Bluesky Post Search Scraper

What is Bluesky Post Search Scraper?

Bluesky Post Search Scraper is an Apify Actor for searching public Bluesky posts by keyword, author, date, language, hashtag, mention, linked domain, linked URL, or minimum likes. It runs one query or a list of up to 20 queries per run (searchQueries), needs no Bluesky account, and returns one structured record per post with text, author identity, timestamps, engagement counts, media fields, facets, and source URLs. Newest-first searches return up to 1,000 posts per query, and an only-new mode returns just the posts that appeared since your last run, which suits scheduled monitoring and AI-agent workflows.

Best fit and connected workflows

This Actor fits workflows that start with a targeted Bluesky search and end with a structured list of matching public posts. Typical routing includes:

• topic monitoring from a phrase or Lucene-style query
• author-scoped searches for public accounts
• language, hashtag, mention, domain, and linked-URL filtering
• keeping only posts with a minimum number of likes, reposts, or replies
• scheduled monitoring that returns only posts new since the previous run
• Use follow-up review of selected posts with Bluesky Posts Scraper - Search, Feeds & Threads when you need a downstream workflow that operates on a returned public URL or ID

Because each dataset row represents one post, the output is easy to filter, store, enrich, or hand to an agent for later steps.

Practical scenario

A product marketer tracks launch chatter for her brand and two competitors. She runs searchQuery: "notion" with searchQueries: ["obsidian app", "logseq"], sets lang to en, and keeps maxResults at 25. The Actor returns rows with postUrl, text, authorHandle, createdAt, likeCount, repostCount, replyCount, quoteCount, tags, and externalUrl. She spots one post that links to a release note page and has strong engagement. Her next action is to open the post URL for review and pass the public URL into a downstream enrichment workflow.

Input

The Actor needs searchQuery (or a searchQueries list), plus optional server-side filters.

Field	Type	Purpose
searchQuery	string	Focused search text or supported Lucene-style query
searchQueries	array[string]	More queries to run in the same run (up to 20), each with the same filters
sort	string	latest or top ordering from Bluesky search
maxResults	integer	Maximum posts per query, from 1 to 1000 (latest order; top order stops at 100)
since	string	Lower bound for indexed search time
until	string	Exclusive upper bound for indexed search time
lang	string	BCP 47 language tag
author	string	Public Bluesky author handle or DID
mentions	string	Mentioned account handle or DID
domain	string	Linked domain hostname
url	string	Absolute linked URL
tags	array[string]	Hashtag facets, up to 10 tags
includeReplies	boolean	Whether reply posts may appear in results
minLikes	integer	Keep only posts with at least this many likes (skipped posts are not charged)
minReposts	integer	Keep only posts with at least this many reposts
minReplies	integer	Keep only posts with at least this many replies
onlyNewSinceLastRun	boolean	Return only posts newer than the last post this search delivered (needs latest order)
monitoringStoreName	string	Optional name for the memory used by only-new mode (default bluesky-post-search-monitor)

Focused input example:

{
  "searchQuery": "open source",
  "sort": "latest",
  "maxResults": 25,
  "since": "2026-07-01",
  "until": "2026-07-15",
  "lang": "en",
  "author": "jay.bsky.team",
  "mentions": "",
  "domain": "github.com",
  "url": "",
  "tags": ["opensource"],
  "includeReplies": true
}

Output

Each dataset row contains one validated public Bluesky post.

Field	Type	Purpose
searchQuery	string	Exact query used for the row
sort	string	Ranking mode used for the search
rank	integer	One-based rank in the run
postUri	string	Stable AT Protocol URI
postCid	string	CID of the returned post version
postUrl	string	Human-readable Bluesky post URL
text	string	Public post text
authorDid	string	Author DID
authorHandle	string	Author handle
authorDisplayName	string or null	Public display name when available
authorAvatarUrl	string or null	Public avatar URL when returned
createdAt	string	Post record timestamp
indexedAt	string	AppView index timestamp
langs	array[string]	Declared language tags
likeCount	integer	Likes at collection time
repostCount	integer	Reposts at collection time
replyCount	integer	Replies at collection time
quoteCount	integer	Quote posts at collection time
engagementCount	integer	Sum of engagement counts
isReply	boolean	Reply indicator
replyRootUri	string or null	Reply root AT URI
replyParentUri	string or null	Direct parent AT URI
links	array[string]	Rich-text link URLs
mentionedDids	array[string]	Mention facet DIDs
tags	array[string]	Hashtags found on the post
labels	array[string]	Content labels returned on the post view
imageUrls	array[string]	Public image URLs
imageAltTexts	array[string]	Image alt text aligned by index
videoPlaylistUrl	string or null	HLS playlist URL for video embeds
videoThumbnailUrl	string or null	Video thumbnail URL
externalUrl	string or null	External card URL
externalTitle	string or null	External card title
externalDescription	string or null	External card description
externalThumbUrl	string or null	External card thumbnail
quotedPostUri	string or null	Quoted post AT URI
collectedAt	string	Dataset normalization timestamp
sourceEndpoint	string	Bluesky AppView endpoint used
_note	string or null	Agent-readable continuation note

