots` 6.2.4 on Nuxt 4.6.0, with the Node server and `nuxt generate`.
+Covers `@nuxtjs/robots` 7. Runtime behaviour was tested with 6.2.4 on Nuxt 4.6.0; version 7 changes are listed under Version limits.
 The module serves `/robots.txt`, sends `X-Robots-Tag`, and adds `<meta name="robots">` to each SSR page.
 Site config decides if the whole site is indexable. Route rules decide per path.
 
@@ -23,6 +23,8 @@ export default defineNuxtConfig({
 })
 ```
 
+Included in `@nuxtjs/seo` v6. If the app uses `@nuxtjs/seo`, configure the `robots` key and do not add the module again.
+
 Site-wide indexing is a `site` key (nuxt-site-config), not a `robots` key.
 
 ## Automatic behaviour
@@ -31,7 +33,7 @@ Site-wide indexing is a `site` key (nuxt-site-config), not a `robots` key.
 - Indexable: robots.txt is `User-agent: *` and an empty `Disallow:`. Pages get `index, follow, max-image-preview:large, max-snippet:-1, max-video-preview:-1`.
 - Not indexable: robots.txt is `Disallow: /`. Every response gets `X-Robots-Tag: noindex, nofollow`, including API routes, `/robots.txt`, and `/_nuxt/` files.
 - `/_nuxt/**` always gets `X-Robots-Tag: noindex`.
-- The module merges a hand-written robots.txt from `public/_robots.txt`, `assets/robots.txt`, `pages/robots.txt`, or `robots.txt`, all relative to the project root. It ignores `app/assets/` and `app/pages/` without a warning. Nuxt renames `public/robots.txt` to `public/_robots.txt` in your source tree each time it loads the module, `nuxt prepare` included.
+- The module merges a hand-written robots.txt from `public/_robots.txt` or `robots.txt` in the project root, or from `assets/robots.txt` or `pages/robots.txt` in `srcDir` (`app/` on Nuxt 4). Nuxt renames `public/robots.txt` to `public/_robots.txt` in your source tree each time it loads the module, `nuxt prepare` included.
 - `nuxt generate` prerenders `/robots.txt`. A static preset that reads `_headers`, such as Netlify, gets the header rules there.
 - To turn parts off, set `robots.robotsTxt`, `robots.header`, or `robots.metaTag` to `false`. `useRobotsRule()` still writes a meta tag when `metaTag` is `false`. `robots.enabled: false` turns off everything, and the composables do nothing.
 
@@ -64,7 +66,7 @@ Decide per request, during SSR. Route rules cannot match a query string:
 import { useRobotsRule, useRoute } from '#imports'
 
 const route = useRoute()
-// Call it only to noindex. useRobotsRule(true) would make a staging page indexable.
+// A string rule replaces the default rule for this response.
 if (route.query.q)
   useRobotsRule('noindex, follow')
 </script>
@@ -103,15 +105,18 @@ export default defineNuxtConfig({
 })
 ```
 
-Server examples use Nitro auto-imports. On Nuxt 4.6, importing them from `#imports` in a server file fails `nuxt typecheck`.
+Import server functions from `#robots/server` and app composables from `#robots/app`. Auto-imports also work.
+Do not import server functions from `#imports`. On Nuxt 4.6, that fails `nuxt typecheck`.
 
 Turn off indexing for one host of a multi-domain deployment. This is per request:
 
 ```ts
 // server/plugins/robots-host.ts
+import { getNitroOrigin } from '#site-config/server'
+
 export default defineNitroPlugin((nitroApp) => {
   nitroApp.hooks.hook('site-config:init', ({ event, siteConfig }) => {
-    if (getRequestHost(event).startsWith('staging.'))
+    if (new URL(getNitroOrigin(event)).hostname.startsWith('staging.'))
       siteConfig.push({ _context: 'staging-host', indexable: false })
   })
 })
@@ -123,17 +128,20 @@ Detect bots from the `User-Agent` header:
 
