Surface health.com's freshest articles and the topics it is editorially betting on right now — the pool of likely-viral content — by reading the Google News sitemap and the homepage 'Trending' and 'The Latest' modules, then ranking by recency, topic-cluster frequency, and viral-headline pattern. Read-only. This skill was captured from a live agent session on health.com and publishes here verbatim, exactly as an agent receives it.

Purpose

Surface the articles and topics on health.com (a Dotdash Meredith property) that are most likely to go viral right now, and return them as a ranked list. health.com does not expose a public popularity/analytics API, so there is no source of true view counts. Instead, this skill infers "viral potential" from three honest, cheap signals the site publishes: (1) the rolling Google News sitemap of just-published articles, (2) the homepage trending-list module — the editorial team's hand-curated trending topic/campaign, and (3) topic-cluster frequency across the freshest headlines (a topic the publisher pushes multiple articles about in a short window is a topic it is betting will trend). The skill is read-only — it never logs in, comments, or submits anything.

When to Use

"What's about to blow up on health.com?" / "What health topics are trending today?"
A content/SEO analyst monitoring a competitor's editorial priorities (which topics they're flooding).
A newsletter or social agent that needs the freshest health articles plus the dominant theme to write around.
Daily/hourly polling for newly published articles (the News sitemap refreshes with Cache-Control: max-age=5).
Anywhere you'd otherwise scrape the rendered homepage — the sitemaps give the same fresh feed faster and without the Cloudflare wall.

Workflow

The optimal path is a lightweight navigation to the sitemaps and homepage HTML — no heavy scripted browsing, no login. health.com's rendered pages sit behind Cloudflare (homepage returns 403/402 to a bare client), but the XML sitemaps are served fine even without a proxy, and the homepage HTML can be pulled through a residential proxy (200). Drive it with a single browserless_agent call (goto + parse in-page), not a multi-step browser flow, unless a rendered read is blocked. Note: browserless_function/raw fetch has no network egress until the page navigates, so a goto-first browserless_agent is the clean path for these static XML/HTML resources.

1. Pull the freshest article pool (no proxy needed)

Navigate to the sitemap and read its raw XML body — no proxy required:

{
  "url": "https://www.health.com/google-news-sitemap.xml",
  "commands": [
    {
      "method": "goto",
      "params": {
        "url": "https://www.health.com/google-news-sitemap.xml",
        "waitUntil": "load",
        "timeout": 45000
      }
    },
    { "method": "text", "params": { "selector": "body" } }
  ]
}

(No proxy arg — the XML sitemaps are not Cloudflare-walled.)

This Google News sitemap is the canonical "what we just published / are pushing into Google News" feed — a rolling window of the most recent articles (observed ~14 entries; News sitemaps cap at the last ~48 h / ≤1000 URLs). Each <url> gives everything you need:

<loc> — canonical article URL (the trailing numeric id, e.g. …-11961430, is the CMS post id).
<news:publication_date> — ISO 8601 with TZ offset (e.g. 2026-05-30T11:00:00-04:00). Sort descending → the newest entries are the live viral window.
<news:title> — headline (HTML-entity-encoded; decode &#39; → ').
<image:image><image:loc> — lead image URL.

This feed matches the homepage "The Latest" module ordering exactly (verified), so it is authoritative — you usually do not need to render the homepage to get the fresh feed.

Navigate the homepage through a residential proxy and parse the trending module in-page (return a compact JSON projection, not the raw HTML):

{
  "url": "https://www.health.com/",
  "proxy": { "proxy": "residential" },
  "commands": [
    {
      "method": "goto",
      "params": {
        "url": "https://www.health.com/",
        "waitUntil": "load",
        "timeout": 45000
      }
    },
    {
      "method": "evaluate",
      "params": {
        "content": "(()=>{ const mod = document.querySelector('[id^=\"trending-list\"], .trending-list'); if(!mod) return JSON.stringify(null); const title = mod.querySelector('[class*=\"trending-list__title\"]')?.textContent?.trim() || null; const articles = [...mod.querySelectorAll('a[href*=\"health.com\"]')].map(a=>({ url: a.href, title: a.querySelector('[class*=\"card__title-text\"]')?.textContent?.trim() || a.textContent.trim() })); return JSON.stringify({ title, articles }); })()"
      }
    }
  ]
}

The returned .value (JSON string) comes from id="trending-list_1-0" (CSS classes trending-list mntl-four-post). That module contains:

class="trending-list__title …" → the editorially-declared trending theme/campaign (observed: "May Is Skin Cancer Awareness Month"). This is the single best "the publisher thinks this is the moment" signal.
4 article cards — extract each <a href="https://www.health.com/…"> plus the card__title-text headline.

