Skip to main content
Scrape a Medium link and the result comes back with the page’s data in a defined structure: a story with its text, a writer’s stories, a publication, a tag’s stories, search results, a list or a story’s responses. Send "output": "json" to get just that data, or add a prompt or schema to shape it your own way (how results come back). Each link costs 1 credit, and a link that can’t be read costs nothing. This works on medium.com, on writers’ and publications’ name.medium.com addresses, and on publications’ own domains, such as netflixtechblog.com.

Example

The item’s result, shortened to one story:

More rows

Writers, publications, tags, search, lists and responses read 25 rows by default. Add count to the link for more, up to 1,000, or count=all:
The first 100 rows cost the link’s 1 credit, and every further 100 is 1 more. A list that runs out sooner costs less. On a story link, ?responses=100 also returns that many responses, and ?clappers=100 that many of the people who clapped for it (clapperList), billed the same way.

Member-only stories

Medium shows visitors who aren’t signed in only the opening of a member-only story. You still get every detail of the story (claps, responses, word count, tags and the rest), with isLocked: true and isPreviewOnly: true, and the text is that opening. If the writer shared a Friend Link (it has ?sk= in it), scrape that link as it is and you get the whole story, with friendLinkUsed: true. A Friend Link that no longer works gives the preview, with a note that says so.

Fields

Story in a list: id, title, subtitle, url, canonicalUrl, slug, author (id, name, username, url, bio, imageUrl, followers, following, membershipTier, isMember, isBookAuthor, twitter, customDomain), publication (id, name, slug, url, domain, description, followers), publishedAt, updatedAt, readingTimeMinutes, wordCount, claps, clappers, responsesCount, tags, topics, isLocked, visibility, language, isSeries, isShortform, isPinned and previewImageUrl. Tag and search rows also have rank. Story: the same, plus markdown, text, paragraphs (each with its type, text, image, code language or embed), imageUrls, links, embeds, codeBlockCount, paragraphCount, topHighlight, highlights, isPreviewOnly and friendLinkUsed, plus responses and clapperList when you ask for them. Writer: id, username, name, bio, imageUrl, url, followers, following, publicationsFollowing, membershipTier (MEMBER or FRIEND), memberSince, friendSince, isBookAuthor, twitter, customDomain, hasSubdomain, tippingLink, newsletter and posts. Publication: id, name, slug, url, domain, tagline, description, followers, tags, twitter, logoUrl, creator, editors, writers, newsletter and posts. Tag: slug, name, url, followers, postCount, relatedTags and posts (or writers on /top-writers, each a writer or a publication). Response: id, url, text, author, publishedAt, updatedAt, claps, clappers, repliesCount, depth (0 for a response, 1 for a reply), parentId and rootPostId. List: id, name, description, url, creator, storyCount, claps, clappers, responsesCount, createdAt, lastItemAddedAt and posts.

Limits

Only what Medium shows to visitors who aren’t signed in is available. Pages for signed-in members, such as your stats, reading history and the For you feed, fail straight away and cost nothing. So do deleted stories, unknown writers, tags and lists, private lists, and searches with no results. Medium’s most read lists for a tag are short, about 15 to 25 stories. The newest stories go back through the tag’s whole archive.

Templates

Ready-made templates for Medium: a writer’s stories, publication stories, stories by tag, story search, top writers in a topic, writer profiles, responses and people search.
Check Medium’s terms before you collect data from it. Stories belong to their writers. You’re responsible for how you use what you collect.