YouTube transcript panel pull
Use this skill when you need the words from a YouTube video as text: your own episode or webinar to repurpose, or a prospect's talk or podcast appearance to prepare for a call. Produces one timestamped .txt transcript per video. "Get the transcript of this video", "pull captions from my channel", "what did they say on that podcast".
R. Hunter HarrisFounder, Hunter Software ConsultingApplies when someone needs a YouTube video's spoken words as text. Produces one timestamped transcript file per video.
The reliable route is the transcript panel on the watch page, read by the agent in a browser. It works on public videos without signing in.
The play
- Collect the video IDs. For one video, take the
v=value from the URL. For a channel, open its Videos tab and collect the IDs from the page links (snippet inreferences/browser-snippets.md). Skip Shorts and IDs you already have. - Open each watch page. Pause the video, expand the description, then click "Show transcript". The button lives inside the description, so it does not exist until the description is expanded.
- Wait a few seconds, then read the transcript rows. Each row has a timestamp and a text span. Join them as
[m:ss] text. - Get the text off the page through a temporary textarea and a copy command. Return the row count and the first and last rows, check the clipboard matches, then write the file. This keeps a long transcript out of the agent's context.
- Zero rows usually means the panel hasn't loaded; it can also mean no captions or changed markup. Retry once, then report which; never save an empty file.
- Save one file per video with a short header: video ID, title, language and whether the captions are auto-generated. Read the language and caption type from the page; don't assume them.
- When pulling a batch, check that the counts agree: IDs collected, files written, unique IDs on disk.
Open references/browser-snippets.md for the exact selectors and the copy snippet.
What good looks like
- First check: the row count. A real 30-minute episode returns hundreds of rows; a handful means the panel was still loading, or you grabbed a chapter list instead.
- The mediocre version calls YouTube's caption endpoint directly. Without a session token it now returns an empty body, so the agent reports success on blank files.
- The other mediocre version pastes the whole transcript into the chat to save it. It works once, then burns the agent's context on the second video.
- The output is right when every file has rows, the batch counts agree, and a spot-checked timestamp in the file matches the video at that moment.
Rules
- MUST read the transcript panel, not the caption endpoint.
- MUST refuse to save a file with zero rows.
- MUST label auto-generated captions as auto-generated; speakers are not identified in them.
- NEVER quote a transcript line as a person's exact words without checking it against the audio. Auto-captions mishear names, numbers and jargon.
- NEVER pull a private or members-only video the reader has no right to.
references/browser-snippets.md
Browser snippets
Run these in the page with the agent's JavaScript tool. YouTube changes its markup from time to time. If a selector returns nothing, inspect the page and update it rather than falling back to the caption endpoint.
Collect video IDs from a channel's Videos tab
Scroll to the bottom first if the channel has more videos than the first screen loads.
[...new Set([...document.querySelectorAll('a[href*="/watch?v="]')]
.map(a => a.href.match(/v=([\w-]{11})/)?.[1]).filter(Boolean))]
Open the transcript panel
document.querySelector('video')?.pause();
document.querySelector('tp-yt-paper-button#expand')?.click();
document.querySelector('ytd-video-description-transcript-section-renderer button')?.click();
Read the rows and copy them
Wait a few seconds after opening the panel.
const rows = [...document.querySelectorAll('transcript-segment-view-model')].map(e => {
const ts = e.querySelector('[class*=Timestamp]')?.textContent.trim();
const tx = e.querySelector('span[role=text]')?.textContent.replace(/\s+/g, ' ').trim();
return `[${ts}] ${tx}`;
});
const ta = document.createElement('textarea');
ta.value = rows.join('\n');
document.body.appendChild(ta);
ta.select();
document.execCommand('copy');
ta.remove();
rows.length;
navigator.clipboard.writeText is often refused in an agent-driven browser because the page doesn't have focus. The textarea route works there.
On a Mac, write the file with pbpaste > <video-id>.txt. Elsewhere, use the platform's paste command, or have the agent return the text in chunks and write it.
File header
YouTube transcript
Video ID: <id>
Title: <title>
Language: <from the panel's language menu>
Captions: auto-generated | creator-provided
[0:00] ...
Failure signs
| You see | It means | Do |
|---|---|---|
rows.length is 0 | the panel hasn't rendered | wait 3 to 5 seconds and rerun |
| no transcript button | description not expanded, or the video has no captions | expand and retry once, then record "no captions" |
| a few rows with chapter titles | you read the chapters list | check the selector is transcript-segment-view-model |
| caption URL returns an empty body | the endpoint needs a token | use the panel |