YouTube Scraper

Scrape YouTube videos, channels, and metadata without an API key. Search by keyword or scrape a full channel's video catalogue. Returns structured records with video ID, title, view count, channel, duration, publish date, and thumbnail URLs.

What it does

The actor uses YouTube's InnerTube REST API — the same JSON API used by YouTube's own web client. No API key, no browser rendering. Requests run on standard datacenter infrastructure, keeping costs low.

Two scraping modes are available:

• Search mode (mode: search) — submit one or more keyword queries and collect matching video results, paginated via InnerTube continuation tokens.
• Channel mode (mode: channel) — supply a channel URL, handle, or bare channel ID and collect all videos from that channel's Videos tab.

Input

Field	Type	Description
mode	string	search (default) or channel
queries	array	Keywords to search for (search mode only)
channelUrl	string	Channel URL, @handle, or UC... ID (channel mode only)
maxItems	integer	Maximum videos to return per query / total for channel mode

Channel URL formats accepted:

• https://www.youtube.com/@MrBeast
• https://www.youtube.com/channel/UCX6OQ3DkcsbYNE6H8uQQuVA
• Bare ID: UCX6OQ3DkcsbYNE6H8uQQuVA

Output

Each record contains:

Field	Type	Description
videoId	string	YouTube video ID
videoUrl	string	Full watch URL
title	string	Video title
description	string	Description snippet (if shown in search results)
channelId	string	YouTube channel ID (UC...)
channelName	string	Channel display name
channelUrl	string	Channel canonical URL
viewCount	integer	Total view count
likeCount	integer	Like count (null if hidden)
commentCount	integer	Comment count (null if unavailable)
durationSeconds	integer	Duration in seconds
publishedAt	string	Relative publish date (e.g. "1 year ago")
thumbnailUrl	string	Highest-resolution thumbnail URL
tags	string	Comma-separated tag list (if available)
isLive	boolean	Whether the video is a live stream
categoryId	string	Video category name
query	string	The search query or channel URL that produced this record
scrapedAt	string	ISO timestamp when the record was collected

Example: search mode

{
  "mode": "search",
  "queries": ["python tutorial", "javascript crash course"],
  "maxItems": 50
}

Example: channel mode

{
  "mode": "channel",
  "channelUrl": "https://www.youtube.com/@MrBeast",
  "maxItems": 100
}

Performance and cost

• Memory: 256 MB
• Approximate throughput: 20 videos per InnerTube page, ~200 ms delay between pages
• Pricing: pay-per-result (DATA_RECORD events)

Notes

• publishedAt returns YouTube's relative format ("3 days ago", "1 year ago") rather than an absolute ISO date — the InnerTube browse/search API does not expose absolute timestamps in these endpoints.
• likeCount and commentCount are not available from the InnerTube browse/search endpoints and will always be null.
• description is populated only when YouTube includes a snippet in search results; it is typically null in channel mode.
• The actor handles YouTube's EU consent mechanism automatically via a session cookie header.