Bluesky Posts Scraper · Keywords, Hashtags, Only New Posts

Scrape Bluesky posts by keyword, hashtag, profile or post URL (reply thread + quote posts). Every post becomes one flat row with exact like, repost, reply, quote and bookmark counts, hashtags, links, media and reply/quote references. No login, no Bluesky account, no API key.

Why this scraper

• Search that keeps going past the first page. Logged-out Bluesky search refuses the cursor after page one (HTTP 403) and throttles single IPs hard. This Actor pages back through time instead (until stepping, deduplicated at the page borders) and sends every search page through a fresh proxy session. Measured on 23 Sep 2026: 15 pages × 100 posts in a row, 0 failed pages. If a page still fails after retries, the run summary says so per keyword — it never comes back short silently.
• Only new posts, for scheduled monitoring. Turn on onlyNewPosts and schedule the Actor. It remembers the newest post per keyword / hashtag / profile / thread and on the next run returns only what is newer. You are not charged again for posts you already have.
• Real hashtag matches. A hashtag search returns only posts that actually carry the tag, not posts that merely contain the word. Posts dropped by this or any other filter are never charged.
• Honest thread counts. For a post URL you get the post, every reply Bluesky's thread API returns (with depth), and optionally every quote post. The summary shows replyCountOnPost next to repliesReturnedByBluesky, so you can see exactly what Bluesky did and did not hand out.
• A hard limit you can trust. maxPosts stops the run cleanly; you are never charged for more posts than that, and never for duplicates.

Input example

{
    "searchTerms": ["web scraping", "\"open source\""],
    "hashtags": ["buildinpublic"],
    "profiles": ["bsky.app"],
    "postUrls": ["https://bsky.app/profile/bsky.app/post/3l6oveex3ii2l"],
    "includeQuotes": true,
    "sinceDate": "2026-09-01",
    "language": "en",
    "minLikes": 5,
    "onlyNewPosts": true,
    "maxPosts": 5000
}

Search filters: language, fromAuthor, mentions (brand monitoring), domain (posts linking to a site), sinceDate / untilDate. Bluesky search syntax inside a keyword works too.

Output example

{
    "url": "https://bsky.app/profile/bsky.app/post/3l6oveex3ii2l",
    "uri": "at://did:plc:z72i7hdynmk6r22z27h6tvur/app.bsky.feed.post/3l6oveex3ii2l",
    "authorHandle": "bsky.app",
    "authorDisplayName": "Bluesky",
    "text": "👋  Bluesky is an open social network that gives creators independence from platforms, ...",
    "createdAt": "2024-10-17T07:06:51.491Z",
    "likeCount": 63696,
    "repostCount": 9531,
    "replyCount": 8592,
    "quoteCount": 708,
    "bookmarkCount": 254,
    "languages": ["en"],
    "hashtags": [],
    "links": [],
    "images": [],
    "externalLink": null,
    "quotedPostUrl": null,
    "isReply": false,
    "replyParentUri": null,
    "sourceType": "thread-root",
    "source": "at://did:plc:z72i7hdynmk6r22z27h6tvur/app.bsky.feed.post/3l6oveex3ii2l"
}

Field	Meaning
likeCount, repostCount, replyCount, quoteCount, bookmarkCount	Exact integers as Bluesky's AppView returns them at scrape time
createdAt, indexedAt	ISO 8601 UTC
hashtags, links, mentionedDids	From the post's rich-text facets
images, video, externalLink, quotedPostUrl	Embedded media, link card, quoted post
isReply, replyParentUri, replyRootUri, depth	Reply structure (depth for thread replies)
sourceType	search, hashtag, profile, thread-root, thread-reply, quote

A run summary (SUMMARY in the key-value store) lists per source: posts written, fetched, skipped as already seen last run, filtered out (by reason), pages, and why paging stopped.

Honest limits

• Top sort is one page. Logged-out Bluesky search serves relevance-sorted ("Top") results as a single page of about 100 posts. Use latest (the default) for more.
• Threads: Bluesky's thread API does not return every reply on very large threads (e.g. 739 of 8,592 on Bluesky's own welcome post, measured 23 Sep 2026). The summary shows both numbers.
• Search needs the Apify proxy (on by default): Bluesky refuses logged-out search from bare cloud IPs. Profiles, threads and quotes do not use it.
• Counts are a snapshot at scrape time. Deleted and private-list content is not available.
• Posts only. Follower lists and profile details are not part of this Actor.

Pricing (pay per event)

• Actor start: tiny flat fee per run
• Per scraped post: charged only for posts written to the dataset — no charge for duplicates, filtered-out posts, or posts you already got in an earlier onlyNewPosts run