X (Twitter) Account Watchlist

Give it a list of X accounts and it brings back their posts: the text, when it went up, how many likes, reposts, replies and quotes it has, the media and links inside it, and the author's follower count at the moment you read it. Turn on monitor mode, put it on a schedule, and every run after the first returns only what has been posted since the last one. That is the part a social listening subscription charges by the month for.

There is no login, no API key and no X developer account. It reads the same public embed timeline X itself serves to any web page that puts a profile widget on it, so nothing here depends on an account that can be suspended.

This is a watchlist, not a search engine. If you want every post mentioning a keyword across all of X, this is the wrong tool. If you want to know what a specific set of accounts said, it is exactly the right one.

What it costs

One flat price per post delivered, platform usage and residential proxies included, plus a small fee for each account actually read. Handles that do not exist, are suspended or are protected are never billed. Neither is an account we could not reach. A monitor run that finds nothing new costs the start fee and the per-account check, nothing more.

	Price	Billed when
Post	$0.20 per 1,000	A post is delivered in the results
Account checked	$0.003 each	The account's public timeline was read and parsed
Actor start	$0.001 per run	Once per run

At the list price a watchlist of 10 accounts pulling 50 posts each costs about 13 cents. The volume leader in this category charges $0.40 per 1,000 posts.

What you get

One record per post (recordType: "post"):

Field	Description
account, handle	The account as you typed it, and the handle X resolved it to
authorName, authorId, authorUrl, authorProfileImageUrl	Who wrote it
authorFollowers, authorVerified	Follower count and verification, read at the time of the run
postId, url	The post id as a string (they are 64-bit, so never read them as numbers) and a direct link
postedAt	When it went up, ISO 8601 UTC
text, textLength	The full text. For a repost this is the original post, not the truncated "RT @" version
lang	The language X detected
likeCount, repostCount, replyCount, quoteCount	Engagement at the time of the run
isReply, isSelfThread, replyToHandle, replyToPostId, replyToUrl	Whether it answers an earlier post, and which one. An account replying to itself is how threads are written, so those are flagged as threads
isRepost, repostOfHandle, repostOfPostId, repostOfUrl, repostOfPostedAt	Where a repost came from
isQuote, quotedHandle, quotedPostId, quotedUrl, quotedText	The post being quoted, with its text
hasMedia, mediaTypes, mediaUrls	Photos, videos and GIFs attached, with direct image URLs
videoUrl, videoDurationSeconds	The best MP4 X exposes for a video post, and how long it runs
links, hashtags, mentions, symbols	Links resolved past t.co, plus hashtags, @mentions and cashtags
linkPreviewUrl, linkPreviewTitle, linkPreviewSite	The link card under the post, when there is one
possiblySensitive	X's own sensitive-content flag
conversationId	The thread this post belongs to
isNewSinceLastRun, isBaseline	Monitor mode only: whether this is fresh or part of the first snapshot
scrapedAt	Run timestamp

And one summary per account (recordType: "account"):

Field	Description
handle, authorName, authorId, authorBio, authorLocation	The profile as X shows it
authorFollowers, authorFollowing, authorPostsTotal, authorLikesTotal, authorListedCount	The full counter set
authorVerified, authorVerifiedType, authorProtected, authorCreatedAt	Verification, protection and when the account was opened
authorProfileImageUrl, authorBannerUrl, profileUrl	Pictures and the profile link
status, statusDetail	ok, unavailable, rate-limited or error, and a sentence saying why. Only ok is billed
postsInFeed, postsFetched, postsDelivered	What the public timeline held, what matched your filters, what was billed
latestPostIdInFeed	The newest post id X had at read time
lastPostId, previousLastPostId	Monitor mode: where the watch stands now and where it stood before
includeRetweets	The filter this run used

Media is reported, not downloaded, so you learn a post carried a 4K video without paying to transfer it.

Who uses it

Brand and comms teams watching competitors, executives and industry accounts so a launch or an incident is seen the hour it goes up rather than the morning after. Crypto and trading desks following project and founder accounts for announcements. Newsrooms watching official and government accounts. Investor-relations and VC teams following portfolio founders. Researchers building a dated corpus of what a defined set of accounts published, without a personal account in the loop. Anyone who used to pay for a social listening seat and only ever used the watchlist half of it.

Input

Field	What it does
handles	The accounts to watch, up to 200 per run. A name (nasa), a handle (@nasa) or a profile link all work
maxPostsPerAccount	Posts to return per account, newest first (1–100 — the public timeline holds 100)
sinceDays	Skip anything older than this many days. 0 means no date limit
onlyNewSinceLastRun	Monitor mode: return only posts that were not there on the previous run
includeRetweets	On by default. Off keeps original posts only
slackWebhookUrl	Optional Slack incoming webhook for a digest of what the run found. Never charged

The 200-account ceiling is what a default 3600-second run finishes comfortably (a healthy account takes about 3.5 seconds). The run also watches its own clock: if X is throttling and accounts start costing retries, it stops starting new ones while there is still time to charge and deliver everything it collected, and finishes with a status saying how many accounts were read and that raising the run timeout — or watching fewer accounts — is the fix. It never gets killed mid-list.

