RSS Improvements

Enter your suggestions in details:

I got feedback that RSS posts would be useful if they had more preview context. For example, this is what blasts into Slack...

I haven't looked at other feed experiences. It could be variable.

I looked into this a bit, I think the Node.js® is a free, open-source, cross-platform... is not coming from the RSS config, but rather the RSS is pointing to a page which Slack (or other tools which perform link previews) is reading a meta tag from. The meta tag looks to be driven from the individual article's description frontmatter property, or if none exists then the site's description property.

My notes on how all this flows together

https://nodejs.org/en/feed/blog.xml contains this object:

<title>
<![CDATA[ Node.js 22.22.0 (LTS) ]]>
</title>
<link>https://nodejs.org/en/blog/release/v22.22.0</link>
<guid isPermaLink="false">/blog/release/v22.22.0?1768313889314</guid>
<pubDate>Tue, 13 Jan 2026 14:18:09 GMT</pubDate>
</item>

Grabbing https://nodejs.org/en/blog/release/v22.22.0 contains the page description in the <meta name="description" content="..." /> tag:

For the blog specifically, that metadata is being generated into Next.js using a generateMetadata export:

nodejs.org/apps/site/app/[locale]/blog/[...path]/page.tsx

Lines 17 to 20 in f0cb063

    
           // This generates each page's HTML Metadata 
        
           // @see https://nextjs.org/docs/app/api-reference/functions/generate-metadata 
        
           export const generateMetadata = ({ params }: PageParams) => 
        
             basePage.generateMetadata({ params, prefix: 'blog' });

nodejs.org/apps/site/next.dynamic.page.mjs

Lines 26 to 44 in f0cb063

    
           /** 
        
            * This generates each page's HTML Metadata 
        
            * 
        
            * @see https://nextjs.org/docs/app/api-reference/functions/generate-metadata 
        
            * 
        
            * @param {{ params: Promise<{ path: Array<string>; locale: string }>, prefix?: string }} props 
        
            * @returns {Promise<import('next').Metadata>} the metadata for the page 
        
            */ 
        
           export const generateMetadata = async ({ params, prefix }) => { 
        
             const { path = [], locale = defaultLocale.code } = await params; 
        
             const pathname = dynamicRouter.getPathname(path); 
        
             return dynamicRouter.getPageMetadata( 
        
               locale, 
        
               // If there's a prefix, `join` it with the pathname 
        
               prefix ? join(prefix, pathname) : pathname 
        
             ); 
        
           };

nodejs.org/apps/site/next.dynamic.mjs

Lines 188 to 262 in f0cb063

    
             /** 
        
              * This method generates the Next.js App Router Metadata 
        
              * that can be used for each page to provide metadata 
        
              * 
        
              * @param {string} locale 
        
              * @param {string} path 
        
              * @returns {Promise<import('next').Metadata>} 
        
              */ 
        
             const _getPageMetadata = async (locale = defaultLocale.code, path = '') => { 
        
               const pageMetadata = { ...PAGE_METADATA }; 
        
               const { source = '' } = await getMarkdownFile(locale, path); 
        
               const { data } = matter(source); 
        
               const getUrlForPathname = (l, p) => 
        
                 `${baseUrlAndPath}/${l}${p ? `/${p}` : ''}`; 
        
               // Default Title for the page 
        
               pageMetadata.title = data.title 
        
                 ? `${siteConfig.title} — ${data.title}` 
        
                 : siteConfig.title; 
        
               pageMetadata.description = data.description 
        
                 ? data.description 
        
                 : siteConfig.description; 
        
               // Default Twitter Title for the page 
        
               pageMetadata.twitter.title = pageMetadata.title; 
        
               // Default Open Graph Image for the page 
        
               pageMetadata.openGraph.images = [ 
        
                 ENABLE_STATIC_EXPORT 
        
                   ? `${defaultLocale.code}/next-data/og/announcement/Run JavaScript Everywhere` 
        
                   : `${defaultLocale.code}/next-data/og/${data.category ?? DEFAULT_CATEGORY_OG_TYPE}/${pageMetadata.title}`, 
        
               ]; 
        
               // Default canonical URL for the page 
        
               pageMetadata.alternates.canonical = 
        
                 data.canonical ?? getUrlForPathname(locale, path); 
        
               // Default alternate URL for the page in the default locale 
        
               pageMetadata.alternates.languages['x-default'] = getUrlForPathname( 
        
                 defaultLocale.code, 
        
                 path 
        
               ); 
        
               // Retrieves a matching blog feed for the category of the blog post 
        
               // If no matching blog feed is found, we simply fallback to the default blog feed 
        
               const matchingBlogFeed = siteConfig.rssFeeds.find( 
        
                 feed => feed.category === data.category 
        
               ); 
        
               // Adds the RSS Feed URL to the page metadata, if a matching feed is found 
        
               // otherwise, we fallback to the default blog feed 
        
               pageMetadata.alternates.types['application/rss+xml'] = getUrlForPathname( 
        
                 locale, 
        
                 `feed/${matchingBlogFeed?.file ?? 'blog.xml'}` 
        
               ); 
        
               // Iterate all languages to generate alternate URLs for each language 
        
               availableLocaleCodes.forEach(currentLocale => { 
        
                 pageMetadata.alternates.languages[currentLocale] = getUrlForPathname( 
        
                   currentLocale, 
        
                   path 
        
                 ); 
        
               }); 
        
               return pageMetadata; 
        
             }; 
        
             // Creates a Cached Version of the Page Metadata Context 
        
             const getPageMetadata = cache(async (locale, path) => { 
        
               return await _getPageMetadata(locale, path); 
        
             });

Data from_getPageMetadata is either sourced from the specific markdown article's frontmatter content or from the entire site's config:

nodejs.org/apps/site/site.json

Line 3 in f0cb063

    
           "description": "Node.js® is a free, open-source, cross-platform JavaScript runtime environment that lets developers create servers, web apps, command line tools and scripts.",

Well, the problem is that we don't really have summaries for our blog posts. Just the content, and authors aren't used on writing summaries.

