Feature: client-side site search (Pagefind)

.github/workflows/deploy.yml

@@ -44,6 +44,15 @@ jobs:
       - name: Build site
         run: pelican content -o output -s publishconf.py
 
+      - name: Build search index (Pagefind)
+        # Static same-origin search: writes the UI assets and chunked index to
+        # output/pagefind/, all fetched same-origin, so a strict script-src /
+        # connect-src 'self' CSP keeps working. Indexing is scoped by
+        # data-pagefind-body (articles + content pages only), which keeps the
+        # noindexed /drafts/ pages OUT of the search index. Version pinned;
+        # the runner's preinstalled Node is sufficient.
+        run: npx --yes pagefind@1.5.2 --site output
+
       - name: Add CNAME file
         run: echo rivassec.com > output/CNAME
 

pelicanconf.py

@@ -127,7 +127,8 @@ def _asset_version() -> str:
     # tags.html and archives.html are noindex,follow (see base.html), so keep
     # them out of the sitemap too - never advertise a URL we ask Google not to
     # index. Surfaced by GSC 'Discovered - not indexed' 2026-08-16.
-    'exclude': ['tag/', 'tags.html', 'archives.html'],
+    # search/ is a noindex client-side search shell (see pagefind_search.html).
+    'exclude': ['tag/', 'tags.html', 'archives.html', 'search/'],
 }
 STATIC_PATHS.append('extra')
 # Client-side tools ship verbatim: content/tools/ -> /tools/ (no remap, no
@@ -187,6 +188,7 @@ def _asset_version() -> str:
 MENUITEMS = [
     ('About', 'https://rivassec.com/pages/about.html'),
     ('Categories', 'https://rivassec.com/categories.html'),
+    ('Search', 'https://rivassec.com/search/'),
     ('Tools', 'https://rivassec.com/tools/iam-blast-radius/'),
     ('GitHub', 'https://github.com/rivassec'),
 ]
@@ -199,6 +201,9 @@ def _asset_version() -> str:
     'llms_txt.html': 'llms.txt',
     'llms_full_txt.html': 'llms-full.txt',
     'llms_txt_wellknown.html': '.well-known/llms.txt',
+    # Client-side search shell; the index under /pagefind/ is generated at
+    # deploy time (see deploy.yml's Pagefind step). noindex + sitemap-excluded.
+    'pagefind_search.html': 'search/index.html',
 }
 
 # Single source of truth for the llms.txt header blockquote, referenced by all

themes/Flex/static/js/search-init.js

@@ -0,0 +1,13 @@
+/* Initializes the Pagefind UI on /search/. Lives in an external file on
+ * purpose: the site ships zero inline scripts (script-src 'self',
+ * precondition enforced by scripts/csp_audit.py in CI). Loaded with defer
+ * after pagefind-ui.js, so PagefindUI is defined and the DOM is parsed. */
+(function () {
+  "use strict";
+  if (typeof PagefindUI === "undefined") { return; }
+  new PagefindUI({
+    element: "#search",
+    showSubResults: true,
+    showImages: false,
+  });
+})();

themes/Flex/templates/article.html

@@ -61,7 +61,10 @@ <h1 id="{{ article.slug }}" itemprop="headline">{{ article.title }}</h1>
   </aside>
   {% endif %}
 
-  <div itemprop="articleBody">
+  {# data-pagefind-body scopes the search index: with the attribute present
+     anywhere, Pagefind indexes ONLY elements carrying it - so drafts (which
+     also render through this template) stay out of the search index. #}
+  <div itemprop="articleBody"{% if article.status != 'draft' %} data-pagefind-body{% endif %}>
     {{ article.content }}
   </div>
 </article>

themes/Flex/templates/page.html

@@ -24,7 +24,7 @@
     {{ translations.translations_for(page) }}
     <h1 id="{{ page.slug }}">{{ page.title }}</h1>
   </header>
-  <div>
+  <div{% if page.status != 'draft' and page not in hidden_pages %} data-pagefind-body{% endif %}>
     {% block before_content %}{% endblock %}
     {% block page_content %}{{ page.content }}{% endblock %}
     {% block after_content %}{% endblock %}

themes/Flex/templates/pagefind_search.html

@@ -0,0 +1,31 @@
+{% extends "base.html" %}
+{% block title %}Search - {{ SITETITLE|default(SITENAME) }}{% endblock %}
+{% block canonical %}<link rel="canonical" href="{{ SITEURL }}/search/">{% endblock %}
+{% block meta %}
+  <meta name="author" content="{{ AUTHOR }}" />
+  <meta name="description" content="Search the posts on rivassec.com." />
+  {# Search shells are thin content; keep out of the index but followable.
+     Matched by the sitemap exclude for search/ so the canonical/noindex
+     guard stays satisfied. #}
+  <meta name="robots" content="noindex, follow" />
+  <link rel="stylesheet" href="{{ SITEURL }}/pagefind/pagefind-ui.css" />
+{% endblock %}
+{% block content %}
+<article class="single">
+  <header>
+    <h1 id="search-heading">Search</h1>
+  </header>
+  <p>Search every published post. Results are computed in your browser from a
+  static index; no query leaves this page.</p>
+  <div id="search" aria-labelledby="search-heading"></div>
+  <noscript>
+    <p>Search needs JavaScript. Browse by
+    <a href="{{ SITEURL }}/categories.html">category</a> or read the
+    <a href="{{ SITEURL }}/devsecops-guide.html">DevSecOps Guide</a> instead.</p>
+  </noscript>
+</article>
+{% endblock %}
+{% block additional_js %}
+<script src="{{ SITEURL }}/pagefind/pagefind-ui.js" defer></script>
+<script src="{{ SITEURL }}/theme/js/search-init.js" defer></script>
+{% endblock %}