Bluesky Scraper — Posts, Profiles, Feeds & Threads, No Login

Scrape public Bluesky posts, profiles, custom feeds, lists and reply threads with no login, and get clean JSON rows with text, engagement counts, links, mentions, hashtags, image alt text, quoted posts and author details.

Use it for social listening, research datasets, brand monitoring or AI pipelines: the smallest run is one search query, such as apify, which returns up to 30 recent posts in a few seconds for about $0.03 (100 posts cost about $0.09 on the Free plan, less on paid plans).

Table of contents

• What it does
• Quick start
• Use cases
• Input
• Output
• Pricing and limits
• Switching from another Bluesky scraper
• API and integrations
• Troubleshooting
• FAQ
• Learn more
• Support

What it does

Bluesky Scraper reads the public Bluesky AppView API (AT Protocol) without an account, a password or cookies. It has five modes:

Mode	You give	You get
searchPosts	Keywords, #hashtags, phrases, operators like from:bsky.app, lang:en, domain:example.com	Up to 100 matching posts per query, latest or top
authorFeed	Handles, DIDs or profile links	Posts from each account, newest first, with optional replies and reposts
profiles	Handles, DIDs or profile links	One profile row per account: bio, follower/following/post counts, avatar, pinned post
thread	Post links or at:// URIs	The post, its parent posts and the replies below it, with threadRole and threadDepth
feed	Custom feed links or list links	Posts from the feed or list

An account, post or feed that does not exist returns a not_found row instead of silently disappearing, so the number of answers always matches your input.

Good fit

• Track what people post about a brand, product, hashtag or domain on Bluesky.
• Collect the posts and profiles of a list of accounts for research or lead lists.
• Pull a full reply thread for sentiment or discussion analysis.
• Feed clean, typed Bluesky rows into an AI agent, a spreadsheet or a database.

Not a fit / limits

• No private data: no DMs, no muted or blocked lists, nothing that needs login.
• No followers or following lists in this version.
• Search returns at most 100 posts per query, because Bluesky blocks further search pages without login. The Actor does not try to get around this.
• Counts (likes, reposts, followers) are a snapshot at fetch time.

Quick start

1. Open the Input tab.
2. Keep What to scrape on Search posts and add a word to Search queries, e.g. apify.
3. Keep Max posts per query at the prefilled 30 (the runtime default is 100).
4. Click Start.
5. Open the Posts view in the Output tab and check text, author.handle and likeCount.

Expected result: up to 30 of the newest posts that match the query, one row per post.

Use cases

Buyer job	Input path	Output entity	Key fields	Next step
Brand or keyword listening	mode: searchPosts, queries	post	text, createdAt, likeCount, author.handle, links	Schedule the Actor, export to Sheets or a database
Account research	mode: authorFeed, handles, since	post	text, isRepost, replyCount, externalLink	Compare accounts, build datasets
Influencer or lead list	mode: profiles, handles	profile	displayName, description, followersCount, postsCount	CRM import
Discussion analysis	mode: thread, postUrls	post	threadRole, threadDepth, text, replyTo	Sentiment or topic analysis
Curated topic monitoring	mode: feed, feedUrls	post	text, hashtags, images[].alt	AI summarisation

Input

Input fields

Field	Plain name	Type	Required	Default / prefill	Allowed values	Effect	Example
mode	What to scrape	string	yes	searchPosts	searchPosts, authorFeed, profiles, thread, feed	Picks the list that is read	"authorFeed"
queries	Search queries	string[]	for searchPosts	prefill ["apify"]	up to 100	One search per line	["#bluesky", "from:bsky.app lang:en"]
handles	Handles	string[]	for authorFeed, profiles	none	handle, @handle, DID, profile link; up to 1,000 (100 for authorFeed)	Accounts to read; a bare name becomes name.bsky.social	["bsky.app", "did:plc:z72i7hdynmk6r22z27h6tvur"]
postUrls	Post links	string[]	for thread	none	bsky.app post links or at:// URIs; up to 100	Threads to read	["https://bsky.app/profile/bsky.app/post/3l6oveex3ii2l"]
feedUrls	Feed or list links	string[]	for feed	none	…/feed/… or …/lists/… links; up to 50	Feeds or lists to read	["https://bsky.app/profile/bsky.app/feed/whats-hot"]
maxItems	Max posts per input	integer	no	100, prefill 30	1–5,000	Cap for each query, account, thread or feed; ignored in profiles	500
since	From date	string	no	none	YYYY-MM-DD or ISO date-time, UTC	Only posts created on or after it	"2026-10-01"
until	To date	string	no	none	same; must be after since	Only posts created before it	"2026-10-08"
searchSort	Search order	string	no	latest	latest, top	Order of search results	"top"
language	Post language	string	no	all	language code, e.g. en, pt-BR	Search only	"en"
includeReplies	Include replies	boolean	no	true		Drop replies in search and account posts when off	false
includeReposts	Include reposts	boolean	no	false		Account posts and feeds: also return reposts, marked isRepost	true
threadDepth	Reply depth	integer	no	10	0–100	Reply levels below the post; 0 = post and parents only	3