Illustrative output record:

{
  "searchQuery": "open source",
  "sort": "latest",
  "rank": 1,
  "postUri": "at://did:plc:example/app.bsky.feed.post/3abc",
  "postCid": "bafyreiexample",
  "postUrl": "https://bsky.app/profile/alice.bsky.social/post/3abc",
  "text": "A new open-source release is available today.",
  "authorDid": "did:plc:example",
  "authorHandle": "alice.bsky.social",
  "authorDisplayName": "Alice",
  "authorAvatarUrl": null,
  "createdAt": "2026-07-15T10:00:00.000Z",
  "indexedAt": "2026-07-15T10:00:01.000Z",
  "langs": ["en"],
  "likeCount": 42,
  "repostCount": 7,
  "replyCount": 5,
  "quoteCount": 2,
  "engagementCount": 56,
  "isReply": false,
  "replyRootUri": null,
  "replyParentUri": null,
  "links": ["https://example.com/release"],
  "mentionedDids": [],
  "tags": ["opensource"],
  "labels": [],
  "imageUrls": [],
  "imageAltTexts": [],
  "videoPlaylistUrl": null,
  "videoThumbnailUrl": null,
  "externalUrl": "https://example.com/release",
  "externalTitle": "Release notes",
  "externalDescription": "What's new in this version.",
  "externalThumbUrl": null,
  "quotedPostUri": null,
  "collectedAt": "2026-07-15T11:00:00.000Z",
  "sourceEndpoint": "https://api.bsky.app/xrpc/app.bsky.feed.searchPosts",
  "_note": null
}

The Actor also writes compact OUTPUT and diagnostic RUN_SUMMARY records to the key-value store.

How it works

This Actor searches public Bluesky posts through Bluesky's public AppView API. Each query runs with the same optional server-side filters; posts found by more than one query are saved once. The implementation uses the public app.bsky.feed.searchPosts route and normalizes the returned post view into dataset rows. The dataset schema keeps all listed fields explicit, with null or empty arrays where applicable, so downstream tools can read the data consistently.

Going past 100 posts. One Bluesky search page holds at most 100 posts, and the public route does not allow page-by-page continuation. In latest order the Actor therefore moves a time window backwards: after each page it searches again for posts older than the oldest one it has, removes the repeated boundary post, and continues until it reaches maxResults, runs out of matches, or has scanned 3,000 posts for that query. top order has no time direction, so it stays at one page of up to 100 posts. If Bluesky changes this behavior, the run still returns what it collected and says so in warnings.

Only new posts. With onlyNewSinceLastRun, each query and filter combination keeps a small record of the newest post it delivered, in a named storage of your account. The next run asks only for newer posts. If more new posts exist than maxResults, the oldest of them are not delivered later, so raise maxResults or schedule runs more often. Changing the query or any filter starts a fresh memory.

Time limit. A run stops a little before its time limit and finishes as PARTIAL with the posts already saved, rather than timing out.

When direct Bluesky search is access-restricted, the Actor retries through a managed unblocker route (at most six pages per run), using the same query and filters; this fallback is covered because Apify platform usage is included. OUTPUT.unblocker and RUN_SUMMARY.unblocker report whether that fallback was used. If every route stays unavailable, the run reports UPSTREAM_FAILED without charging for posts.

Pricing

This Actor uses Pay per event with Apify platform usage included.

• Actor start: charged once per run, scaled by the run's memory allocation
• post-scraped: charged for each schema-valid post persisted to the default dataset (posts found by several queries are charged once)
• Apify platform usage: included in the event price

A run that persists one hundred posts charges one start event plus one hundred post-scraped events; Apify platform usage is included. With searchQueries, the cap is maxResults posts per query. Posts skipped by minLikes, minReposts, minReplies, includeReplies, or only-new mode are not charged. Higher Apify plans see lower per-post prices on the live Pricing tab, which shows the current billable values and totals.