We could source the first paragraph form the article itself ... but not sure if that'd be a good idea.

so we'd have to institute a process change for releasers to write some sort of top-level description, right? certainly puts a bit of a wrench in the automation

Maybe we could generate summaries with AI

Maybe we could generate summaries with AI

That'd be a bit expensive... Where would we store them or when to generate them?

so we'd have to institute a process change for releasers to write some sort of top-level description, right? certainly puts a bit of a wrench in the automation

🫠

Maybe we could generate summaries with AI

That'd be a bit expensive... Where would we store them or when to generate them?

Why expensive? The posts aren't that long. We can automate this using actions and can save the result on the post's frontmatter

Maybe we could generate summaries with AI

That'd be a bit expensive... Where would we store them or when to generate them?

Why expensive? The posts aren't that long. We can automate this using actions and can save the result on the post's frontmatter

Who paying the AI bill, you?

Cause automated (non-human-originating) AI agentic/task usage is against the ToS of pretty much every provider out there.

👀 I think we'd be fine to set the RSS description to the first paragraph of the article, with links + markup stripped out.

let's try Matt idea it's seem good idea

Navigation Menu

Repository navigation

RSS Improvements #8544

Enter your suggestions in details:

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

	// This generates each page's HTML Metadata
	// @see https://nextjs.org/docs/app/api-reference/functions/generate-metadata
	export const generateMetadata = ({ params }: PageParams) =>
	basePage.generateMetadata({ params, prefix: 'blog' });

	/**
	* This generates each page's HTML Metadata
	*
	* @see https://nextjs.org/docs/app/api-reference/functions/generate-metadata
	*
	* @param {{ params: Promise<{ path: Array<string>; locale: string }>, prefix?: string }} props
	* @returns {Promise<import('next').Metadata>} the metadata for the page
	*/
	export const generateMetadata = async ({ params, prefix }) => {
	const { path = [], locale = defaultLocale.code } = await params;

	const pathname = dynamicRouter.getPathname(path);

	return dynamicRouter.getPageMetadata(
	locale,
	// If there's a prefix, `join` it with the pathname
	prefix ? join(prefix, pathname) : pathname
	);
	};

	/**
	* This method generates the Next.js App Router Metadata
	* that can be used for each page to provide metadata
	*
	* @param {string} locale
	* @param {string} path
	* @returns {Promise<import('next').Metadata>}
	*/
	const _getPageMetadata = async (locale = defaultLocale.code, path = '') => {
	const pageMetadata = { ...PAGE_METADATA };

	const { source = '' } = await getMarkdownFile(locale, path);

	const { data } = matter(source);

	const getUrlForPathname = (l, p) =>
	`${baseUrlAndPath}/${l}${p ? `/${p}` : ''}`;

	// Default Title for the page
	pageMetadata.title = data.title
	? `${siteConfig.title} — ${data.title}`
	: siteConfig.title;

	pageMetadata.description = data.description
	? data.description
	: siteConfig.description;

	// Default Twitter Title for the page
	pageMetadata.twitter.title = pageMetadata.title;

	// Default Open Graph Image for the page
	pageMetadata.openGraph.images = [
	ENABLE_STATIC_EXPORT
	? `${defaultLocale.code}/next-data/og/announcement/Run JavaScript Everywhere`
	: `${defaultLocale.code}/next-data/og/${data.category ?? DEFAULT_CATEGORY_OG_TYPE}/${pageMetadata.title}`,
	];

	// Default canonical URL for the page
	pageMetadata.alternates.canonical =
	data.canonical ?? getUrlForPathname(locale, path);

	// Default alternate URL for the page in the default locale
	pageMetadata.alternates.languages['x-default'] = getUrlForPathname(
	defaultLocale.code,
	path
	);

	// Retrieves a matching blog feed for the category of the blog post
	// If no matching blog feed is found, we simply fallback to the default blog feed
	const matchingBlogFeed = siteConfig.rssFeeds.find(
	feed => feed.category === data.category
	);

	// Adds the RSS Feed URL to the page metadata, if a matching feed is found
	// otherwise, we fallback to the default blog feed
	pageMetadata.alternates.types['application/rss+xml'] = getUrlForPathname(
	locale,
	`feed/${matchingBlogFeed?.file ?? 'blog.xml'}`
	);

	// Iterate all languages to generate alternate URLs for each language
	availableLocaleCodes.forEach(currentLocale => {
	pageMetadata.alternates.languages[currentLocale] = getUrlForPathname(
	currentLocale,
	path
	);
	});

	return pageMetadata;
	};

	// Creates a Cached Version of the Page Metadata Context
	const getPageMetadata = cache(async (locale, path) => {
	return await _getPageMetadata(locale, path);
	});

RSS Improvements #8544

Description

Enter your suggestions in details:

Activity

10hendersonm commented on Jan 13, 2026

ovflowd commented on Jan 23, 2026

ovflowd commented on Jan 23, 2026

bmuenzenmeyer commented on Feb 1, 2026

araujogui commented on Feb 1, 2026

ovflowd commented on Feb 8, 2026

ovflowd commented on Feb 8, 2026

araujogui commented on Feb 13, 2026

ovflowd commented on Feb 14, 2026

ovflowd commented on Feb 14, 2026

MattIPv4 commented on Mar 6, 2026

AugustinMauroy commented on Mar 7, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions