X (Twitter) Profile & Post Engagement Extractor

Collect public X profiles and posts without login, cookies, developer keys or API tokens. The actor returns a profile row, post rows with engagement and media, and explicit status rows when X limits or hides something.

What it collects

• Profile header from the public profile page: handle, display name, bio, bio links and emails, location, website, join date, follower/following/post/media/like counters, verification type, avatar, banner, public birth date, pinned post and professional label.
• Latest posts from the profile's server-rendered originals timeline, strictly newest first, topped up through the cursor-paged timeline when the target is higher than the 5-9 server-rendered rows. A pinned post is marked isPinned but stays in its chronological position (and may exceed the cap by one).
• Top posts ranked by total engagement. X exposes no public Top tab to logged-out callers, so the actor combines the guest search Top product with the profile timeline candidate pool and ranks the merged set locally; a pinned post is not promoted above its earned rank.
• Media posts by filtering the collected timelines to photos, videos and GIFs. X does not serve a full media tab to logged-out callers. Filter-heavy runs page the timeline until the requested number of matching rows is found or the source ceiling is reached, so rare GIF matches are not missed by a fixed buffer.
• Replies and Highlights as best-effort groups. Highlights are cursor-paged through the legacy guest query, so large curated sets return their full observed depth; replies remain limited. X sometimes refuses those guest timelines; the actor reports repliesLimited or highlightsLimited instead of pretending the account has none.
• Specific post URLs through postUrls: the focal post plus up to three server-rendered replies per URL.
• Derived activity metrics on profile rows: account age, followers/following ratio, posts per day, last post date, days since the last post, recent average likes/replies/reposts/views and a recent engagement percentage.
• Provenance and timestamps on every row: type, recordType, sourceUrl, collectedAt and scrapedAt.

The actor never signs in, never asks for cookies and never uses a user-supplied token.

Input

Field	Type	Default	Description
usernames	array	–	Optional; supply this, postUrls, or both. Handles (NASA, @NASA), profile links (https://x.com/NASA), numeric ids (11348282, id:11348282) or https://x.com/i/user/11348282.
include	multi-select	["profileInfo", "latestPosts"]	profileInfo, latestPosts, topPosts, media, replies, highlights; legacy about is accepted but reported as unavailable. Groups are additive.
postUrls	array	[]	Exact post URLs or ids; can run alone or alongside profiles. Each returns the focal post and up to 3 replies.
maxPostsPerProfile	integer 1-1000	50	Post target per account. The actor follows X's guest timeline bottom cursor (X's own ceiling is about 3,200 posts). When a target is still unmet, older posts are filled from search windows only when X serves them.
includeReplies	boolean	false	Keep replies the account sent (best effort).
includeRetweets	boolean	false	Keep reposts.
includeHighlights	boolean	false	Also collect the Highlights tab. Cursor-paged up to maxPostsPerProfile; observed depths include 113, 287 and 1,097 rows on different accounts.
dateFrom, dateTo	date picker	–	Client-side post date window. dateTo includes the whole day.
language	text	–	Post language, e.g. en, es, ja. Locale forms such as en-US and pt-BR are normalized to en/pt in code, so old tasks keep working.
minimumLikes, minimumRetweets, minimumReplies, minimumViews, minimumQuotes	integer ≥ 0	–	Client-side engagement minimums. Posts missing the requested count are dropped when the minimum is set.
onlyWithMedia	boolean	false	Keep only posts with photos, videos or GIFs.
mediaType	select	any	any, photos, videos, gifs.
onlyVerified	boolean	false	Skip accounts without a verification badge.
verifiedType	select	any	any, blue, business, government. X and mirror labels are normalized in code.
minFollowers, maxFollowers	integer ≥ 0	0	Account-level follower range. 0 means any.
minAccountAgeDays	integer ≥ 0	0	Skip accounts younger than this. Accounts without a public join date are skipped when set.
withWebsiteOnly	boolean	false	Skip accounts with no website in the profile or bio.
bioKeywords	array	[]	Keep accounts whose name or bio contains every keyword (case-insensitive).
activeWithinDays	integer ≥ 0	0	Skip accounts whose newest public post is older than this.
enrichEngagement	boolean	true	Refresh each collected post through its public status page for views, bookmarks and complete media. One request per post; turn off for the cheapest profile-only runs.
proxyConfiguration	proxy	–	Optional override. The actor works without a proxy and tries the direct connection first.

Examples

Profile only (the daily prefill shape):

{
  "usernames": ["NASA"],
  "include": ["profileInfo"],
  "maxPostsPerProfile": 3,
  "enrichEngagement": false
}

Profile plus latest and top posts:

{
  "usernames": ["NASA", "https://x.com/apify"],
  "include": ["profileInfo", "latestPosts", "topPosts"],
  "maxPostsPerProfile": 100
}

High-engagement photo posts from a date range:

{
  "usernames": ["NASA"],
  "include": ["latestPosts"],
  "mediaType": "photos",
  "minimumLikes": 5000,
  "minimumViews": 1000000,
  "dateFrom": "2026-01-01",
  "dateTo": "2026-10-27"
}

Lead-gen profile filter:

{
  "usernames": ["apify", "NASA"],
  "include": ["profileInfo"],
  "minFollowers": 10000,
  "minAccountAgeDays": 365,
  "withWebsiteOnly": true,
  "bioKeywords": ["space"]
}

Specific posts:

{
  "usernames": ["NASA"],
  "include": ["profileInfo"],
  "postUrls": ["https://x.com/NASA/status/2114801524358406570"]
}

Output

Every row has type (profile, post or status), recordType, collectedAt and scrapedAt. Profile and post rows, and status rows with a known account or post URL, also carry sourceUrl. Empty fields are omitted recursively: the dataset never contains null, "", [] or {}.

Profile row

Keeps the original fields: id, username, name, bio, bioLinks, bioEmails, location, website, joinedAt, birthDate, followers, following, postCount, mediaCount, likesGiven, listedCount, isBlueVerified, verified, verifiedType, label, isProtected, avatarUrl, bannerUrl, pinnedPostId, pinnedTweetId, pinnedTweet, mediaTweets, withReplies, profileStatus, collectedAt.

Adds: profileUrl, websiteDisplayUrl, bioMentions, accountAgeDays, followersToFollowingRatio, postsPerDay, lastPostAt, daysSinceLastPost, recentAvgLikes, recentAvgReplies, recentAvgReposts, recentAvgViews, recentEngagementRatePct, professionalType, professionalCategory, affiliateLabel, highlightCount, recordType, sourceUrl, scrapedAt.

websiteDisplayUrl is the label X shows for the profile's website entity (for example nasa.gov). Bio links stay in bioLinks and are never reused as the website label.

Conditional fields are omitted when X does not expose them. listedCount is not guaranteed by current cookieless payloads.

Post row

Keeps the original fields: id, url, text, createdAt, language, replyCount, retweetCount, likeCount, quoteCount, bookmarkCount, viewCount, engagementRate, isReply, isRetweet, isQuote, isNoteTweet, isPinned, conversationId, threadRootId, threadPosition, inReplyToTweetId, inReplyToUserId, inReplyToUsername, hashtags, mentions, urls, cashtags, media, quotedTweet, quotedTweetId, poll, card, possiblySensitive, place, postedFrom, withheldInCountries, isEdited, editControl, newsActionType, parentTweet, author, username, collectedAt.

Adds: totalEngagement, engagementPerView, mediaTypes, source, isHighlight, authorId, authorUsername, authorName, recordType, sourceUrl, scrapedAt.

source is provenance: profileLatest, profileTop, profileHighlight, profileReplies, syndication, search or status.

conversationId and threadRootId are set on postUrls replies from the focal post when the status page omits them, so every reply can be grouped by conversation.

totalEngagement is emitted only when likes, reposts, replies and quotes are all real numbers. engagementPerView additionally needs a positive view count. No count is invented.

Media item: type (photo / video / gif), url, previewImageUrl, width, height, durationSeconds, viewCount, altText, variants, and where X provides them mediaKey, allowDownload, mediaAvailability, focusRects. Video url prefers the best MP4 at or below 1080p; the full variant list (including 4K) remains in variants.

Status row

Status rows explain a skipped, protected, missing, empty or limited account:

Status	Meaning
noInput	No usable usernames or post URLs were supplied.
invalidUsername	A username/link/id could not be parsed.
notFound	The account does not exist, was removed or is suspended.
rateLimited	X temporarily limited the profile request.
skipped	An account-level filter excluded the account. The message names the filter.
protected	The account is protected; only its public header is returned.
repliesLimited	Replies were requested but X served none or fewer than the target this run; achieved and target are included.
highlightsLimited	Highlights were requested but X returned no rows or refused the request this run.
noMediaPosts	Media mode found no matching photos, videos or GIFs.
noPosts	Posts were collected but every one was filtered out.
noVisiblePosts	The account exists but X shows no public posts.
timelineUnavailable	The profile loaded but its posts could not be collected.
searchLimited	Fewer posts than requested were available; achieved and target are included.
aboutUnavailable	A legacy task requested X's login-only About panel.
error	An unexpected per-profile failure.

Data source and limits

• Primary: the public profile HTML, including X's server-rendered GraphQLRequestStream.Completed payload. This is the same page any logged-out visitor sees.
• Fallbacks: guest GraphQL (UserByScreenName header, cursor-paged UserTweets ranked timeline, Highlights), the syndication timeline, and the public FxTwitter mirror. Guest UserByScreenName is merged before FxTwitter; FxTwitter is a third-party mirror and remains the last resort for post views or a missing header field.
• No credentials: no cookies, login, API key or user proxy is required.
• Proxy: the actor tries a direct connection first, then the Apify AUTO datacenter tier after a real block. A paid residential tier is hardcoded as a last resort and is never a schema default. A user-supplied proxyConfiguration overrides the built-in ladder.

Honest limits

• About panel: login-only. The legacy about value remains in the include enum for saved-task compatibility and is labeled as unavailable; requesting it emits an aboutUnavailable status instead of failing the task.
• Full replies: X does not serve the full reply tab to logged-out callers, and the guest reply timeline/search refuses most attempts. Replies remain best effort and report repliesLimited when empty; use postUrls for a focal post plus up to three server-rendered replies.
• Full media tab: X's /media route is a login wall for guests and the guest UserMedia operation returns 404. Media mode filters the collected timelines instead; cursor pagination can now collect far more media posts than the old one-page path.
• Deep history: the guest UserTweets bottom cursor pages a profile timeline back to X's ceiling of about 3,200 posts, and maxPostsPerProfile supports 1-1000. Guest search is not reliable, so older-row top-ups beyond the cursor are best effort.
• Highlights: the old userId Highlights query pages the curated set through its bottom cursor, not a fixed 20-row page. Observed depths include 113 (second-opinion sample), 287 (NASA) and 1,097 (elonmusk), bounded by the account's Highlights set and maxPostsPerProfile. The live screenName query remains a one-page fallback.
• Syndication quota: about 30 profile timeline lookups per IP per 15 minutes. The actor falls back to guest GraphQL and FxTwitter instead of retrying forever.
• listedCount: best effort only. Syndication returns a placeholder 0 and that value is never published; only a search author snapshot carries the real count, and guest search itself is limited.
• postedFrom, place: conditional. postedFrom is filled when the timeline row carries a top-level source; place is not exposed by current cookieless payloads and is omitted.
• Views: recent posts get views from the status page; tweets from before X introduced view counts have no view count anywhere.
• 4K video: the 4K MP4 can be hundreds of megabytes. media.url prefers 1080p or lower; the 4K URL stays in variants.

FAQ

Do I need an X account or developer key? No. The actor only reads public pages and needs no credential input.

Why are there fewer posts than maxPostsPerProfile? The account may have fewer public posts in the reachable timeline, the filters may remove rows, or X may refuse the guest search top-up. The actor merges and cursor-pages every public source, then reports searchLimited with the achieved count only when a real source cap leaves the target unmet.

Why is a count missing instead of 0? A missing field means X did not expose it. The actor never substitutes a sentinel or an invented zero. A real zero (for example quoteCount: 0) is preserved.

Why did my account get a skipped row? An account-level filter (onlyVerified, verifiedType, follower range, account age, website, bio keywords or activeWithinDays) excluded it. The message says which one.

Can I collect protected accounts? Only their public header. Posts require approval from the account owner, which the actor does not attempt.

Is the data live? Yes. Every run reads X at request time; schedule the actor for hourly, daily or weekly refreshes.

How do I use the dataset? Use the run page, the Apify API, webhooks, or a no-code connector such as Zapier, Make or Google Sheets.

Is this affiliated with X? No. It is an independent third-party actor that reads publicly available pages.

Development

Run the offline tests from the actor folder:

$python -m pytest tests/ -q

Run locally with the Apify CLI:

$apify run --purge --input-file input.json