TikTok Explore Video Scraper

TikTok Explore Video Scraper reads the public TikTok Explore page and returns a bounded sample of source-backed video cards exposed by that page. It preserves public titles or captions, creator identity, hashtags, visible metrics, thumbnails, canonical video URLs, and observed position. It does not log in, access private accounts, or turn a blocked page into an empty successful result.

Get started

1. Open the Actor Store page and select Try for free or open the Input tab.
2. Set limit to the number of public video cards you want, up to 20.
3. Keep the default limits for the first run. Enable includeDiagnostics only if you want a single minimal row when the page is blocked or empty.
4. Start the Actor, open the dataset, and export records as JSON, CSV, or Excel. Open Run summary (OUTPUT_SUMMARY) to see the access outcome and transport.

Input

All fields are optional; the defaults below are ready to use. Unsupported fields and out-of-range values are rejected before requests start.

Field	Type	Required?	Default	Meaning
limit	integer	Optional	10	Maximum distinct public video records to emit, from 1 to 20.
includeDiagnostics	boolean	Optional	false	Emit one minimal {url,error,errorCode,scrapedAt} row when the public page is blocked or empty.
requestTimeoutSecs	integer	Optional	20	Bounded timeout for the public page request, from 10 to 120 seconds.
maxRequestRetries	integer	Optional	1	Bounded retry count for the direct public page request, from 0 to 2.
proxyConfiguration	object	Optional	Not supplied	Optional Apify Proxy or custom HTTP(S) proxy. Proxy use does not bypass access controls.
fixturePath	string	Optional	Not supplied	QA-only relative path below test/fixtures, up to 240 characters. Fixture mode is explicit and is never used as a live fallback.

Full default input, excluding the optional proxy object:

{
  "limit": 10,
  "includeDiagnostics": false,
  "requestTimeoutSecs": 20,
  "maxRequestRetries": 1
}

The Actor requests the public Explore URL once and parses only the video cards present in that response. It does not scroll, paginate, sign in, or use a personalized recommendation endpoint.

Output

The dataset contains one row per source-backed video card, or one minimal diagnostic row when includeDiagnostics is enabled and the page is blocked or empty. In the saved 2026-10-25 local run, TikTok returned an access challenge and includeDiagnostics was false, so no rows were written. The saved dataset is an empty array:

[]

The fields below describe the row shape when public data is available; none was observed in that run. Optional values are omitted when the page does not expose them, and no placeholder rows are written.

Field	Type	Availability	Meaning
videoId	string	When published	Public TikTok video identifier.
videoUrl	string	When published	Public TikTok video URL.
videoTitle	string	When published	Public title or caption shown for the video.
creatorName	string	When published	Public creator display name or handle.
creatorUrl	string	When published	Public creator profile URL.
description	string	When published	Public video description or caption.
hashtags	string array	When published	Hashtags visible in the public caption or structured card data.
durationSeconds	integer	When published	Public duration in seconds when exposed.
durationText	string	When published	Public duration text when exposed.
playCount	integer	When published	Public play or view count when exposed.
likeCount	integer	When published	Public like count when exposed.
commentCount	integer	When published	Public comment count when exposed.
shareCount	integer	When published	Public share count when exposed.
metricText	string	When published	Source metric text when it cannot be safely converted to an integer.
thumbnailUrl	string	When published	Public video thumbnail URL.
source	string	When published	Provenance source label.
sourceUrl	string	When published	Public Explore URL that produced the record.
pageNum	integer	When published	Observed source page number; Explore is a single bounded page.
position	integer	When published	Position in the observed Explore batch.
scrapedAt	string	When published	UTC record timestamp.
url	string	In a diagnostic row	Requested URL in a minimal diagnostic row.
error	string	In a diagnostic row	Bounded diagnostic message.
errorCode	string	In a diagnostic row	Stable diagnostic code, such as ACCESS_BLOCK.

A diagnostic row contains only url, error, errorCode, and scrapedAt. It is written only when includeDiagnostics is true; the saved run kept the default false, which is why the dataset is empty.

Run summary

Run summary is stored separately from the dataset in the run's key-value store under OUTPUT_SUMMARY. It records the access outcome, transport, response format, and retry attempts even when no dataset row is written.

Field	Meaning
source	Source label recorded for the run, tiktok.com.
sourceUrl	Public Explore URL requested.
status	Run outcome: success, blocked, empty, or failed.
normalCount	Number of video records saved.
diagnosticCount	Number of diagnostic dataset rows saved; zero unless includeDiagnostics is enabled.
responseFormat	Detected page format, such as embed_state_json, html, or unknown.
transport	Transport used for the request, such as direct_http, proxy, or apify_proxy.
readerFallback	Whether a reader fallback was used.
responseStatus	HTTP status of the last response, when one was received.
responseBytes	Size of the last response body in bytes, when reported.
contentType	Content type of the last response, when reported.
accessSignal	Access-challenge signal detected in the response, when present.
attempts	Array of route attempts with status, transport, size, and outcome.
completedAt	Time the summary was written, in ISO UTC.

Optional response fields appear only when the route reports them; the saved run did not expose responseStatus, responseBytes, contentType, or accessSignal.

Status	Meaning
success	Public video cards were saved without an extraction problem.
blocked	TikTok returned an access challenge or denial, so no video cards were saved.
empty	The page loaded without an access challenge but exposed no usable public video cards.
failed	A request or parse failure occurred before any usable record was saved.

Example OUTPUT_SUMMARY from the saved 2026-10-25 run:

{
  "source": "tiktok.com",
  "sourceUrl": "https://www.tiktok.com/explore",
  "status": "blocked",
  "normalCount": 0,
  "diagnosticCount": 0,
  "responseFormat": "unknown",
  "transport": "apify_proxy",
  "readerFallback": false,
  "attempts": [
    {
      "route": "direct_public_page",
      "transport": "apify_proxy",
      "statusCode": 200,
      "responseBytes": 369235,
      "contentType": "text/html; charset=utf-8",
      "outcome": "blocked",
      "error": "HTTP 200 (captcha)"
    }
  ],
  "completedAt": "2026-10-25T21:32:40.650Z"
}

Coverage and limitations

Explore is a single public page, not a complete or personalized feed. The Actor captures only the cards present in the response it receives, up to limit; it does not scroll, paginate, or follow creator profiles. Position is the order observed in that response and can change between requests.

TikTok can respond with a CAPTCHA, login wall, empty shell, or generic error. Those outcomes stop the run and are recorded in OUTPUT_SUMMARY; when includeDiagnostics is enabled, the same outcome is also written as one minimal dataset row with url, error, errorCode, and scrapedAt. A blocked response is never converted into a successful empty result. fixturePath exists only for offline QA and is never used as a live fallback. Proxy use does not bypass access controls.

Usage and support

Cost depends on Apify compute, request time, dataset size, and any proxy traffic. Keep limit small, start with the default single request, and inspect OUTPUT_SUMMARY before scheduling repeated runs. See the Actor's Pricing tab for current account rates.

Use the Actor Issues page to report a problem. Include the run ID, the input used, and the run summary; do not include tokens, cookies, or private data. This Actor is not affiliated with, endorsed by, or sponsored by TikTok. Use it only for public data and in accordance with the source's terms, robots guidance, and applicable law. Public pages can contain personal information; use the output only for lawful purposes.