X (Twitter) Replies Scraper — Thread Context

X (Twitter) Replies Scraper exports visible comments and nested replies from public X posts. Paste one or more post URLs or numeric IDs to get clean reply text, parent-thread context, engagement metrics, media, quoted-post data, and detailed public author profiles.

The Actor follows X's recent reply pages automatically. Users do not need to supply an X login, cookies, API key, or proxy settings—add the posts and click Start.

What can this X replies scraper do?

• Scrape replies from multiple public X or Twitter post URLs in one run.
• Follow reply pages automatically, with no manual cursor or next-page token.
• Distinguish direct comments from nested replies in the same conversation.
• Resolve the parent post, replied-to username, and replied-to display name when X returns the required context.
• Provide clean display text while preserving the full native reply text.
• Export engagement, media, quoted posts, and detailed public author profiles.
• Store each completed page immediately and keep those results if a later page fails.
• Export the Dataset as JSON, JSONL, CSV, Excel, XML, or RSS, or retrieve it through the Apify API.

Why use this Actor?

Capability	How this Actor handles it
Setup	No user-supplied X account, cookies, API key, or proxy configuration
Pagination	Automatically follows the available reply pages for every source post
Thread context	Parent post ID, replied-to user ID, username, display name, and direct/nested classification
Analysis-ready text	Clean reply_display_text plus the complete native reply_text
Reliability	Automatic retries, duplicate removal, and retention of results already collected
Output design	Readable Overview and Detailed views, normalized flat fields, and optional raw X objects
Coverage claim	Reports the visible replies X returns instead of promising inaccessible or deleted comments

This combination is useful when a flat list of comment text is not enough and the relationship between replies matters.

How to scrape X replies and comments

1. Open the Actor's Input tab.
2. Paste one or more X post URLs or numeric post IDs into X post IDs or URLs.
3. Set Maximum replies per post. Keep the default 100, choose another limit, or enter 0 to continue through the available recent replies.
4. Leave raw objects disabled unless your workflow needs X's internal response fields.
5. Click Start, then open the Dataset to preview or export the results.

Input fields

Field	Description	Default
tweets	Up to 50 unique numeric post IDs or full x.com / twitter.com status URLs. Duplicate IDs are requested once, and invalid values do not stop other valid posts.	Required
maxRepliesPerTweet	Maximum unique visible replies per source post. Use 0 to continue until X provides no more new visible replies.	100
includeRawUser	Adds the native reply-author object as raw_user from the existing X response.	false
includeRawReply	Adds the native Tweet object as raw_reply from the existing X response.	false

Example input:

{
  "tweets": [
    "https://x.com/naval/status/2100128104283091247",
    "2097815633828172097"
  ],
  "maxRepliesPerTweet": 100,
  "includeRawUser": false,
  "includeRawReply": false
}

Posts are processed sequentially. A run accepts up to 50 unique valid source posts; split larger batches across multiple runs. More posts and larger reply limits increase runtime and the number of billable results.

When should raw X objects be enabled?

Both raw options reuse data already present in the same TweetDetail response and do not create another X request. raw_user contains the reply author's native user object. raw_reply contains the native Tweet object and normally embeds author data again, so enabling both creates duplication and can make exports several times larger.

Keep both options disabled for normal exports. Enable them for debugging or when a native field that has not been normalized is required. Authentication cookies, tokens, passwords, and proxy details are never included.

What X reply data can you extract?

Reply and conversation fields

• Reply post ID, URL, full text, clean display text, timestamp, language, and posting client.
• Conversation ID, parent post ID, replied-to user ID, username, and display name.
• is_direct_reply, which identifies replies to the requested source post versus nested replies returned in the same conversation.
• Reply, repost, like, quote, bookmark, and view counts when X returns them.
• Hashtags, mentions, expanded URLs, sensitivity status, images, and video variant metadata.

Quoted-post context

• Quoted post ID, URL, text, timestamp, and language.
• Quoted author ID, username, and display name.

Reply-author profile

• User ID, username, display name, bio, location, account creation time, profile URL, avatar, and banner.
• Follower, following, post, media, and like counts.
• Verification, professional categories, affiliation, labels, pinned posts, website, and public profile flags.

Output example

Each successful Dataset row represents one visible reply. The example below is shortened; the Dataset schema contains the complete normalized record.

{
  "status": "success",
  "source_tweet_id": "2097815633828172097",
  "source_tweet_url": "https://x.com/i/web/status/2097815633828172097",
  "reply_tweet_id": "2097842509372092418",
  "reply_tweet_url": "https://x.com/JordanSchachtel/status/2097842509372092418",
  "reply_text": "@FoxNews He spent six weeks at Anthropic, guys",
  "reply_display_text": "He spent six weeks at Anthropic, guys",
  "reply_created_at": "Thu Sep 10 00:20:04 +0000 2026",
  "reply_language": "en",
  "conversation_id": "2097815633828172097",
  "reply_to_tweet_id": "2097815633828172097",
  "reply_to_username": "FoxNews",
  "reply_to_name": "Fox News",
  "is_direct_reply": true,
  "reply_reply_count": 3,
  "reply_retweet_count": 3,
  "reply_like_count": 121,
  "reply_view_count": 3503,
  "reply_media": [],
  "reply_is_quote": false,
  "author_id": "535643852",
  "author_username": "JordanSchachtel",
  "author_name": "Jordan Schachtel",
  "author_followers_count": 302813,
  "author_blue_verified": true,
  "author_profile_url": "https://x.com/JordanSchachtel"
}

The Overview view emphasizes clean text, time, engagement, quote context, author identity, audience size, and clickable links. The Detailed view adds every normalized field. Optional native objects remain available in the complete Dataset record.

