Study a Creator's Entire Channel in an Afternoon

The feature everyone asks for, described honestly

On 4 September 2026 we read the channel-import code that is actually deployed — the two edge functions that run when you paste a YouTube channel into BrainTube — rather than the page that describes them. The same day we read Rescroll's homepage, which sells this exact workflow. Both products do the job. The honest version of the job is less dramatic than either of us would like to write.

A channel study costs you an afternoon, a picker and a few pauses. What you get back at the end is worth more than the import, and almost nobody talks about that half.

Step one: resolving the channel

You paste a URL. The deployed youtube-fetch-channel-videos function accepts several shapes of it, and we know the list precisely because the function's own error string enumerates them: /@handle, /channel/UC..., /c/Name, /user/Name, and /playlist?list=PLxx. A handle, a channel ID, a custom name, a legacy username and a plain playlist link all land in the same place.

Whichever you paste, it resolves to the channel's uploads playlist and returns the first page of that playlist: up to 50 videos, each with title, thumbnail, duration and published date.

Two consequences are worth stating plainly rather than burying. First, 50 is a page, not a catalogue. A creator with four hundred uploads hands you their recent fifty through this door, so what you are building is a study of where they are now, not a full history; for the back catalogue you will be resolving playlists by hand. Second, videos already in your library and videos you have previously excluded come back flagged, so the picker can grey them out. That matters most on the second visit: come back to the same channel in November and you see what is new, rather than trying to remember what you took in September.

Step two: the picker, and the arithmetic of it

You select. Then the deployed youtube-bulk-import function takes the selected video IDs — up to 10 per call, rate-limited to 3 bulk imports per hour. Plan caps on bulk YouTube import are Free 10, Starter 50, Pro 150 and Pro+ 400.

Do the arithmetic and the marketing evaporates in a useful way. Fifty videos is five calls; at three calls an hour, that is two sittings rather than one. It is an afternoon with a list in front of you, and the selecting is the part that decides whether the corpus is any good. A study built from fifty videos you chose is a different object from one built from the fifty that happened to be most recent. We would rather say that than claim a catalogue lands in one click, because it does not.

The half that actually matters

Import is plumbing. The reason to spend an afternoon on it is that a body of work supports questions a single video cannot answer.

Recurrence. Which claim does this person make in nearly every video? One video tells you they said it. Forty tells you it is load-bearing, restated in different words each time. Search across the corpus (search_knowledge, deep_search) turns a hunch into a count.

Change of mind, and when. Constrain the same query by date — search_by_date_range exists for this — and run it against the early uploads and the late ones separately. Where the answers diverge you have found the month something shifted, which is the least visible thing from any single video.

Adjacency. get_related answers the question "what else sits near this claim in their work", which is how you find the video where they qualified something they stated flatly somewhere else.

Absence. What does this person never address? Absence is the hardest thing to establish from a corpus, because a search that returns nothing looks identical to a search you phrased badly. Treat a null result as a prompt to rephrase twice more before writing it down as a finding. No tool, ours included, tells those two cases apart for you.

Every answer resolves to a specific item you can open and check, which is the whole point. A summary you cannot trace back to the video it came from is a rumour with better typography.

What the neighbours ship

Rescroll — a creator-focused tool whose homepage we read on 4 September 2026 — sells this workflow directly and well. Their copy: "Decode an entire niche in an afternoon. Paste any YouTube channel, pull 50 videos in one click. Map a competitor's catalog before lunch. Nobody else does this." Their FAQ puts the same number on it: "Bulk YouTube channels: import up to 50 videos from any channel in one go." They shipped it, and they describe it in plain language.

The exclusivity claim is theirs, not ours to endorse or attack; a reader can see that two tools now do a fifty-video channel import. One thing we will note, because it decides whether this workflow is usable outside one app: the string "MCP" appears zero times on their page. If you want to interrogate the catalogue from Claude Code or any other client, that absence is the whole story.

Recall's users have been asking for the subscription version of this for a while — their most-voted open feature request is subscribe-to-channels with automatic summarising, at 625 votes, under review. Nobody has fully answered it, us included: BrainTube has no channel subscription that watches for new uploads. You come back and re-run the import.

Where BrainTube stands

What works today. A channel URL in any of the five documented shapes resolves to the uploads playlist. The first page of up to 50 videos comes back with titles, durations and dates, with library duplicates and prior exclusions flagged so the picker can grey them out. Selected videos import and their text is indexed, so the corpus is searchable rather than merely stored. deep_search and get_related handle the cross-cutting questions; search_by_date_range handles the change-of-mind ones. export_corpus gives the whole thing back as a ZIP of Markdown with a signed download URL, so an afternoon's work is not trapped. All 44 tools on the one read-write MCP server at https://mcp.brain-tube.com/mcp are available on every plan, Free included, which means the interrogation can happen from a client you already use rather than only in our app.

What does not work. The first page of 50 is a hard boundary — there is no "fetch page two" in the deployed function, so a long-running channel gives you its recent uploads and nothing older. Imports run 10 video IDs per call and 3 calls per hour, so a fifty-video study spans two hours whether or not you are in a hurry. The bulk caps are real gates, not metering: Free stops at 10, Starter 50, Pro 150, Pro+ 400. Our YouTube sync covers playlists and Liked Videos, and that is genuinely all it is — not watch history, not Watch Later, and we will not imply otherwise. There is no channel subscription that notices new uploads for you. And there are no native iOS or Android apps; the mobile experience is add-to-home-screen. If your afternoon of channel study is meant to happen on a train, that is the constraint to weigh first.

The playlist import walkthrough covers the mechanics from the other direction if you already have a playlist rather than a channel.

How these figures were checked

The channel-resolution formats, the 50-video first page, the duplicate and exclusion flags, the 10-IDs-per-call limit and the 3-imports-per-hour rate limit were read on 4 September 2026 from the youtube-fetch-channel-videos and youtube-bulk-import functions as deployed in production, not from documentation about them. Bulk-import plan caps and the MCP tool count of 44 were read the same day, the tool count from https://brain-tube.com/connect.md.

Rescroll's homepage and FAQ copy, and the absence of "MCP" from that page, were read live on 4 September 2026; their dollar prices did not appear in the server-rendered HTML, so we have not quoted any. Recall's most-voted feature request and its 625-vote count are as of the packet date, 4 September 2026, not re-verified live — their feature board renders its posts in JavaScript and the fetch returned navigation only.

If you spot something here that has gone stale, tell us and we will correct it with the date attached.