Input examples

Watch three accounts, the 20 most recent posts each:

{
  "handles": ["nasa", "BBCBreaking", "@Space_Station"],
  "maxPostsPerAccount": 20
}

A daily alert feed. Schedule this and every run after the first returns only what is new, with a Slack digest:

{
  "handles": [
    "https://x.com/nasa",
    "@BBCBreaking"
  ],
  "onlyNewSinceLastRun": true,
  "maxPostsPerAccount": 100,
  "slackWebhookUrl": "https://hooks.slack.com/services/T000/B000/xxxx"
}

Original posts only from the last week, no reposts:

{
  "handles": ["nasa"],
  "sinceDays": 7,
  "includeRetweets": false,
  "maxPostsPerAccount": 100
}

How monitor mode works

The first run with "only posts published since the last run" turned on gives you the current picture and remembers the newest post it delivered for each account. Every run after that returns only what came later. Each account keeps its own position in a key-value store named x-account-watchlist-state, one record per account under the key account-<handle> (lower-cased), so adding or removing an account from the list never disturbs the others.

Post ids are compared as 64-bit integers, which is what makes the mark exact. If a run is cut short by a charge limit, the posts it did not deliver are not marked as seen, so they come back as new next time instead of being lost.

If an account has posted more since the last run than your maxPostsPerAccount allows, monitor mode gives you the oldest of those first and the rest on the following run, so nothing is skipped over. In normal (non-monitor) mode the same cap gives you the newest, which is what you want when you are asking for "the latest N".

Things worth knowing before you run it

The public timeline holds up to 100 posts per account and there is no way to page further back on this surface. For a busy account that is a few hours of history; for most accounts it is weeks. Setting maxPostsPerAccount above 100 does not reach further, it just caps at what X serves.

The timeline is not in strict date order. X mixes pinned posts and older high-engagement ones in among the recent ones. Everything this actor returns is sorted newest first by the real posting date, and sinceDays filters on that date, so the feed's own ordering never leaks into your results.

You get the "Posts" tab, which is originals, threads, reposts and quote posts. Replies to other people are not on this surface at all, at any price, so there is no switch for them. Threads are complete: an account replying to itself is kept and flagged with isSelfThread.

X applies a request quota per exit IP on this surface. The actor takes a fresh residential exit for every account in your list, which is why a long watchlist works, and it stops and tells you rather than hammering if the pool has a bad minute. If some accounts come back rate-limited, re-running a few minutes later picks them up.

FAQ

Do I need an X API key or a developer account? No. Nothing here touches the paid X API, and there is no key to supply.

Does this scrape private or protected accounts? No. Only accounts whose posts are public. A protected, suspended, deleted or misspelled handle comes back with status: "unavailable" and is not charged.

How far back can it go? Up to 100 posts per account, which is what the public timeline exposes. There is no deeper history on this surface at any price.

Can I search X for a keyword instead? Not with this actor. Search on X needs a logged-in account, which is a path we do not take. This one watches accounts you name.

Can I get an account's replies to other people? No. The public timeline this reads is the "Posts" tab, and replies to other accounts are not in it. Threads the account writes to itself are included in full.

Can I get the replies under a specific post? No, this reads account timelines. Each post does carry its replyCount and, for a thread, a replyToUrl pointing at the post above it.

Is the engagement count live? It is the count at the moment of the run. Run it again later and you get the updated numbers for the same post.

What is the difference between this and a Twitter scraper? Most of them are bulk search scrapers priced per tweet. This one is built for the watchlist job: a fixed set of accounts, a new-since-last-run delta, per-account status reporting, and a price that stays flat because the proxies are on us.

Why is there a per-account fee as well as a per-post price? Because a monitor run that finds nothing new still has to read every account's timeline. The per-account fee covers that work, and it is what lets the per-post price stay at $0.20 per 1,000 instead of being padded to cover empty runs.

What happens if I put a bad handle in the list? It is reported in the run's status message and in an unavailable account record, and it is not billed. The run still returns everything else. This matters more than it sounds: a handle that does not exist gets a perfectly normal-looking page from X rather than an error, so an actor that does not check for it would charge you for typos.

Related actors

• Telegram Channel Monitor for the same watchlist job on public Telegram channels, with keyword alerts.
• Threads Scraper for Meta's Threads.
• Brand News Monitor for press and news coverage of the same names.
• LinkedIn Hiring Signals Monitor for what those companies are doing on the hiring side.

---

X scraper, Twitter scraper, X account monitor, Twitter account monitor, tweet scraper, X timeline scraper, Twitter timeline to JSON, social listening, X alerts, new tweet alerts, competitor monitoring on X, brand monitoring Twitter, X data export, Twitter API alternative, no API key Twitter scraper.