The reply_*_count fields describe engagement on the reply, not the source post. reply_display_text applies X's native display range so leading reply mentions do not obscure the comment, while reply_text preserves the complete text returned by X. reply_source_url comes from X's native source field; it is not constructed by the Actor.

How much does it cost to scrape X replies?

This Actor uses pay-per-event pricing. One successfully stored reply produces one reply-result event.

Apify plan	Price per successful reply	Example: 1,000 replies
Free	$0.00035	$0.35
Bronze	$0.00030	$0.30
Silver	$0.00025	$0.25
Gold, Platinum, and Diamond	$0.00020	$0.20

An apify-actor-start event costs $0.00005 per run. On the Free tier, 100 successful replies cost approximately $0.03505, and 1,000 cost approximately $0.35005, including one start event. Platform usage is included in the configured event prices.

Diagnostic rows with invalid_input, not_found, no_results, partial, request_failed, or charge_limit_reached do not create a reply-result charge. Use maxRepliesPerTweet and Apify's Maximum cost per run setting to control spending. The Actor page's Pricing tab is authoritative if prices change.

Pagination, retries, and result limits

Reply pages are followed automatically. If a temporary request problem occurs, the Actor retries the same page up to five times and changes the request route when needed, so a retry does not intentionally skip part of the conversation.

Completed pages are stored immediately, and duplicate reply IDs are removed separately for each source post. Pagination stops when X provides no next page, repeats the same page, reaches the requested result or charge limit, or produces three consecutive pages without a new unique reply. The final log reports how many pages were requested and why collection stopped.

X may occasionally include unrelated posts in a conversation response. The Actor checks that every stored reply belongs to the requested conversation and has a parent post.

Why can the result count be lower than the number shown on X?

X decides which replies are exposed through its recent conversation timeline. Deleted, protected, withheld, spam-filtered, restricted, or otherwise unavailable replies may not be returned. X can also display counts that include content unavailable to the active session.

For that reason, the Actor returns all unique visible replies provided by the paginated X response, not a guaranteed copy of every reply represented by X's displayed counter. Nested replies are included when X returns them in the same conversation timeline.

Diagnostic statuses

Status	Meaning	Reply-result charge
invalid_input	The submitted value is not a numeric post ID or supported X / Twitter status URL; other valid inputs continue normally.	No
not_found	The source post was deleted, unavailable, or does not exist.	No
no_results	The source post exists, but X returned no visible replies.	No
partial	Earlier pages were stored before a later page failed.	No charge for the diagnostic row; stored replies remain billable
request_failed	The first page failed before a reply was stored.	No
charge_limit_reached	The run stopped at the configured maximum cost.	No

X replies scraper API and integrations

Use the Apify API to start a run from Python, JavaScript, a backend service, or any HTTP client:

curl -X POST \
  "https://api.apify.com/v2/acts/arjun_code~x-twitter-replies-scraper/runs?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "tweets": ["https://x.com/naval/status/2100128104283091247"],
    "maxRepliesPerTweet": 100
  }'

Runs can be automated with Apify schedules and webhooks. Results can be sent to Google Sheets, Make, Zapier, cloud storage, databases, or another system through the Dataset API and Apify integrations.

Frequently asked questions

Do I need an X account, cookies, or an API key?

No. Users provide only public post URLs or IDs. Proxy routing and request credentials are handled inside the Actor.

Can it scrape replies from several X posts in one run?

Yes. Add multiple URLs or IDs to tweets. Every source post has its own pagination state, reply-ID deduplication set, and result limit.

Does it collect direct comments and replies to other comments?

Yes, when X returns them in the same conversation timeline. Use is_direct_reply, reply_to_tweet_id, reply_to_username, and reply_to_name to reconstruct the visible relationship.

Can I scrape every reply to a post?

Set maxRepliesPerTweet to 0 to continue through the available recent replies. However, no scraper can return replies that X does not expose. See the result-limit explanation above.

Can I export X comments to CSV or Excel?

Yes. Open the run's Dataset and choose JSON, JSONL, CSV, Excel, XML, or RSS. You can also retrieve records through the Dataset API.

Is scraping X replies legal?

This Actor processes publicly exposed post and profile data. Whether a particular use is lawful depends on the data, jurisdiction, and purpose. Do not collect or use personal data unlawfully, and follow X, Apify, and applicable privacy requirements.

Related X (Twitter) Actors

Actor	Best for
X (Twitter) Media Downloader	Post details plus downloadable images, videos, GIFs, and X Articles
X (Twitter) Retweeters Scraper	Users who reposted or quoted a post, with quote-post context
X (Twitter) Followers & Followings Scraper	Complete visible follower or following lists with profile data
X (Twitter) People Search Scraper	Public X profiles matching keywords or search queries

Privacy and support

Internal cookies, proxy details, and authentication data are never written to the result Dataset. This independent Actor is not affiliated with, endorsed by, or sponsored by X Corp.

If a run fails or X changes its response format, open the Actor's Issues tab and include the Apify run ID, input, and expected behavior. Never include passwords, cookies, tokens, or other secrets.