YouTube Shorts Finder Pro — Search, Channels & Hashtags

Find any YouTube Short by channel, keyword, hashtag, or direct URL — with views, displayed likes and comment counts, hashtags, plus a viral score and engagement rate on every row. No login, no API key, no quotas.

What it does

YouTube Shorts are where the views are, but YouTube search and channel pages make Shorts discoverable. This actor scrapes Shorts in 4 input modes with per-line auto-detection:

1. Channel — @handle, youtube.com/channel/…, /c/… or /user/… URLs. Reads the channel's Shorts tab (with pagination).
2. Keyword — any YouTube search (cooking tips), filtered to the Shorts shelf (with pagination).
3. Hashtag — #asmr and friends, resolved through YouTube search so you get real Shorts, not long-form videos.
4. Direct URL — paste a youtube.com/shorts/<id> (or watch?v=, youtu.be/) link to enrich one specific video.

Every Short comes back with: video id, URL, title, full description, hashtags, view counts and displayed like/comment counts, duration in seconds, ISO publish date, thumbnail, caption availability, comments-disabled flag, channel data (id, name, URL, subscribers, verified badge) — plus two computed metrics:

• viralScore = viewCount / channelSubscribers — how many times over its own audience a Short went. A score above 1.0 means it broke out of the channel's subscriber base.
• engagementRate = (likeCount + commentCount) / viewCount — likes and comments per view.

Optional top comments per Short (author, text, likes, replies, publish time).

How to use it

1. Add one entry per line in Channels, keywords, hashtags or Shorts URLs: channels (@MrBeast), plain keywords (cooking tips), hashtags (#asmr), or Shorts URLs.
2. Set Max Shorts per entry (maxResults, 1–1000) and the global cap (maxItems) to control spend.
3. Optionally filter by date (Published after accepts 2026-01-01 or relative values like 7 days, 2 weeks, 1 month — free, no extra charge), pick the sort order (newest, popular, oldest), and enable top comments.
4. Run. Results land in the default dataset; comments land in the same dataset with type: "comment". Failed inputs are reported as typed error items (CHANNEL_NOT_FOUND, NO_SHORTS, DATE_FILTER_EMPTY, VIDEO_UNAVAILABLE) in the default dataset and summarized in the ERRORS key-value record — the run never dies because of one bad line.

Example input

{
  "maxResults": 5,
  "sortOrder": "newest",
  "startUrls": ["@MrBeast"]
}

Example output (real run)

{
  "videoId": "aNI2jv7lBPQ",
  "url": "https://www.youtube.com/shorts/aNI2jv7lBPQ",
  "title": "Giving $10,000 To A Random Subscriber",
  "description": null,
  "hashtags": [],
  "viewCount": 19211316,
  "likeCount": 604000,
  "commentCount": 14000,
  "durationSeconds": 40,
  "publishedAt": "2026-10-09T16:00:01.000Z",
  "thumbnail": "https://i.ytimg.com/vi/aNI2jv7lBPQ/hqdefault.jpg",
  "captionsAvailable": true,
  "commentsDisabled": false,
  "viralScore": 0.03687392706333973,
  "engagementRate": 0.03216854066634477,
  "channelId": "UCX6OQ3DkcsbYNE6H8uQQuVA",
  "channelName": "MrBeast",
  "channelUrl": "https://www.youtube.com/@MrBeast",
  "channelSubscribers": 521000000,
  "channelVerified": true,
  "source": "channel",
  "query": "@MrBeast",
  "scrapedAt": "2026-10-12T07:21:49.240Z"
}

Use cases: find viral Shorts in your niche, track competitors' Shorts output, harvest proven hooks and hashtags, measure which topics break out (viralScore > 1), or enrich a list of Shorts URLs with public stats.

Pricing

Pay per event — you only pay for data you receive:

Event	Price (USD)	What counts
short	$0.002	Each Short with full metadata, viral score and engagement rate
comment	$0.0004	Each top comment (only when top-comments mode is on)

Typical runs:

• 100 Shorts from one channel ≈ $0.20
• 1,000 Shorts across 10 keywords ≈ $2.00
• 50 Shorts with 20 comments each (50 shorts + 1,000 comments) ≈ $0.50

Date filtering, sorting, deduplication and error reporting have no separate event charge.

How it compares

Channel-only Shorts scrapers exist (including the official one). This actor differs in measurable ways:

	This actor	Typical channel-only scrapers
Inputs	Channels, keywords, hashtags, direct URLs	Channels only
Discovery	Search YouTube for any topic/hashtag	Limited to channels you already know
Date filter	Included free	Often a separately billed extra
Derived metrics	viralScore + engagementRate per row	Raw counts only
Price per Short	$0.002	Up to 2× more
Engine	Plain HTTP, no browser (fewer failure points)	Often headless browsers

Metadata is parsed from YouTube's public pages; unavailable fields are null. Rounded counts (likes, comments or subscribers shown as 2.6K by YouTube itself) are returned as their numeric value (2600).

FAQ

Is this legal / against YouTube's terms? The actor only reads public pages (the same HTML any logged-out browser receives). It needs no login, no API key, and optional comment rows include public author names, channel IDs and avatars. Respect YouTube's Terms of Service for your use case.

How accurate are the numbers? Views come from the public watch page. Likes, comment counts and subscriber counts may be rounded by YouTube (for example, 14K comments becomes 14000). Derived metrics use those displayed numbers. Missing counts are null.

How is pagination handled? Channel tabs, search shelves and hashtag results are paginated with continuation tokens until maxResults per entry or the global maxItems cap is reached.

Which region are results from? Requests run with US/English locale. YouTube still personalizes some ordering by proxy egress region; search/hashtag modes follow YouTube relevance order unless you sort newest/popular/oldest after collection.

Why did an entry return an error item? CHANNEL_NOT_FOUND (bad handle/URL), NO_SHORTS (valid page, no Shorts found), DATE_FILTER_EMPTY (Shorts exist but all are older than your filter — loosen publishedAfter), VIDEO_UNAVAILABLE (deleted/private video). Other lines still run.

Do I need proxies? YouTube blocks datacenter IPs, so the actor defaults to Apify Residential proxies (changeable in the input). It retries with exponential backoff on 429/5xx and keeps going when a single video fails.

Comments? Turn on Include top comments to get up to maxComments (1–200) top-level comments per Short, billed as comment events in the default dataset. Videos with disabled comments are flagged (commentsDisabled: true, commentCount: null).

Integrations

• Apify API: run via POST /v2/acts/<actor-id>/runs, read defaultDatasetId items; comments in the same dataset.
• Make / Zapier / n8n: use the Apify integration, trigger on dataset items.
• AI agents (MCP): the dataset has a table view (overview) and a fixed schema, so agents can query Shorts by views, viral score or hashtag directly.

Support

Found a page layout this actor misreads, or hitting blocks at scale? Open a run with a minimal input that reproduces it (one line in startUrls) and describe what you expected — layout fixes ship fast because parsing is defensive and covered by tests.