Lemmy Scraper

Two Lemmy servers answer the same question differently. Each one holds only the communities it has federated in, and each keeps its own blocklist: lemmy.world links more than eleven thousand instances and blocks a couple of hundred of them outright. So the host is the first thing you choose here, not an afterthought. Front-page feed, one community, or a keyword search, over the public v3 REST API. No account, no key.

You pick a host and you get what that host holds. lemmy.world is the default; lemmy.ml, beehaw.org and sh.itjust.works all work the same way.

What you get back

{
  "ok": true,
  "id": 41822190,
  "title": "Framework ships the Laptop 16 refresh",
  "url": "https://example.com/article",
  "body": "Spec sheet in the link. The keyboard is one piece now.",
  "author": "someuser",
  "authorActor": "https://lemmy.world/u/someuser",
  "community": "technology",
  "communityTitle": "Technology",
  "score": 812,
  "comments": 147,
  "upvotes": 871,
  "downvotes": 59,
  "nsfw": false,
  "thumbnail": "https://lemmy.world/pictrs/image/abc123.webp",
  "published": "2026-08-21T09:14:02.118Z",
  "postUrl": "https://lemmy.world/post/41822190"
}

Two different links per row, and the difference matters: url is the external page the post points at, which is null on a text-only post. postUrl is the Lemmy thread itself and is always there.

body is the post text. Lemmy stores it as markdown and it comes back as markdown. Stray HTML tags get cleaned out, but **bold** stays **bold**. authorActor is the federated user URL, which is what you want as a stable identity across instances.

Input

{
  "instance": "lemmy.world",
  "mode": "community",
  "query": "technology",
  "sort": "TopWeek",
  "maxItems": 200
}

Field	Default	Notes
instance	lemmy.world	Bare host, no https://.
mode	feed	feed (front page), community, or search.
query	empty	Community name in community mode, keywords in search mode. Ignored for feed. The Console prefills technology; an API call has to send its own.
sort	Hot	Hot, Active, New, TopDay, TopWeek, TopMonth, TopAll.
maxItems	100	1 to 1,000. Fetched 50 per request.
proxyConfiguration	off	The API is public. Only worth it if an instance rate-limits your IP.

Community names work either bare (technology) or fully qualified (technology@lemmy.world). Use the qualified form when you want a community that lives on a different server than the one you're querying.

Limits

• Comment text isn't returned. You get the count in comments, not the comments themselves.
• downvotes is 0 on instances that disable them. Several do, as a moderation choice. When that's the case score and upvotes will agree and downvotes sits at zero, which is a config fact rather than unanimous approval.
• What you can see depends on the instance. A host that has defederated from another won't show you its posts, and a small host has federated fewer of them.
• Search mode uses the instance's own search, so results and ranking are that server's.
• Posts are deduplicated on ID within a run.
• Bad input, no results, or a network failure each give one diagnostic row with an errorCode. Not billed.

Billing

$1.50 per 1,000 posts, so $0.0015 each, with no run-start fee. Only posts actually returned are charged, so an empty search costs nothing.

Delivery

Dataset export as JSON, CSV, Excel or XML. Set notionConnector and notionParentId to also write one Notion page per post when the run ends. Authorize the connector once under Settings → API & Integrations → MCP connectors. Leave them empty to skip.

How people use it

Thread watching. Pick a community, set sort to New, run it hourly and diff against the previous dataset. New threads land in your inbox instead of you sitting in the feed.

It's also decent for content research. TopMonth on a niche community shows you what actually got read that month, and score next to comments separates the posts people upvoted from the posts people argued about.

Keyword monitoring works, but be realistic about the size of it. Lemmy is small. You'll get genuine mentions and not many of them.