diff --git a/config/urls.py b/config/urls.py
index 35f278f..620dfdd 100644
--- a/config/urls.py
+++ b/config/urls.py
@@ -15,6 +15,7 @@
 from mpicms.personal.views import ContactListView, RawContactListView
 from mpicms.events.views import ics_view
 from mpicms.publications.views import PublicationsFeedView
+from mpicms.publications.views import PublicationsRssFeedView
 
 
 urlpatterns = [
@@ -48,7 +49,8 @@
     path('contactsraw/', RawContactListView.as_view(), name='contactsraw'),
 
     path('events/ics', ics_view, name='ics'),
-    path('feed/last_publications', PublicationsFeedView.as_view()),
+    path('feed/last_publications_xml', PublicationsFeedView.as_view()),
+    path('feed/last_publications', PublicationsRssFeedView.as_view()),
 ] + static(
     settings.MEDIA_URL, document_root=settings.MEDIA_ROOT
 )
diff --git a/mpicms/publications/models.py b/mpicms/publications/models.py
index 5e157e9..de164ec 100644
--- a/mpicms/publications/models.py
+++ b/mpicms/publications/models.py
@@ -10,14 +10,17 @@
 
 from modelcluster.models import ClusterableModel
 
+
 class LatestFirstOrderable(Orderable):
     def save(self, *args, **kwargs):
         if self.pk is None:
             self.sort_order = self.__class__.objects.aggregate(models.Min('sort_order'))['sort_order__min'] - 1
         super(Orderable, self).save(*args, **kwargs)
+
     class Meta:
         abstract = True
 
+
 @register_snippet
 class Publication(index.Indexed, ClusterableModel, LatestFirstOrderable):
     title = RichTextField(_('title'), features=['bold', 'italic', 'link'])
diff --git a/mpicms/publications/views.py b/mpicms/publications/views.py
index 9c9bac4..fc3dd5e 100644
--- a/mpicms/publications/views.py
+++ b/mpicms/publications/views.py
@@ -3,16 +3,22 @@
 
 from .models import Publication
 
+
 class PublicationsFeedView(View):
 
+    # IMPORTANT: Quick hack to get experiments going, This assumes, that the
+    # editor of the publications is trusted to put only expected values into the
+    # fields.
+    #
     # 2DO: check for legal XML characters everywhere https://www.w3.org/TR/xml/#NT-Char
     # 2DO: Put HTML content in <![CDATA[...]]> ?
     # If we don't make DOI into element:
-    #   2DO: escape invalid characterds in attribute ( &quot; &amp; &lt; )
-    # If we don't convert the rich text fields to plain text)
-    #    2DO: Remove editor artefacts from ricch text fields ("<p>blabla</p>") ?
+    #   2DO: escape invalid characterds in attribute ( &quot; &amp; &lt; &gt; )
+    # If we don't convert the rich text fields to plain text maybe:
+    #    2DO: Remove <p> editor artefacts from rich text fields ("<p>blabla</p>") ?
     #    2DO: Remove <i> from source fields ("<p><i>Mol Cell.</i></p>") ?
     # maybe use xml.dom for that?
+    # 2DO: ... ?
 
     def get(self, request, *args, **kwargs):
         response = HttpResponse(content_type="application/rss+xml; charset=utf-8")
@@ -20,10 +26,49 @@ def get(self, request, *args, **kwargs):
         response.write('<publication-list>\n')
         for item in Publication.objects.order_by('sort_order')[:3]:
             response.write(f'  <publication doi="{item.doi}">\n')
-            response.write(f'    <authors>{item.authors}</authors>\n');
-            response.write(f'    <title>{item.title}</title>\n');
-            response.write(f'    <source>{item.source}</source>\n');
-            response.write(f'    <groups>{item.groups}</groups>\n');
+            response.write(f'    <authors>{item.authors}</authors>\n')
+            response.write(f'    <title>{item.title}</title>\n')
+            response.write(f'    <source>{item.source}</source>\n')
+            response.write(f'    <groups>{item.groups}</groups>\n')
             response.write('  </publication>\n')
         response.write('</publication-list>\n')
-        return response;
+        return response
+
+
+class PublicationsRssFeedView(View):
+
+    # IMPORTANT: Quick hack to get experiments going, This assumes, that the
+    # editor of the publications is trusted to put only expected values into the
+    # fields.
+    #
+    # 2DO: check for legal XML characters everywhere https://www.w3.org/TR/xml/#NT-Char
+    # 2DO: Put HTML content in <![CDATA[...]]> ?
+    # If we don't make DOI into element:
+    #   2DO: escape invalid characterds in attribute ( &quot; &amp; &lt; &gt; )
+    # If we don't convert the rich text fields to plain text maybe:
+    #    2DO: Remove <p> editor artefacts from rich text fields ("<p>blabla</p>") ?
+    #    2DO: Remove <i> from source fields ("<p><i>Mol Cell.</i></p>") ?
+    # maybe use xml.dom for that?
+    # 2DO: ... ?
+
+    def get(self, request, *args, **kwargs):
+        response = HttpResponse(content_type="application/rss+xml; charset=utf-8")
+        response.write('<?xml version="1.0" encoding="utf-8" ?>\n')
+        response.write('<rss version="2.0">\n')
+        response.write('  <channel xmlns:pl="http://www.molgen.mpg.de/xml/publication_feed">\n')
+        # the next three elements are required https://www.rssboard.org/rss-specification
+        response.write('    <title>Last Three Publications</title>\n')
+        response.write('    <link>https://intranet.molgen.mpg.de/</link>\n')
+        response.write('    <description>The last three publications published on the Intranet site</description>\n')
+        for item in Publication.objects.order_by('sort_order')[:3]:
+            # all elements optional, but either title or description allowed
+            response.write(f'    <item pl:doi="{item.doi}">\n')
+            response.write(f'      <title>{item.title}</title>\n')
+            response.write(f'      <pl:authors>{item.authors}</pl:authors>\n')
+            response.write(f'      <pl:title>{item.title}</pl:title>\n')
+            response.write(f'      <pl:source>{item.source}</pl:source>\n')
+            response.write(f'      <pl:groups>{item.groups}</pl:groups>\n')
+            response.write('     </item>\n')
+        response.write('  </channel>\n')
+        response.write('</rss>\n')
+        return response