Lemmy Scraper: posts and comments from any Lemmy instance, with votes, authors and communities

Lemmy Scraper collects posts and comments from Lemmy, the open, federated alternative to Reddit, without an account, a login or an API key. Point it at any instance (lemmy.world by default, or lemmy.ml, sh.itjust.works, programming.dev and hundreds more), then pick communities, search terms, post links or simply the front page, in any sort Lemmy offers: Hot, Active, New, Top of the day, week, month or all time, Most comments and more. Each post comes back as one clean row with title, body, link, upvotes, downvotes, score, comment count, author, community and dates. Turn on comments to get each post's discussion as separate rows, with depth and parent comment.

It is used to build text datasets for AI training and RAG, to monitor what open-source, privacy and tech communities say about a product or a topic, to research trends across the fediverse, and to archive communities over time.

At a glance

What	Price
Post delivered	US$ 0.40 per 1,000 posts on the Free plan (US$ 0.0004 each)
Comment delivered (only with comments on)	US$ 0.40 per 1,000 comments
Paid Apify plans	Bronze US$ 0.38, Silver US$ 0.36, Gold and above US$ 0.34 per 1,000
Run start	US$ 0.005 per run

• 5 communities with 30 posts each = 150 posts = US$ 0.07 on the Free plan, run start included.
• The same 150 posts with 20 comments each = 3,150 items = US$ 1.27.
• Error rows and duplicates are never charged. maxItems is a hard cap, and a spending limit set on the run stops it cleanly with everything delivered so far.

Quick start

Hot posts of one community:

{
  "communities": ["technology"],
  "maxPostsPerSource": 30
}

Top posts of the week from communities on two instances, with the 20 best comments of each post:

{
  "communities": ["linux@lemmy.ml", "https://lemmy.world/c/selfhosted"],
  "sort": "TopWeek",
  "includeComments": true,
  "maxCommentsPerPost": 20
}

Search posts on another instance:

{
  "instance": "programming.dev",
  "searchQueries": ["rust async", "python packaging"],
  "sort": "TopMonth",
  "maxPostsPerSource": 50
}

Leave communities, searches and links empty to get the instance front page in the chosen sort.

Input

Field	What it does
communities	Community names on the chosen instance (technology), federated names (linux@lemmy.ml) or community links.
searchQueries	Keywords to search in posts. With exactly one community, the search runs inside it.
startUrls	Post (/post/123), comment (/comment/456) or community (/c/name) links from any instance.
instance	The Lemmy server to read from (default lemmy.world).
sort	Hot (default), Active, New, Old, Scaled, Controversial, MostComments, NewComments, TopHour to TopAll.
listingType	For front page and search: All (federated, default) or Local (this instance only).
maxPostsPerSource	Posts per community, search term or front page (default 30; 0 = as many as the instance returns).
maxItems	Cap for the whole run, posts and comments together (0 = no cap).
skipNsfw	Leave out posts marked NSFW (default off).
includeComments	Add each post's comments as rows (default off).
maxCommentsPerPost	Comments per post when comments are on (default 20; 0 = all).
commentSort	Top (default), Hot, New, Old or Controversial.
proxyConfiguration	Apify Proxy by default.

community, query, keyword, searchQuery, urls, sortBy, maxResults and limit are accepted too, so inputs written for other Lemmy tools keep working.

Output

Every row has type (post, comment or error).

Post row (type: post):

Field	What it is
id, postUrl	Post id and link on the instance you read from
originalUrl	Link on the instance where the post was written (federation)
title, body	Post title and text (Markdown)
link, linkTitle, linkDescription, thumbnailUrl	Shared link and its preview
upvotes, downvotes, score, comments	Vote counts and number of comments
publishedAt, updatedAt, newestCommentAt	Dates in UTC
nsfw, locked, featured, isLocal	Post flags; isLocal is true when the post was written on this instance
author, authorDisplayName, authorHandle, authorUrl, authorIsBot	Author's public username, display name, name@instance handle, profile link and bot flag
community, communityTitle, communityHandle, communityUrl	Community name, title, name@instance handle and link
instance, position, sourceUrl, scrapedAt	Where it was read, order in the source, what you sent and when

Comment row (type: comment): id, commentUrl, originalUrl, text, upvotes, downvotes, score, replies, depth (1 = direct reply to the post), parentId (the comment it answers), publishedAt, updatedAt, distinguished, postId, postTitle, postUrl, plus the same author and community fields.

