Medium Articles Scraper · Claps per Follower + Outlier Posts

Scrape Medium articles by tag, keyword, author or publication — and see which ones actually work. Every article row carries its claps, the number of distinct people who clapped, responses, claps per 1,000 author followers and how it compares with the author's own median. Authors and publications get a profile row with followers, posting frequency and median engagement. No login, no API key, no cookies.

Built for content strategists, ghostwriters, newsletter and Medium writers researching what to write next, marketers looking for writers to partner with, and anyone who needs Medium data with engagement instead of a bare RSS list.

Why this scraper

• Reach is not resonance. Big accounts get claps because they are big. clapsPer1kFollowers and vsAuthorMedian separate the articles that outperformed their author from the ones that just rode a large following. Measured on 24 Sep 2026: Software 2.0 by Andrej Karpathy has 61,025 claps = 33.9× his own median of 1,800.5; in the ai agents search, a post by a writer with 442 followers got 572 claps — 71.5× that writer's median of 8.
• Claps vs. clappers. One reader can clap up to 50 times. We return claps and clappers (distinct people) plus clapsPerClapper, so a handful of heavy clappers cannot pass as broad interest.
• Real engagement numbers, not RSS. RSS feeds carry titles and dates only — no claps, no responses, no reading time, no member-only flag. This Actor reads Medium's public web API, the same one medium.com uses for logged-out visitors.
• Member-only flag without the paywall. isMemberOnly shows which articles sit behind Medium's paywall; filter with memberOnly: "free" or "memberOnly". We return metadata and counters only, never paywalled text.
• Filters that save money. publishedWithinDays, minClaps, memberOnly. Filtered-out articles are never charged. Hard limits maxArticles / maxProfiles stop the run cleanly — you are never charged beyond them.
• Tag feeds by time window. TOP_WEEK, TOP_MONTH, TOP_YEAR, HOT or NEW — plus the tag's total article and follower count in the run summary.

Input example

{
    "tags": ["product-management"],
    "tagFeedMode": "TOP_MONTH",
    "searchTerms": ["ai agents"],
    "authors": ["@karpathy"],
    "publications": ["netflixtechblog.com"],
    "maxArticlesPerSource": 50,
    "publishedWithinDays": 90,
    "minClaps": 10
}

Output example (article row, measured 24 Sep 2026)

{
    "type": "article",
    "title": "Software 2.0",
    "url": "https://karpathy.medium.com/software-2-0-a64152b37c35",
    "authorName": "Andrej Karpathy",
    "authorFollowers": 61927,
    "publishedAt": "2017-11-11T22:18:53.751Z",
    "readingTimeMinutes": 8.8,
    "wordCount": 2146,
    "isMemberOnly": false,
    "tags": ["machine-learning", "artificial-intelligence", "programming", "software-development", "future"],
    "claps": 61025,
    "clappers": 8884,
    "clapsPerClapper": 6.9,
    "responses": 194,
    "clapsPer1kFollowers": 985.43,
    "authorMedianClaps": 1800.5,
    "vsAuthorMedian": 33.89,
    "source": "author:karpathy"
}

Field	Meaning
claps / clappers / responses	Exact counters as Medium reports them. clappers = distinct people who clapped.
clapsPer1kFollowers	Claps per 1,000 followers of the author. Articles in large publications can reach far beyond the author's followers — then this number gets very high; compare with publicationFollowers.
authorMedianClaps / vsAuthorMedian	Median claps over the author's latest 25 posts older than 48 hours, and this article's claps divided by it. null when the median is 0 (a ratio against 0 is not meaningful).
isMemberOnly	Medium's paywall flag.
isFresh	Published less than 48 hours ago — claps are still coming in.
readingTimeMinutes / wordCount	As computed by Medium.

Author / publication rows (type: "author" / "publication"): followers, lastPostDate, daysSinceLastPost, postsLast30Days, postsLast90Days (exact, or null with postsLast90DaysAtLeast when the window was not fully read), medianClaps, medianClappers, medianResponses, medianClapsPer1kFollowers, medianReadingTimeMinutes, memberOnlyShare, top post, top tags, X/Twitter link.

A run summary (SUMMARY in the key-value store) lists articles found per source, tag sizes, how many articles each filter dropped, failed or unknown inputs and HTTP status counts.

Honest limits

• Public data only. No login, no paywalled text. Member-only articles are returned as metadata and counters.
• Tag feeds end after about 100 articles per tag and mode (measured 24 Sep 2026 on artificial-intelligence: NEW 100, HOT 97, TOP_WEEK 99, TOP_MONTH 100). For more, use several modes, several tags, or search terms — search returns up to ~1,000 articles per term.
• TOP_WEEK / TOP_MONTH are Medium's own rankings; they can include articles older than the window that are clapped a lot right now.
• Publications that have left Medium (for example Towards Data Science, now on its own site) are listed as "not found" in the summary.
• Author followers and publication followers are Medium's public counts at scrape time.

Pricing (pay per event)

• Actor start: small flat fee per run
• Per scraped article: only articles that pass your filters and are written to the dataset
• Per scraped profile: one per requested author or publication

The author comparison (compareWithAuthor) costs extra requests on our side but is not charged separately.