Yandex News Story Clusters Across Russia & the CIS

Yandex News groups coverage of the same event from many outlets into a story cluster. This actor reads the public Dzen News surface that now serves Yandex News, returns the clusters, and lists every linked article inside each one. Search by keyword, browse a category, read the homepage sections, or paste a story URL. No login, API key, or proxy is needed.

What this actor does

• Four ways to collect news: keyword search, 16 category feeds, the 15 homepage sections, or specific story URLs and cluster IDs.
• Story clusters, not just headlines: each story carries its cluster ID, source-reported coverage count, actual article count, unique source list, topics, image, country/region and timestamps.
• Every linked article: article records include the publisher, headline, clean article URL, publication date, image and source flags.
• Filters and sorting on the returned set: date window, source domain, publisher, keyword, relevance, newest, oldest or most covered.
• Automatic deduplication by story ID and article ID, and empty fields are omitted rather than stored as nulls.

Output per story

Field	Description
recordType	"story"
storyId	Story cluster ID (UUID); shared with its article records
persistentId	Yandex's numeric story ID
storyUrl	Canonical story page on Dzen News
headline	Cluster headline
summary	Short cluster summary, when Yandex provides one
category	Broad rubric name, when available
topics / topicTags	Rubric or topic labels attached to the cluster
topicIds	Numeric rubric IDs
articleCount	Number of publications Yandex reports for the cluster
articlesReturned	Number of article records actually collected
sourceCount	Number of distinct outlet names among the collected articles
sources	Outlet names, when articles were collected
leadSource	Outlet behind the lead article
leadArticleUrl	Publisher URL of the lead article
publishedAt	Cluster timestamp, ISO UTC
publishedAtRaw	Timestamp label as shown by Yandex
imageUrl	Cluster image, when present
country / region	Story location, when Yandex reports it
importance / relatedStories	Yandex story metrics
commentsCount / likesCount	Interaction counts, when present
domain	ru
scrapedAt	UTC timestamp when collected

articlesReturned, sourceCount and sources are emitted only when includeArticles is true. With it off, no article records are fetched, so those article-derived fields are omitted.

Output per article

Field	Description
recordType	"article"
storyId	Parent story cluster ID
docId	Yandex document ID
headline	Outlet's headline
summary	Article snippet, when Yandex provides one
sourceName	Outlet name
sourceUrl	Outlet site root
articleUrl	Clean link to the article
publishedAt	Publication date (YYYY-MM-DD)
publishedAtIso	Publication timestamp, ISO UTC
publishedAtRaw	Timestamp label as shown by Yandex
imageUrl	Outlet logo or article image
isForeignAgent / isDzenArticle	Yandex flags
agencyId	Yandex outlet ID
domain	ru
scrapedAt	UTC timestamp when collected

Output per homepage record

topStories (and the legacy getTrending mode) emits one recordType: "trending" record with topics (homepage section names), sectionCounts, domain and scrapedAt, followed by the story records for those sections.

Input

Field	Type	Default	Description
mode	string	searchNews	searchNews, browseCategory, topStories or byStoryUrls
searchQueries	array	–	Keywords or phrases (mode=searchNews); the daily-test prefill uses ["Яндекс"]
category	string	politics	Rubric slug or all (mode=browseCategory)
storyUrls	array	–	Story/instory URLs or cluster UUIDs (mode=byStoryUrls)
maxStories	integer	20	Maximum story clusters across the run (1–500)
sortOrder	string	newest	relevance, newest, oldest or mostCovered (mode=searchNews)
timeRange	string	any	any, day, week or month (mode=searchNews)
publishedAfter	string	–	Inclusive YYYY-MM-DD start date (mode=searchNews)
publishedBefore	string	–	Inclusive YYYY-MM-DD end date (mode=searchNews)
includeDomains	array	[]	Keep clusters with an article from any listed domain
excludeDomains	array	[]	Drop clusters with an article from any listed domain
includePublishers	array	[]	Keep clusters matching any outlet name substring
excludePublishers	array	[]	Drop clusters matching any outlet name substring
includeKeywords	array	[]	Keep clusters matching any keyword in the title, summary or articles
excludeKeywords	array	[]	Drop clusters matching any keyword
includeDetails	boolean	true	Fetch cluster details such as topics, reported count, image and metrics
includeArticles	boolean	true	Emit one record per linked article; when off, article-derived story fields are omitted
maxArticlesPerStory	integer	100	Maximum articles collected per cluster (1–1000)

Example: brand monitoring by keyword

{
  "mode": "searchNews",
  "searchQueries": ["Яндекс", "Яндекс Такси"],
  "maxStories": 50,
  "timeRange": "week",
  "includeArticles": true
}

Example: all categories, story records only

{
  "mode": "browseCategory",
  "category": "all",
  "maxStories": 200,
  "includeArticles": false
}

Example: homepage top stories

{
  "mode": "topStories",
  "maxStories": 40
}

Example: resolve specific story clusters

{
  "mode": "byStoryUrls",
  "storyUrls": [
    "https://dzen.ru/news/story/85c864f8-36d7-5b96-ab8d-937bea5d9c9b",
    "c29b2d3c-c183-546b-9a23-b9aba12e981e"
  ],
  "maxArticlesPerStory": 300
}

Use cases

• Media monitoring: follow a company, person or event across many Russian outlets.
• Share-of-voice measurement: use the reported and returned article counts to size coverage.
• PR and communications: see which outlets picked up a story and how the headline evolved.
• Newsroom research: export a cluster's full article list with dates and publisher links.
• Risk and compliance: monitor keywords, source domains and publication windows.

FAQ

What happened to Yandex News? The public Yandex News pages now redirect to Dzen News, which provides the same story-cluster model. This actor reads that public surface.

How many stories can one search return? Yandex search returns 20 clusters per page and its depth varies by query; observed runs reached 9–22 pages (roughly 175–400 unique clusters) before the result set ended. maxStories is a hard run-wide cap.

Why is articleCount different from articlesReturned? articleCount is the number Yandex reports for the cluster, while articlesReturned is the number of article records actually collected. The two can differ in either direction: Yandex's live count changes as outlets publish or remove articles (observed 96 reported vs 122 returned, and 246 vs 362), and very large clusters may be capped by maxArticlesPerStory.

Are the date and sort filters applied by Yandex? Yandex's public search page applies them in the browser rather than in the server response. This actor applies the same filters and sorting to the set it returns, so a very old custom date range may return fewer results than a dedicated archive search.

Which categories are available? All, politics, society, SVO, business, world, sport, incidents, culture, technology, science, cars, army, show business, health, Moscow and interesting. all reads every feed.

Do I need a proxy or login? No. The actor connects directly and handles the site's anonymous session automatically. No cookies or credentials are required.

Can I search other regional editions or languages? The current public surface serves one Russian-language feed; the old ru/kz/by/uz/com edition selector no longer changes the content. Search queries can still be written in Russian, English or any language Yandex indexes.

Why is there no trending keyword list? The old “what people discuss” panel is no longer present on the homepage. topStories returns the homepage sections and their top clusters instead.

Does this actor download full article text? No. It returns article metadata and the publisher's URL. Full text stays on the publisher's site and may be paywalled or geo-restricted.

Is this affiliated with Yandex or Dzen? No. This is a third-party actor reading public pages; it is not affiliated with, endorsed by, or sponsored by Yandex or Dzen.