Lines pasted with several values are split, and duplicates are removed. A value that cannot be read (e.g. not a handle!) returns an invalid_input row, and the other values are still processed. A wrong field combination, such as profiles with no handles or since after until, stops the run at once with a clear message and no rows.

Interacting settings

When	Input A	Input B	Effective behavior
Search with a lone hashtag and dates	queries: ["#ai"]	since / until	Sent to Bluesky as a tag search, which is the only form it accepts with dates
Account posts with dates	handles	since	Reading stops once the feed reaches posts older than since
Account posts with strict filters	includeReplies: false, since / until	—	If fewer than 1 in 100 feed items match after 1,000 items, the account is stopped as partial with a warning. Use a search with from:handle instead
Any mode	maxItems	search	Search never returns more than 100 posts per query, whatever maxItems says

Input examples

Search posts about a topic

{
  "mode": "searchPosts",
  "queries": ["apify", "#webscraping"],
  "maxItems": 50,
  "searchSort": "latest",
  "language": "en"
}

Expected output: up to 50 post rows per query with text, author, likeCount, links, hashtags.

Recent posts from accounts

{
  "mode": "authorFeed",
  "handles": ["bsky.app", "@jay.bsky.team"],
  "since": "2026-10-01",
  "includeReposts": true,
  "maxItems": 200
}

Profiles

{
  "mode": "profiles",
  "handles": ["bsky.app", "jay.bsky.team", "https://bsky.app/profile/pfrazee.com"]
}

A reply thread

{
  "mode": "thread",
  "postUrls": ["https://bsky.app/profile/bsky.app/post/3l6oveex3ii2l"],
  "threadDepth": 3,
  "maxItems": 200
}

Output

Every row goes to the default Dataset and has recordType: post, profile or status. The Output tab has Posts, Profiles and Not found & errors views. Views choose columns, not rows: in a posts run the Profiles view shows the same rows with empty profile columns, and status rows appear in the Posts view with empty post columns. To get only one type, filter by recordType in your export or code. The run summary is in the key-value store as RUN_SUMMARY.

Post fields

Field	Meaning	Type	Can be empty?
uri, cid	AT Protocol identifiers of the post	string	cid can be null
url	bsky.app link to the post	string	rarely null
author	did, handle, displayName, avatar, url	object	name and avatar can be null
text	Post text	string	empty for image-only posts
createdAt, indexedAt	Created (by the author's record) and indexed by Bluesky, ISO 8601	string	can be null
langs	Declared languages	string[]	often empty
replyCount, repostCount, likeCount, quoteCount, bookmarkCount	Counts at fetch time	number	null when Bluesky does not report it (never a fake 0)
isReply, replyTo, rootPost	Reply status, parent and root post	boolean, object	objects are null for top-level posts
isRepost, repostedBy, repostedAt	Repost marker in account posts and feeds	boolean, object, string	null when not a repost
isPinned	Pinned on the author's profile	boolean
isQuote, quotedPost	Quoted post: uri, url, authorHandle, text	boolean, object	null when not a quote
embedType	images, video, external, record, recordWithMedia	string	null without an embed
images	url, thumbnailUrl, alt, width, height	object[]	empty without images
video	playlistUrl, thumbnailUrl, alt	object	null without a video
externalLink	Link card: url, title, description	object	null without a card
links	All links in the text and the link card	string[]
hashtags, mentions	Tags and mentioned accounts (did, handle)	array
labels	Moderation labels	string[]
threadRole, threadDepth	anchor, parent or reply, and level (thread mode)	string, number	null outside thread mode
mode, input, fetchedAt, source	Which input produced the row and when	string

Profile fields

did, handle, displayName, description, url, avatar, banner, followersCount, followsCount, postsCount, listsCount, feedsCount, starterPacksCount, isLabeler, verifiedStatus, pinnedPostUrl, labels, createdAt, indexedAt, plus mode, input, fetchedAt, source.

Status rows

status	When	errorCode examples
not_found	The account, post, feed or list does not exist	NOT_FOUND
invalid_input	The value cannot be read as a handle, link or query	INVALID_INPUT
error	Bluesky failed or refused the request after retries	RATE_LIMITED, SOURCE_FORBIDDEN, SOURCE_UNAVAILABLE

A source failure is never shown as "no results": it is an error row and the run status says so.

Run summary (RUN_SUMMARY)

Field	Meaning
status	COMPLETE, PARTIAL, NO_RESULTS, CAPPED, ERROR, INVALID_INPUT or CANCELLED
errorCode	Set when the whole run failed for one reason, e.g. INVALID_INPUT or RUN_TIMEOUT_TOO_SHORT; otherwise null
message	One plain-English sentence about the result
counts	posts, profiles, notFound, errors, invalidInputs, duplicatesSkipped, chargedPosts, chargedProfiles, chargedEmptyResults, unprocessedInputs
inputs	One entry per input value (input, status, items, message): complete, partial, no_results, not_found, error, invalid_input or skipped, with the row count
warnings	Plain-English notes, e.g. the 100-per-query search limit

Output example

A fictional post, trimmed to the main fields:

{
  "recordType": "post",
  "status": "complete",
  "mode": "searchPosts",
  "input": "opendata",
  "uri": "at://did:plc:aaaaaaaaaaaaaaaaaaaaaaaa/app.bsky.feed.post/3mexample2abc",
  "url": "https://bsky.app/profile/author.example/post/3mexample2abc",
  "author": {
    "did": "did:plc:aaaaaaaaaaaaaaaaaaaaaaaa",
    "handle": "author.example",
    "displayName": "Sample Author",
    "url": "https://bsky.app/profile/author.example"
  },
  "text": "Our open dataset is live! Thanks @team.example #opendata https://data.example/downloads",
  "createdAt": "2026-10-12T09:15:00.000Z",
  "likeCount": 2431,
  "repostCount": 318,
  "replyCount": 154,
  "isReply": false,
  "isRepost": false,
  "embedType": "images",
  "images": [
    { "url": "https://cdn.bsky.app/img/feed_fullsize/plain/did:plc:aaaaaaaaaaaaaaaaaaaaaaaa/bafkexample@jpeg", "alt": "Chart of monthly downloads" }
  ],
  "links": ["https://data.example/downloads"],
  "hashtags": ["opendata"],
  "mentions": [{ "did": "did:plc:bbbbbbbbbbbbbbbbbbbbbbbb", "handle": "team.example" }],
  "quotedPost": null,
  "threadRole": null,
  "fetchedAt": "2026-10-13T01:52:14.278Z",
  "source": "Bluesky AppView API (AT Protocol)"
}

null means Bluesky did not provide the value; it never stands for zero.

Pricing and limits

Pricing unit

You pay per event. Platform usage (compute, proxy, storage) is included in the price.

Event	Free plan	Starter (Bronze)	Scale (Silver)	Business (Gold) and higher
Post row	$0.0009	$0.00085	$0.0007	$0.00055
Profile row	$0.0027	$0.0026	$0.0021	$0.0016
Empty result	$0.0005	$0.0005	$0.0005	$0.0005
Actor start	$0.001	$0.001	$0.001	$0.001

• Post row / profile row: one per data row written.
• Empty result: one per input value that produced no data: a search query with no matches, an account with no matching posts, a profile, post or feed that does not exist, or a value that cannot be read (for example a typo in a link). If every value is unreadable, the run ends as NO_RESULTS with a message saying so, and each value is one empty result. It covers the requests made for that value.
• Actor start: charged by Apify once per run (per GB of memory; the default 256 MB counts as one).
• Free: error rows (Bluesky failed), skipped inputs, duplicate posts and duplicate input values (the same account, post or feed written another way, e.g. @Name.bsky.social and its profile link).

A run whose input is rejected before any request (e.g. profiles without handles) costs only the start fee.

Cost examples (Free plan prices)

Scenario	Input	Charged events	Cost
Prefill search	1 query, 30 posts	start + 30 posts	about $0.028
1,000 posts from accounts	2 accounts × 500	start + 1,000 posts	about $0.90
100 profiles	100 handles	start + 100 profiles	about $0.27
Search with no matches	1 query	start + 1 empty result	$0.0015

Spending limit

Set Maximum cost per run in the run options. The Actor checks the limit before every row; when the next row would exceed it, the run stops with status CAPPED and keeps everything written so far.

Limits

• Search: up to 100 posts per query (Bluesky's limit without login). Split the topic into narrower queries or date ranges to see more.
• Account posts: up to 10,000 feed items scanned per account. A scan also stops early when fewer than 1 in 100 items match the filters after the first 1,000 items, so a sparse filter does not read the whole history.
• Feeds: the same sparse stop applies.
• Speed at 256 MB: one account's 1,030 posts in about 5 s; 10,000 account posts took about 2–3 minutes in tests. Large jobs need a longer run timeout than the default 300 s; an input cut by the timeout is marked partial with its row count.
• since / until compare the post's createdAt (for reposts, repostedAt). Account feeds are ordered by when Bluesky indexed the post, so posts imported later with an older date can be outside the scanned range.

Switching from another Bluesky scraper

Field names differ from other Bluesky Actors. Typical mapping:

Other Actor	This Actor
mode: search-posts	mode: searchPosts
mode: posts	mode: authorFeed
mode: profiles	mode: profiles
mode: threads / post-urls	mode: thread
sort	searchSort
maxResults	maxItems
authorHandle, authorName, authorDid, authorAvatar	author.handle, author.displayName, author.did, author.avatar
lang	langs
mediaUrls, hasMedia	images[].url, video.playlistUrl, embedType
embedUrl, embedTitle	externalLink.url, externalLink.title (also in links)
scrapedAt	fetchedAt

Extra here: image alt text, video, quoted post, repost details, thread roles, custom feeds and lists, and not_found rows.

API and integrations

• API: start a run with POST https://api.apify.com/v2/acts/cliqtomedia~bluesky-scraper/runs and the JSON input above; read rows from the run's default Dataset.
• Export: the Dataset downloads as JSON, CSV, Excel, XML or RSS from the Output tab or the API.
• Scheduling: use Apify Schedules, e.g. a daily search with since set to yesterday.
• Integrations: Make, Zapier, n8n, Google Sheets and webhooks connect through the Apify integrations tab.

Use with AI agents (MCP)

• MCP server with only this Actor: https://mcp.apify.com?tools=cliqtomedia/bluesky-scraper.
• Smallest useful input: {"mode": "searchPosts", "queries": ["apify"], "maxItems": 30}.
• Key fields: text (post text), author.handle (who posted), createdAt (when), likeCount (engagement), url (link to cite).
• A small search finishes in a few seconds and costs about $0.03. At most 100 posts per search query; not_found and error rows explain missing data.

Troubleshooting

Search stops at 100 posts

Symptom: you set maxItems above 100, a query returns 100 posts, its entry in inputs is partial and there is a warning about 100 posts per query. With maxItems of 100 or less the query is complete. Reason: Bluesky serves only the first search page without login. Action: add operators (lang:en, from:, domain:) or a narrower since / until.

An account returns no posts

Check the account's entry in inputs in RUN_SUMMARY. not_found means the handle does not exist. no_results means the filters matched nothing; partial with the sparse-scan warning means the filters matched fewer than 1 in 100 items. Loosen since / until, turn on replies or reposts, or use search with from:handle.

The run ends as PARTIAL with skipped inputs

The run hit its timeout. Raise the timeout in the run options or split the input into smaller runs.

The run fails with RUN_TIMEOUT_TOO_SHORT

The run timeout leaves no time to fetch data and save the summary. Set the timeout to at least 60 seconds; the default 300 seconds is enough for small runs. Nothing is charged except the start.

Rows with error status

Bluesky failed or refused the request after retries (RATE_LIMITED, SOURCE_FORBIDDEN, SOURCE_UNAVAILABLE). These rows are free; run the failed values again later.

FAQ

Do I need a Bluesky account or app password?

No. The Actor reads only public data from the public Bluesky API and never asks for login.

Why is search limited to 100 posts per query?

Bluesky serves further search pages only to logged-in clients. The Actor respects that and tells you in the run summary when a query hit the limit.

Are the counts live?

They are a snapshot at fetchedAt. Run again to refresh them.

Is this data complete?

It is what the Bluesky API returns at fetch time. Deleted posts, blocked content and posts hidden by moderation are not included, and counts can be null when Bluesky does not report them.

Can I scrape followers or DMs?

Not in this version. DMs are private and will never be supported.

Is scraping Bluesky allowed?

The Actor uses only public endpoints at a modest request rate and does not bypass any limit. You are responsible for how you store and reuse the content, including personal data and authors' rights.

Learn more

• Bluesky API directory: the public AppView endpoints this Actor reads.
• Search API reference: query operators such as from:, lang: and domain:.
• AT Protocol: how DIDs, handles and at:// URIs work.
• Bluesky Terms of Service.

Support

Use the Issues tab of this Actor for bugs, questions and feature requests. Include the run ID, the mode and the error text. Never post tokens, cookies or private data.