Gouda Studios Agent Tools

How to download YouTube transcripts in bulk for AI and RAG

Short answer: Pass a list of video URLs or IDs to the YouTube Transcript Extractor. Duplicates are removed, each video becomes one JSON row with a clean text field ready for chunking and embedding, and you can export everything as JSON, CSV or Excel or pull it with the API. At $4 per 1,000 transcripts it is cheaper than most transcript scrapers.

Tool: YouTube Transcript Extractor | Run it on Apify | MCP: https://gouda-studios-agent-tools.vercel.app/mcp

Steps

  1. Collect the video URLs or IDs, for example from a channel or playlist export.
  2. Run the tool with videos set to the full list and maxVideos high enough to cover it.
  3. Keep includeSegments on if you want to cite timestamps in answers; turn it off for smaller rows.
  4. Chunk the text field and send it to your embedding model or LLM, or download the dataset as CSV.

Example input

{
  "videos": [
    "UF8uR6Z6KLc",
    "https://youtu.be/8jPQjjsBbIc",
    "https://www.youtube.com/shorts/aqz-KE-bpKQ"
  ],
  "languages": [
    "en"
  ],
  "includeSegments": true,
  "maxVideos": 1000
}

Use it from an AI agent

Connect any MCP client to https://gouda-studios-agent-tools.vercel.app/mcp, or see the Claude and Cursor guide.

FAQ

How many videos can one run handle?

Up to 1,000 by default through maxVideos. Raise it for bigger batches.

Can an AI agent call this directly?

Yes. Connect any MCP client to the Gouda Studios MCP endpoint, or call the Apify API with run-sync-get-dataset-items and get the transcripts back in the same request.

What does it cost?

$0.004 per transcript returned plus $0.0005 per run. Failed or caption-less videos are free.

More guides

Contact: m0f@mail.grokbot.com | Machine-readable: llms.txt, llms-full.txt, tools.json, openapi.json | Source: GitHub

Gouda Studios Agent Tools badge