Update from code changes: offline search in mint export

deploy/export.mdx

@@ -12,7 +12,7 @@
 
 This is useful when you need to distribute documentation to users who can't access your live site, such as for on-premise customers or internal compliance reviews.
 
 The CLI bundles the icons, fonts, and LaTeX styles your site uses into the archive, so exported docs work in air-gapped environments without internet access. Exports are served from the root of a domain unless you set a [base path](#serve-under-a-subpath). Review the [limitations](#limitations) before you deploy an export.
 
 ## Prerequisites
 
@@ -21,11 +21,16 @@
 
 ## Limitations
 
-- Offline exports do not support search. To add offline search, index the exported HTML with a static search tool.
 - Offline exports do not support the Mintlify assistant.
 - Offline exports do not include analytics or other [integrations](/organize/settings-integrations) configured in your `docs.json`.
 - Images and other files that your pages reference by external URL are not bundled. Store them in your project to include them in the archive.
 
+### Offline search
+
+The search bar works in exported docs without a network connection. When you run `mint export`, the CLI builds a search index from the exported pages and saves it in the archive. The browser searches that index locally.
+
+Offline search follows the same rules as search on your live site. It skips pages that are [hidden from search](/organize/hidden-pages), such as pages with `noindex: true` or `searchable: false` in their frontmatter. Results only include pages in the reader's current language and version.
+
 ### Bundled assets
 
 When you run `mint export`, the CLI downloads assets that your site loads from external hosts and saves them in the archive. This includes:
@@ -84,7 +89,7 @@
 The exported archive is a static, unauthenticated bundle. Anyone with the zip file can view every page it contains.
 
 - The `--groups` flag controls which group-restricted pages Mintlify includes in the archive at export time. It does not enforce authentication after export.
 - Only share archives with recipients who are allowed to see the content they contain.
 - Regenerate and redistribute the archive to publish new changes. Anyone running an older zip continues to see the older content.
 
 ## What's in the zip
@@ -97,6 +102,7 @@
 | `index.txt` files | Page data for client-side navigation, saved next to each page's HTML |
 | `_next/static/` | Bundled CSS, JavaScript, and other static assets |
 | `_mintlify/export-assets/` | Icons, fonts, and other assets downloaded from external hosts |
+| `_mintlify/search-index.json` | Search index for [offline search](#offline-search) |
 | `llms.txt` | Index of the exported pages for AI tools |
 | `README.md` | Instructions for viewing the export and hosting it on your own server |
 | `serve.js` | Zero-dependency Node.js HTTP server that serves the site |
@@ -139,7 +145,7 @@
 To host an export on your own web server instead of `serve.js`, serve the unzipped folder as static files from the root of a domain. If you exported with `--base-path`, serve the folder under that path instead, and apply these rules to the part of the request path after the base path:
 
 1. **Pages:** Serve `<path>/index.html` for a request to `/<path>`.
 2. **Client-side navigation:** For a request with an `RSC: 1` header or an `_rsc` query parameter, serve `<path>/index.txt` with `Content-Type: text/x-component` and `Vary: RSC`. Without this rule, every link click loads a full page.
 3. **Caching:** Serve `_next/static/*` and `_mintlify/export-assets/*` with `Cache-Control: public, max-age=31536000, immutable`. Serve pages and `index.txt` files with `Cache-Control: no-cache`.
 
 `serve.js` implements all three rules, so you can use it as a reference.