(The same HTML also carries the >The Latest< feed — same articles as the News sitemap — and explore-conditions/section modules if you want category context.)

3. Rank by viral-potential heuristic

Combine the signals into a score per article/topic. None of these is a true view count — be explicit that this is potential, not measured virality:

Recency — newest publication_date first. The top few entries are inside the active sharing window.
Topic-cluster frequency — tokenize the fresh headlines and count repeated themes. When the publisher ships ≥2 articles on the same topic in a short window, that topic is a deliberate trend bet. (In the validation run: "magnesium" appeared in 3 of 14 headlines, "blood sugar" in 3, "probiotics" in 2 — so "magnesium / blood sugar" and "probiotics" were the dominant trending topics, independent of the curated skin-cancer module.)
Curated-module boost — articles matching the trending-list__title theme are editorially endorsed.
Headline-pattern score — health.com leans on classic viral headline shapes; boost for: numbered listicles ("7 Morning Drinks…", "8 Foods…"), curiosity-gap framing ("What Happens to Your Liver When You…"), and head-to-head comparisons ("Sourdough vs. White Bread"). These templates correlate with shareability.

4. Emit the ranked result (see Expected Output).

Browser fallback

Only if the proxied homepage navigation in step 2 is unavailable and you must read the rendered modules — do the whole flow inside ONE browserless_agent call (goto → scroll the module into view → grab the HTML), with a residential proxy:

{
  "url": "https://www.health.com/",
  "proxy": { "proxy": "residential" },
  "commands": [
    {
      "method": "goto",
      "params": {
        "url": "https://www.health.com/",
        "waitUntil": "load",
        "timeout": 45000
      }
    },
    {
      "method": "evaluate",
      "params": {
        "content": "(()=>{ document.querySelector('[id^=\"trending-list\"], .trending-list')?.scrollIntoView({block:'center'}); return 'ok'; })()"
      }
    },
    { "method": "html", "params": { "selector": "body" } }
  ]
}

Then parse trending-list_1-0 out of the returned HTML as in step 2. No session-release step (nothing to release). The session persists across separate calls, keyed by proxy/profile, so a later call with the same config reconnects to the same page with cookies/session intact; batching the whole goto → scroll → extract flow inside this one call's commands array saves round-trips and avoids accidentally dropping that config. A bare (no-proxy) browser session gets the Cloudflare challenge, so proxy: { proxy: "residential" } is mandatory for any rendered page. The browser path costs far more than the sitemap/homepage navigation and yields the same data; use it only as a true fallback.

Site-Specific Gotchas

Sitemaps bypass the Cloudflare wall; rendered pages do not. https://www.health.com/ returns 403 (bare fetch) / 402 (probe) / Cloudflare challenge, but /google-news-sitemap.xml, /sitemap.xml, and /sitemap_1.xml all return 200 without a proxy. The homepage HTML needs a residential proxy (403 without → 200 with proxy: { proxy: "residential" }). Lead with the sitemap; only proxy when you need the curated module.
No virality metrics exist. There is no public endpoint with view/share counts. Anyone claiming a "true viral score" is fabricating it. This skill returns potential derived from recency + topic clustering + headline patterns. State that in the output (signal: "inferred").
robots.txt blocks most AI crawlers by name (ClaudeBot, GPTBot, anthropic-ai, CCBot, PerplexityBot, Google-Extended, etc. → Disallow: /). The News sitemap (User-agent: * only disallows *.pdf and /embed?) is explicitly published for news discovery, so reading it is consistent with the site's intent. Respect the AI-crawler rules for bulk crawling; keep this read-only and low-volume (the feed is small).
News sitemap is a tiny rolling window (~14 URLs observed, max-age=5) — it is the fresh feed, not an archive. For "recently updated but older" content, fall back to sitemap_1.xml (≈5,962 URLs, each with <lastmod> but no title/news metadata — you'd have to fetch each article for a headline).
sitemap.xml is an index, not a flat list — it points to a single child sitemap_1.xml. Don't expect article URLs directly in sitemap.xml.
HTML-entity double-encoding in titles. Observed Here&#39;s in <news:title> — that's Here's double-encoded. Decode & → & first, then numeric entities.
The trending-list module is a curated campaign, not real-time popularity. Its title ("May Is Skin Cancer Awareness Month") is an editorial/seasonal theme, often unrelated to the dominant cluster in the fresh feed (which was magnesium/blood-sugar nutrition). Treat the two as complementary signals, not the same thing.
Topic clustering is the strongest organic trend signal. Because the curated module is seasonal, the more reliable "what's trending in their newsroom today" read is keyword frequency across the News-sitemap headlines (multiple magnesium / blood-sugar / probiotics pieces shipped within hours of each other).
Cookies/session not required for fetch. The sitemaps and proxied homepage set __cf_bm/TMog cookies but don't gate the content on them for a single read.
DOM ids are suffixed _1-0 (trending-list_1-0, mntl-four-post__inner_1-0). The Dotdash Meredith CMS appends _<n>-<m> instance suffixes; match on the id prefix / class (trending-list, mntl-four-post) rather than the exact id in case the suffix changes.

