Extract YouTube video transcripts and subtitles via YouMind API — no yt-dlp, no proxy, no local dependencies.
Batch extract up to 5 videos at once with parallel processing.
Saves videos to your YouMind board with timestamped transcripts in markdown.
Works from any IP (cloud, VPS, CI/CD, corporate networks).
Use when user wants to "get YouTube transcript", "extract video subtitles",
"transcribe YouTube video", "batch transcribe videos", "get video captions",
"summarize YouTube video", "YouTube 字幕", "YouTube 文字起こし", "YouTube 자막",
or "download YouTube transcript".
Extract YouTube video transcripts and subtitles via YouMind API — no yt-dlp, no proxy, no local dependencies.
Batch extract up to 5 videos at once with parallel processing.
Saves videos to your YouMind board with timestamped transcripts in markdown.
Works from any IP (cloud, VPS, CI/CD, corporate networks).
Use when user wants to "get YouTube transcript", "extract video subtitles",
"transcribe YouTube video", "batch transcribe videos", "get video captions",
"summarize YouTube video", "YouTube 字幕", "YouTube 文字起こし", "YouTube 자막",
or "download YouTube transcript".
Batch extract YouTube video transcripts with timestamps — up to 5 videos at once, no yt-dlp, no proxy, no local setup. Videos are saved to your YouMind board and transcripts are output as clean markdown.
Why YouMind? Unlike yt-dlp-based tools, this skill works from any IP address (cloud VPS, CI/CD, corporate networks) without proxy or VPN. YouMind handles the extraction server-side. And batch mode means you can process multiple videos in one go.
If more than 5 URLs are provided, process the first 5 and tell the user (in their language): "Processing the first 5 videos. Please submit the remaining ones in a follow-up message."
Extract id as materialId from the response. Build the YouMind link:
https://<endpoint>/boards/<boardId>?material-id=<materialId>&utm_source=youmind-youtube-transcript (endpoint = youmind.com or preview.youmind.com). Do NOT use /material/<id> — that URL does not work.
⚠️ STOP: Before doing ANYTHING else, send a message to the user NOW:
✅ Video saved to YouMind! View here: <link>
Extracting transcript, usually takes 10-20 seconds...
(Use the user's language. The above is a Chinese example.)
Do NOT proceed to Step 4 until you have sent this message. The user needs something to click while waiting.
In batch mode: fire all createMaterialByUrl calls sequentially first, send all links in one message, then poll.
Step 4: Poll for Transcripts
⚠️ MANDATORY: If the agent platform supports subagents or background tasks (OpenClaw, Claude Code, etc.), spawn a subagent/background task for the polling + extraction work. Return control to the user immediately. The subagent sends the result when done. See references/long-running-tasks.md for the full pattern.
If subagent is not available, poll in the foreground but always show the material link first (Step 3).
In batch mode: poll all materials in a round-robin loop. Each iteration, check all pending materials. Remove from the pending list once resolved.
Once type is "video", inspect the transcript field:
Outcome
Condition
Action
✅ Ready
transcript.contents[0].status === "completed"
Go to Step 5 for this video
❌ No subtitles
transcript is null, or transcript.contents is empty
Tell user: "[Video Title] does not have subtitles. Transcript extraction is not supported for this video." Link: https://<endpoint>/boards/<boardId>?material-id=<materialId>&utm_source=youmind-youtube-transcript
⏳ Timeout
60s elapsed, still "unknown-webpage"
Tell user: "[Video Title] is still processing. Check later at https://<endpoint>/boards/<boardId>?material-id=<materialId>&utm_source=youmind-youtube-transcript"
Replace <YOUTUBE_URL> with the actual YouTube URL. The script writes a markdown file and prints the filename and YouMind link.
This command does everything in one step: parse JSON, extract fields, format markdown, write file, and print summary.
File naming: transcript-<video-title-slug>.md — derived from the video title, not the video ID. Examples: transcript-never-gonna-give-you-up.md, transcript-一口气了解韩国经济.md.
⚠️ MANDATORY: Send the transcript file as an attachment. The transcript is too long to display inline. Always write the file first, then send it as an attachment (use the platform's file upload capability). Include a brief summary message alongside the file — title, language, word count. Do NOT paste the entire transcript as text in the chat.
In batch mode, send each transcript file as a separate attachment, then show a final summary table:
When you receive a 402 error (codes: InsufficientCreditsException, QuotaExceededException, DailyLimitExceededException, LimitExceededException), immediately show this message (translated to user's language):
You've reached your free plan limit. Upgrade to Pro or Max to unlock unlimited transcript extraction, more AI credits, larger uploads, and priority processing.
