FeedMCP

Data APIs / YouTube

YouTube transcript API

The full transcript of a YouTube video with timed segments, in a language of your choice; uses YouTube captions, any length.

Item Value
Endpoint GET https://api.feedmcp.com/v1/youtube/videos/transcript/get
ID youtube.videos.transcript.get
Price $0.002 / call
Returns One Transcript object
Auth Authorization: Bearer $FEEDMCP_API_KEY

Uses YouTube captions, auto-generated included, any length; language picks the track, falling back to the default one.

Parameters

Name Type Required Description
id string one of Video ID (the 11 characters after v=)
url string one of Video or post link
language string no Caption language, a two-letter code such as en, zh or ja; the default track if left out

Every endpoint also takes view=full to include the platform's raw data as raw, at the same price.

Returned fields

data.item is one Transcript object, with the same fields on every platform:

Field Type Description
id string The video or post ID, as the platform’s detail endpoint takes it
platform string Platform
url string | null Video or post link
language string | null Caption language (when the platform says)
text string The full transcript
segments {start, end, text}[] Timed segments in seconds (YouTube, TikTok, Instagram and Reddit; empty elsewhere)
duration_seconds number | null Video length in seconds
source 'captions' | 'ai' | null captions (the platform’s own), ai (AI transcription) or null (not known)

When you are charged

$0.002 per call that returns data. Not found, no data, bad parameters, platform errors and timeouts are free.

Each answer carries meta.billing: what this call cost and the balance left.

Request

cURL

curl "https://api.feedmcp.com/v1/youtube/videos/transcript/get?id=Hz033eUq6Qw" \
  -H "Authorization: Bearer $FEEDMCP_API_KEY"

Python

import os, requests

r = requests.get(
    "https://api.feedmcp.com/v1/youtube/videos/transcript/get",
    params={"id": "Hz033eUq6Qw"},
    headers={"Authorization": f"Bearer {os.environ['FEEDMCP_API_KEY']}"},
    timeout=30,
)
print(r.json())

JavaScript

const res = await fetch("https://api.feedmcp.com/v1/youtube/videos/transcript/get?id=Hz033eUq6Qw", {
  headers: { Authorization: `Bearer ${process.env.FEEDMCP_API_KEY}` },
});
console.log(await res.json());

MCP

From an AI with FeedMCP installed, the call tool:

{
  "endpoint": "youtube.videos.transcript.get",
  "params": {
    "id": "Hz033eUq6Qw"
  },
  "max_price": "0.002"
}

Response

The data below is made up; the shape is the real one.

{
  "code": "ok",
  "request_id": "req_7Qm2kXb9TzLp",
  "data": {
    "item": {
      "id": "Hz033eUq6Qw",
      "platform": "youtube",
      "url": "https://www.youtube.com/watch?v=Hz033eUq6Qw",
      "language": "English",
      "text": "Need some help? It is getting windy now.",
      "segments": [
        {
          "start": 97.4,
          "end": 101.58,
          "text": "Need some help?"
        },
        {
          "start": 101.6,
          "end": 104,
          "text": "It is getting windy now."
        }
      ],
      "duration_seconds": null,
      "source": "captions"
    }
  },
  "page": null,
  "warnings": [],
  "meta": {
    "endpoint": "youtube.videos.transcript.get",
    "platform": "youtube",
    "fetched_at": "2026-11-19T08:00:00.000Z",
    "source": "FeedMCP",
    "catalog_revision": "d09a736bd39e8b3f",
    "billing": {
      "charged": true,
      "amount": "0.002",
      "currency": "USD",
      "balance": "9.98"
    }
  }
}

Errors are described in Errors: each says whether you were charged (charged) and whether to retry (retry).

Other YouTube endpoints

Questions

How much does the YouTube transcript endpoint cost?

$0.002 per call that returns data. Not found, no data, bad parameters, platform errors and timeouts are free.

Do I need a platform account?

No. Only public data is read.

A video ID is the 11 characters after v= in a link; a channel is the channel ID starting with UC (author.id in video and search results); @handles are not supported yet.

How do I use it from an AI?

Install the MCP server and ask; your AI calls youtube.videos.transcript.get.