Use with AI agents (MCP)

This Actor is usable through Apify MCP as a tool for public Bluesky post search. The exact Actor identity is khadinakbar/search-bluesky-posts.

The tool is suitable used with a focused query, optional filters, and a bounded maxResults so an agent can read the returned dataset and continue with a second step if needed.

Search public Bluesky posts for "open source" from the last month, keep results in English, and return up to 25 posts. Then summarize the most relevant post URLs and engagement counts.

Output interpretation:

• dataset rows contain one post each
• rank shows the order of persisted rows in the run
• sourceEndpoint identifies the Bluesky-hosted AppView route
• collectedAt records when the row was normalized
• _note may help an agent continue after a capped execution Provenance and scope:
• results come from public Bluesky AppView search
• the Actor returns public posts, authors, timestamps, engagement, media, and URL fields
• The search scope is post search, not profiles, followers, or private data

Pagination and cost guidance:

• maxResults can be set from 1 to 1000 per query in latest order; top order stops at 100
• for a recurring agent job, turn on onlyNewSinceLastRun so each run returns only unseen posts
• tighter filters, including minimum likes, make follow-up runs more specific
• each persisted post creates a post-scraped event charge; skipped posts do not

JavaScript example

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });

const run = await client.actor('khadinakbar/search-bluesky-posts').call({
  searchQuery: 'open source',
  sort: 'latest',
  maxResults: 25,
  lang: 'en'
});

const dataset = await client.dataset(run.defaultDatasetId).listItems();
console.log(dataset.items);

Best results and outcome guidance

Start with one focused query, add related phrases to searchQueries for monitoring several topics in one run, and then narrow by author, language, date, hashtag, mention, domain, or URL as needed. If you are building an agent workflow, keep the first run small, inspect text, authorHandle, createdAt, and postUrl, then decide whether a second run should target a more specific slice of the search space. Use includeReplies based on whether reply posts belong in the review set.

Continue the workflow

• Then use Bluesky Posts Scraper — Search, Feeds & Threads to continue from Bluesky Post Search Scraper — Posts & Metrics discovery into content data for the selected records.
• Then use Skool Community Scraper to extend Bluesky Post Search Scraper — Posts & Metrics with a neighboring social-media research source when the brief calls for Skool data.

Builder's note

I built this for brand and topic monitoring. I found that the dataset contract keeps searchQuery, sort, rank, postUrl, text, authorHandle, createdAt, and engagement counts in every row, which makes each record self-describing for later analysis.

FAQ

How many posts can one query return?

Up to 1,000 per query in latest order (set maxResults), and up to 100 in top order. The Actor reaches past Bluesky's 100-post page by moving a time window backwards, so a very busy phrase may still hold more posts than one run collects. Narrow it with since / until or add more phrases to searchQueries.

How does only-new mode work for a daily schedule?

Turn on onlyNewSinceLastRun and schedule the same input. The first run returns up to maxResults posts; each later run returns only posts newer than the last post it delivered, and finishes as VALID_EMPTY when nothing new appeared. Use monitoringStoreName to keep separate memories, for example one per client.

Why does top order stop at 100?

Relevance order has no time direction, so it cannot be walked backwards in time. Use latest order with a date range when you need depth.

Are posts skipped by minimum likes charged?

No. Only posts saved to the dataset create a post-scraped charge. A strict filter on a broad phrase can scan many posts to find a few matches, and a run stops after scanning 3,000 posts per query.

When should I use author instead of searchQuery?

Use author when you want public posts from one Bluesky account and searchQuery when you are tracking a topic, phrase, or query syntax pattern.

When should I use mentions?

Use mentions when the post must include a structured mention facet for a specific account handle or DID.

When should I use domain or url?

Use domain when you want posts linking to a hostname, and url when you need the exact linked web address.

How do tags work?

tags targets structured hashtags. Multiple tags are matched together, and the tags are entered without the # prefix.

What does includeReplies change?

includeReplies controls whether reply posts may remain in the final dataset after search results are returned.

How should I choose between latest and top?

Use latest for newest indexed posts and top for relevance-ranked results.

How can I use this output in a downstream workflow?

Use the returned postUrl, postUri, or authorHandle as the starting point for a second Apify workflow or for agent-side enrichment.

Responsible use

This Actor searches public Bluesky posts only. Use the returned data in line with applicable law, Bluesky and AT Protocol terms, and your own privacy and research obligations. Keep collection proportional to the task, store data carefully, and apply special care when combining public social data with other sources.