YouTube Transcript

@YouTube Transcript fetch the transcript for https://youtube.com/watch?v=dQw4w9WgXcQ and summarize the main topics covered in the first 5 minutes

The MCP retrieved the video's closed captions and stored the transcript in your account. The first 5 minutes cover an introduction to the speaker's background in software architecture, followed by a discussion of microservices trade-offs—specifically latency versus maintainability—and a brief case study from a 2019 migration project. The speaker emphasizes that distributed systems require explicit error-handling contracts between services.

This example shows the read-and-reason workflow: GET_VIDEO_TRANSCRIPT pulls captions (consuming provider credits for uncached videos), then the AI synthesizes the content. The tool waits up to 20 seconds; longer videos may time out and require polling GET_VIDEO_TRANSCRIPT_JOB_STATUS. ASR is never triggered—only existing captions are retrieved.

@YouTube Transcript list all transcripts I've stored in the last 30 days, filtered to videos longer than 20 minutes

Your account holds 14 transcripts from the past month matching that duration threshold. Results include a 47-minute conference talk on Kubernetes networking (stored Jan 12), a 38-minute product demo (Jan 9), and a 22-minute tutorial on React hooks (Jan 6). Each entry shows video title, channel name, duration, storage timestamp, and available language codes. The list is paginated; 11 additional matches are available on the next page.

LIST_STORED_TRANSCRIPTS demonstrates the MCP's search and inventory capabilities without consuming new credits—it only reads what you've already fetched. Pagination means large result sets require multiple calls. This is useful for auditing usage or finding transcripts to reuse across Projects without re-fetching.

@YouTube Transcript get transcripts for these 6 tutorial videos in batch, then draft a combined study guide covering the shared concepts across all of them

The MCP queued a batch job and retrieved captions for all 6 videos (4 were cached; 2 consumed new credits). Shared concepts identified: dependency injection appears in 5 videos, test-driven development in 4, and CI/CD pipeline setup in 3. The draft study guide groups these themes, cross-references timestamps where each instructor explains the same pattern differently, and flags one video that uses outdated tooling (Docker Compose v1 syntax) as a point of caution.

GET_VIDEO_TRANSCRIPTS_BATCH handles 1–10 videos at once, making it efficient for curriculum analysis or competitive research. Cost scales with uncached requests, so batch size affects your monthly budget. The 20-second wait may time out on large batches; use GET_VIDEO_TRANSCRIPTS_BATCH_STATUS to poll completion if needed.

When batch transcript retrieval beats manual note-taking

A three-person content team reviews 15-20 competitor YouTube videos every Monday to spot trends and extract quotes for their own scripts. Without this MCP, someone watches at 1.5x speed and takes notes. With it, they run Get Video Transcripts Batch on Sunday night, then search the stored transcripts during standup using List Stored Transcripts. The win is real if the videos average over eight minutes—shorter clips don't justify the setup cost. The MCP consumes provider credits per uncached video, so teams watching the same channels weekly should track whether the cache hit rate justifies the monthly spend. If your research list changes completely each week, the credit burn adds up fast. Buy this if your team watches more than ten videos per week and needs searchable text, not just summaries.

Why this MCP isn't built for internal meeting transcripts

A five-person customer success team records onboarding calls as unlisted YouTube videos and wants to pull transcripts into Switchy for post-call summaries. This MCP can technically fetch those transcripts using Get Video Transcript, but it's the wrong tool. YouTube's caption system adds 30-90 seconds of lag after upload, and the MCP waits up to 20 seconds per job—so you're looking at two minutes minimum before the transcript lands in Switchy. Teams doing this workflow more than twice a week should use a dedicated meeting transcription MCP that hooks Zoom or Google Meet directly. The YouTube Transcript MCP shines when you're analyzing public or competitor content, not your own recordings. If you're already uploading to YouTube for other reasons and can tolerate the delay, it works, but don't choose YouTube as your recording platform just to use this integration.

When channel-level research justifies the Basic tier

A solo YouTube creator in the productivity niche wants to analyze what topics competitors covered in the last quarter without watching 80 videos. They use Resolve Channel Videos to pull upload metadata for three competitor channels, then run Get Video Transcripts Batch on the 15 most-viewed videos from each. The metadata step requires the Basic tier and counts against a monthly channel allowance, so this workflow makes sense if you're scanning channels once or twice a month, not daily. The stored transcripts stay in the account, so follow-up searches with List Stored Transcripts are free. The break-even point is around six hours of manual watching per month—if you're spending less time than that on competitor research, the API key setup and credit costs aren't worth it. Buy this if you're running quarterly landscape reviews or building a content strategy database, not if you're just checking one video occasionally.