 ```ts
 // server/middleware/block-bots.ts
+import { createError, defineEventHandler } from 'nuxt/server'
+import { getBotDetection } from '#robots/server'
+
 export default defineEventHandler((event) => {
   const bot = getBotDetection(event) // { isBot, botName, botCategory, trusted }
   // Search engines and AI crawlers count as trusted. To block AI crawlers, test bot.botCategory === 'ai'.
-  if (event.path.startsWith('/api/') && bot.isBot && !bot.trusted)
-    throw createError({ statusCode: 403 })
+  if (event.url.pathname.startsWith('/api/') && bot.isBot && !bot.trusted)
+    throw createError({ status: 403 })
 })
 ```
 
 In the app, `useBotDetection()` returns the same fields as computed refs.
 `useBotDetection({ fingerprint: true })` also runs BotD in the browser. Its result arrives later, so watch `isBot`.
-Outside Nuxt, `getBotDetection(headers)` from `@nuxtjs/robots/util` reads only a lowercase `user-agent` key. Pass `Object.fromEntries(request.headers)`. A `Headers` instance returns `isBot: false`.
+Outside Nuxt, `getBotDetection(headers)` from `@nuxtjs/robots/util` accepts a `Headers` instance or a plain header object.
 
 ## Integrations
 
@@ -142,25 +150,29 @@ Nuxt Content v3 frontmatter and Nuxt i18n path expansion: [references/integratio
 ## Traps
 
 - A path can be crawlable and noindex, or blocked in robots.txt. It cannot be both. Google never reads `noindex` on a disallowed URL and can still index it from links.
-- `useRobotsRule(true)` sends `robotsEnabledValue` even when the site is not indexable. Pass `true` only to override that.
+- `useRobotsRule(true)` keeps `noindex` when the site is not indexable. To force a rule, pass the string, such as `useRobotsRule('index, follow')`.
+- `useRobotsRule(null)` and `useRobotsRule(undefined)` keep the existing rule.
 - `useRobotsRule()` sets the rule only during SSR. In the browser it only reads the rendered value.
 - `noai` or `noimageai` makes a path non-indexable to the module, even with `index: true`. `getPathRobotConfig` returns `indexable: false`, and `disallowNonIndexableRoutes` disallows the path.
 - An `app.baseURL` other than `/` turns off robots.txt and logs an error. Crawlers only read `/robots.txt` at the host root.
 - Nuxt Content frontmatter `robots` sets the meta tag but not the header, and it needs `zod` 4 in your dependencies. See [references/integrations.md](references/integrations.md).
-- The key is `blockAiBots`. Some docs prose spells it `blockAIBots`.
+- The key is `blockAiBots`, not `blockAIBots`.
 - Use `blockAiBots: 'training'` to leave dedicated AI search and user fetch agents unblocked. Existing rules still apply.
 - Training mode includes Google-Extended, which also controls grounding in Gemini Apps and Vertex AI.
 
 ## Version limits
 
-These hold for 6.2.4:
-
-- `definePageMeta({ robots })` has no effect on Nuxt 4.6. Use a route rule, `defineRouteRules()`, or `useRobotsRule()`.
-- A build that calls `useBotDetection()` fails with `Rolldown failed to resolve import "@vueuse/core"` unless the app installs `@vueuse/core`.
-- `robots.disableNuxtContentIntegration` has no effect.
-- The docs' `robots:config` Nitro hook that edits `ctx.groups` per host leaks. The result is shared, so other hosts get `noindex` until the next `/robots.txt` request. Use the `site-config:init` plugin above.
-- `groups[].comment` must be an array. A string prints one `#` line per character.
-- `disallowNonIndexableRoutes: true` (deprecated) also writes `Disallow: /_nuxt` and `Disallow: /_nuxt/*` into robots.txt.
+Version 7 breaks these 6.x patterns:
+
+- Nuxt 3 and Node 20 are unsupported. Use Nuxt `^4.6.0 || ^5.0.0` and Node `^22.22.3 || ^24.15.0 || >=26.0.0`.
+- `#internal/nuxt-robots` is removed. Import from `#robots/server` or `#robots/app`.
+- Hook events are `RequestEvent` from `nuxt/server`. Read `ctx.event.req.headers.get('host')`, not `ctx.event.headers` or `event.node.req`.
+- A `robots:config` edit made with `ctx.event` changes only that `/robots.txt` response. For per host indexing, use the `site-config:init` plugin above.
+- A route rule beats `definePageMeta({ robots })` on the same path.
+- `assets/robots.txt` and `pages/robots.txt` load from `srcDir`. On Nuxt 4, move them into `app/`.
+- Nuxt Content v2 is unsupported, and Nuxt fails at startup with a version error. Upgrade to Nuxt Content v3.
+- `defineRobotsSchema()` needs `zod` `^4.6.5`.
+- The DevTools panel needs `nuxtseo-devtools-host` as a dev dependency: `pnpm add -D nuxtseo-devtools-host`.
 
 ## Config
 
diff --git a/skills/nuxtjs-robots/references/integrations.md b/skills/nuxtjs-robots/references/integrations.md
index b0c6a01c..b0b30c5f 100644
--- a/skills/nuxtjs-robots/references/integrations.md
+++ b/skills/nuxtjs-robots/references/integrations.md
@@ -33,7 +33,7 @@ const route = useRoute()
 const { data: page } = await useAsyncData(`page-${route.path}`, () => queryCollection('content').path(route.path).first())
 // The meta tag comes from page.seo. Only useRobotsRule() also sets the header.
 useSeoMeta(page.value?.seo || {})
-// Content returns null for an unset key. null sends X-Robots-Tag: null and drops the meta tag.
+// Content returns null for an unset key. Robots 7 ignores null and undefined, so the default rule stays.
 useRobotsRule(page.value?.robots ?? undefined)
 </script>