One real post row:

{
  "type": "post",
  "id": 53214894,
  "postUrl": "https://lemmy.world/post/53214894",
  "title": "Linus Torvalds shares his thoughts on Anti-AI coding rules",
  "link": "https://media.piefed.social/posts/TS/lY/TSlYv9DvdaiTI2U.png",
  "upvotes": 135,
  "downvotes": 14,
  "score": 121,
  "comments": 83,
  "publishedAt": "2026-10-16T17:36:33.541124Z",
  "author": "Ek-Hou-Van-Braai",
  "authorHandle": "Ek-Hou-Van-Braai@piefed.social",
  "community": "linux",
  "communityHandle": "linux@lemmy.ml",
  "instance": "lemmy.world"
}

And one of its comments:

{
  "type": "comment",
  "id": 26368467,
  "text": "I don't think he thinks about you at all",
  "upvotes": 32,
  "downvotes": 3,
  "score": 29,
  "replies": 1,
  "depth": 2,
  "parentId": 26367731,
  "postId": 53214894,
  "community": "linux"
}

How it works

Every Lemmy server publishes the same open API. The Actor reads it and turns the answers into flat rows, so runs are fast: 30 posts usually take one or two seconds. Communities hosted on other servers are read through the instance you choose, exactly as its web page shows them; links keep their own instance.

Errors you may see

Every error is a row with errorCode and is never charged:

errorCode	Meaning
INVALID_URL	Not a Lemmy community, post or comment link
INVALID_INPUT	A field has a wrong value (for example an unknown sort)
NOT_FOUND	The community, post or comment does not exist on that instance
NO_RESULTS	The community or search returned no posts for this sort
NOT_LEMMY	The server answered, but it is not a Lemmy instance
BLOCKED	The instance kept refusing requests (rate limit); run it again in a few minutes
UNREACHABLE	The instance did not answer
NOT_REACHED	The run timeout or the cost guard arrived before this source

Integrations, scheduling and AI agents

• Schedule a run every hour or day in Apify schedules with sort New to collect every new post of your communities.
• Export as JSON, CSV, Excel or XML, or read the dataset from the Apify API.
• Connect it to Make, n8n, Zapier, Slack or Google Sheets through Apify integrations, for example to post brand mentions to a channel.
• AI agents: call it as a tool through the Apify MCP server in Claude, ChatGPT, Cursor or any MCP client, for example "get this week's top posts of linux@lemmy.ml with their comments and summarize the main complaints".

FAQ

Do I need a Lemmy account? No. The Lemmy API is public for reading; no login, cookie or key is used.

Which instance should I use? The default, lemmy.world, is the largest and sees most federated communities. Use another instance when the community you want lives there or is not federated with lemmy.world.

How do I get communities from other servers? Write them as name@server (for example linux@lemmy.ml) or paste their link. Counts can differ slightly between instances, because each server counts the votes it received.

Can I get every comment of a post? Yes: turn on includeComments and set maxCommentsPerPost to 0.

Is it legal? The Actor reads public data only and returns public usernames, never emails or phone numbers. You are responsible for how you use the data under the laws that apply to you.

Something broke or a field is missing? Open an issue on the Actor page; issues are answered within a day.

Limitations

• Lemmy search matches words loosely; very generic terms bring unrelated posts.
• Some instances limit how deep you can page into old posts or rate-limit busy periods.
• Deleted and removed posts and comments are skipped.
• User profiles and private messages are not collected.

Changelog

• 16 Oct 2026: first version: communities, search, post, comment and community links, front page, all Lemmy sorts, comments with depth and parent, any instance.