Lemmy Scraper — Posts, Comments, Communities & Users (Fediverse / Reddit Alternative)

Scrape any Lemmy instance — posts, comments, communities, and users — in one run, with no account, no API key, and no proxy spend. Point it at lemmy.world (the largest instance) or any other Lemmy server and export clean, structured JSON/CSV/Excel ready for analysis, dashboards, or AI pipelines.

Lemmy is the open-source, federated Reddit alternative built on ActivityPub. lemmy.world alone hosts 13,400+ communities, 735,000+ posts, and 6.7 million+ comments (live count, July 2026) — a large, fully public, keyless corpus that most scrapers on the market don't touch.

What it does — four modes, one actor

• Feed — a site-wide post feed by sort (Active / Hot / New / TopDay / TopWeek …) and listing type (All = the whole federated feed the instance sees, or Local only)
• Community — pull posts from any named community (e.g. technology or selfhosted@lemmy.world), across many communities in a single run
• Comments — recent comments site-wide or scoped to specific communities, with a path field for thread reconstruction
• Search — find communities, posts, users, or comments by keyword (or All types at once)
• Unified output — posts, comments, communities and users land in one dataset with consistent, flat field names, ready for spreadsheets, databases, or RAG ingestion

Why use this actor

Most Lemmy tools	The problem	This actor
Single-community scrapers	One community per run — instance-wide coverage needs many runs and glue code	Site-wide feed + multi-community + comments + search, all in one run
Post-only scrapers	No comments, no community/user discovery	Posts + comments + communities + users, unified
Roll-your-own API scripts	You maintain cursor-vs-page pagination, view-object flattening, rate-limit retries, and charge logic	One input, one clean flat schema, pay only per result
Reddit-only scrapers	Miss the open Fediverse entirely	Native Lemmy / ActivityPub coverage across any instance

Built to never crash. Every network call retries transient failures (429 rate-limits, 5xx, timeouts) with exponential backoff and degrades gracefully on hard errors (a missing community is skipped, not fatal) — the run always finishes and delivers whatever it collected. No "under maintenance" dead ends.

Output fields

Posts (type: "post"): id, ap_id, title, body, external_url, thumbnail_url, embed_title, embed_description, published, updated, nsfw, locked, removed, deleted, featured_community, featured_local, language_id, score, upvotes, downvotes, comments_count, newest_comment_time, community_id, community_name, community_title, community_actor_id, creator_id, creator_name, creator_display_name, creator_actor_id, creator_instance_id

Comments (type: "comment"): id, ap_id, content, path, published, updated, distinguished, language_id, score, upvotes, downvotes, child_count, post_id, post_title, community_id, community_name, community_actor_id, creator_id, creator_name, creator_display_name, creator_actor_id

Communities (type: "community"): id, name, title, actor_id, description, published, nsfw, posting_restricted_to_mods, instance_id, icon, banner, subscribers, posts_count, comments_count, users_active_day, users_active_week, users_active_month, users_active_half_year

Users (type: "user"): id, name, display_name, actor_id, bio, published, banned, bot_account, instance_id, avatar, banner, matrix_user_id, post_count, comment_count

Every post, comment, community and user carries its ap_id / actor_id — the canonical ActivityPub URL — so you can join records back to the live Fediverse or across instances.

Usage examples

Site-wide hot feed:

{ "mode": "feed", "instance": "lemmy.world", "postSort": "Hot", "listingType": "All", "maxResultsPerTerm": 500 }

Posts from several communities (local or federated name@instance):

{
  "mode": "community",
  "instance": "lemmy.world",
  "communities": ["technology", "selfhosted@lemmy.world", "privacy"],
  "postSort": "TopWeek",
  "maxResultsPerTerm": 300
}

Recent comments in a community:

{ "mode": "comments", "communities": ["technology"], "commentSort": "New", "maxResultsPerTerm": 200 }

Discover communities and users about a topic:

{ "mode": "search", "searchTerms": ["selfhosted", "homelab"], "searchType": "All" }

Common use cases

• Social listening / brand monitoring — track a product, keyword, or community across the Fediverse, outside walled gardens
• Reddit-alternative monitoring — follow communities that migrated off Reddit to the open web
• Community analytics — rank Lemmy communities by subscribers and active users to find where a topic actually lives
• RAG & LLM training data — feed fresh, openly-licensed public discourse into retrieval-augmented and fine-tuning pipelines
• Academic & network research — collect public posts, comments, and community/user metadata at scale for NLP or social-graph analysis
• Sentiment & trend analysis — measure opinion on trending topics with upvote/downvote/score signals included

Pricing

Pay per result — $0.003 per item (post, comment, community, or user). You are billed only for rows actually delivered to your dataset.

• 1,000 items ≈ $3
• 5,000 items ≈ $15

No monthly fee, no proxy surcharge, no charge for empty runs.

Technical notes

• Uses each instance's official Lemmy REST API v3 (/post/list, /comment/list, /search) — free, public, keyless, no anti-bot bypass required. Verified live against lemmy.world (Lemmy 0.19.19).
• Zero proxy cost — official public API, no residential proxies needed.
• Handles both pagination models for you: /post/list uses opaque cursor pagination (next_page → page_cursor), while /comment/list and /search use classic 1-indexed page. You just set maxResultsPerTerm.
• Resilient by design: transient errors (429/5xx/network) are retried with backoff; hard errors (bad community, unsupported sort) are skipped, not fatal; the run always exits cleanly with partial results rather than failing.
• Polite pacing (~4 requests/sec) stays comfortably inside instance rate limits.
• Automatically stops when your spending limit is reached — results are delivered before they are charged, so no overrun.

Honest scope / limits

• Post ordering honors Lemmy's server-side sort; Top* windows are relative to run time (a snapshot, not a live stream).
• Comments mode returns comments as a flat list with a path field (materialized tree position 0.123.456) — reconstruct threads client-side from path when you need the tree.
• Search depth depends on what the chosen instance has federated and indexed; a niche term may return fewer rows on a small instance than on lemmy.world. Point the actor at the instance closest to your topic for best coverage.
• Output is a snapshot at run time; schedule recurring runs for ongoing monitoring.

Data compliance

This actor accesses only public Lemmy data through each instance's official, publicly documented REST API. No login credentials are used, and no private or authenticated content is touched. Public posts, comments, and communities on Lemmy are intentionally public. Use responsibly, respect each instance's terms of service, and honor the authors of the content you collect.