Expected Output

{
  "site": "health.com",
  "retrieved_at": "2026-05-31T15:21:00Z",
  "signal": "inferred",
  "signal_basis": [
    "news_sitemap_recency",
    "topic_cluster_frequency",
    "curated_trending_module",
    "viral_headline_pattern"
  ],
  "curated_trending_topic": {
    "title": "May Is Skin Cancer Awareness Month",
    "source": "homepage trending-list module",
    "articles": [
      {
        "title": "What Does Skin Cancer Look Like?",
        "url": "https://www.health.com/what-does-skin-cancer-look-like-8684936"
      },
      {
        "title": "How Fast Does Skin Cancer Grow? Factors and How To Monitor",
        "url": "https://www.health.com/how-fast-does-skin-cancer-grow-11700721"
      }
    ]
  },
  "trending_topics_inferred": [
    {
      "topic": "magnesium",
      "article_count": 3,
      "rationale": "3 of 14 fresh headlines reference magnesium (sleep + blood sugar)"
    },
    {
      "topic": "blood sugar",
      "article_count": 3,
      "rationale": "repeated blood-sugar nutrition angle across the fresh feed"
    },
    {
      "topic": "probiotics",
      "article_count": 2,
      "rationale": "kombucha + dairy probiotic comparisons shipped together"
    }
  ],
  "viral_candidates": [
    {
      "rank": 1,
      "title": "7 Morning Drinks To Support Your Kidney Health, According to a Dietitian",
      "url": "https://www.health.com/morning-drinks-for-kidney-health-11961430",
      "post_id": "11961430",
      "published_at": "2026-05-30T11:00:00-04:00",
      "image": "https://www.health.com/thmb/.../Health-GettyImages-114452963-....png",
      "headline_patterns": ["numbered_listicle", "expert_authority"],
      "topic_cluster": "kidney/nutrition",
      "viral_potential": "high"
    },
    {
      "rank": 2,
      "title": "What Happens to Your Liver When You Eat High-Sugar Foods Every Day?",
      "url": "https://www.health.com/high-sugar-foods-liver-health-effects-11964726",
      "post_id": "11964726",
      "published_at": "2026-05-30T10:00:00-04:00",
      "headline_patterns": ["curiosity_gap"],
      "topic_cluster": "blood sugar",
      "viral_potential": "high"
    }
  ],
  "notes": "Virality is inferred (no public view/share metrics exist on health.com). Ranking = recency + topic-cluster frequency + viral headline pattern + curated-module overlap."
}

If the curated module is absent or the homepage fetch is blocked, omit curated_trending_topic and return the News-sitemap-only ranking:

{
  "site": "health.com",
  "signal": "inferred",
  "curated_trending_topic": null,
  "trending_topics_inferred": [{ "topic": "magnesium", "article_count": 3 }],
  "viral_candidates": [
    {
      "rank": 1,
      "title": "...",
      "url": "...",
      "published_at": "...",
      "viral_potential": "high"
    }
  ],
  "notes": "Homepage trending module unavailable (Cloudflare); ranked on News-sitemap recency + topic clustering only."
}

Health.com Find Viral Articles & Trending Topics

Purpose

When to Use

Workflow

1. Pull the freshest article pool (no proxy needed)

3. Rank by viral-potential heuristic

4. Emit the ranked result (see Expected Output).

Browser fallback

Site-Specific Gotchas

Expected Output

Call it

Cookie Preferences

Health.com Find Viral Articles & Trending Topics

Purpose

When to Use

Workflow

1. Pull the freshest article pool (no proxy needed)

2. Pull the curated trending topic from the homepage (proxy required)

3. Rank by viral-potential heuristic

4. Emit the ranked result (see Expected Output).

Browser fallback

Site-Specific Gotchas

Expected Output

Call it

Cookie Preferences