AI Reads YouTube

Transcripts, search, channels and playlists, plus summaries, chat, comments and images. Over REST or MCP, on one key, spending the same credits as the app.

125+ languagesNo blocks, no proxiesREST or MCP, one key

Let your AI agents fetch the transcript and work with it. One MCP server, the same key.

Your first call200 OK
curl "https://api.localhost/v1/transcript?video=iG9CE55wbtY" \
-H "Authorization: Bearer yts_live_..."
{
"language": "en",
"length_text": "19:24",
"cue_count": 412,
"metadata": { "title": "Do schools kill creativity?" },
"cues": [
{ "text": "Good morning. How are you?", "start": 0.0 }
]
}
1M+Transcripts generated
125+Languages covered
10,000Active users
Endpoints

One key. The whole of YouTube.

Ten endpoints, priced by what each one costs to run. The AI ones are not an add-on tier, they are on the same key and the same balance. Each one is also an MCP tool, so an agent can reach it without you writing the call.

YouTube data
AI, on the same key
GET/v1/transcript

The official caption track for any video, watch, Shorts, mobile or embed URL, cleaned, with or without timestamps, in 125+ languages.

1 credit
SummarisingSearch indexesStudy notesRepurposing
Request
curl "https://api.localhost/v1/transcript?video=iG9CE55wbtY&timestamps=true" \
-H "Authorization: Bearer yts_live_..."
Response
{
"video_id": "iG9CE55wbtY",
"language": "en",
"length_seconds": 1164,
"length_text": "19:24",
"cue_count": 412,
"metadata": {
"title": "Do schools kill creativity?",
"channel": "TED"
},
"cues": [
{ "text": "Good morning. How are you?", "start": 0.0, "duration": 3.2 },
{ "text": "It has been great, has it not?", "start": 3.2, "duration": 2.9 }
]
}
MCP server

Hand YouTube to your agent.

Point any MCP client at the server, paste the same key once, and the model decides when to pull a transcript, sweep a channel or read an hour of video. No wrapper code, no second bill.

Add it in the agent.

Claude CodeTerminal
claude mcp add --transport http youtubescribe http://mcp.localhost:3000/v1 --header "Authorization: Bearer yts_live_..."
  1. 01Paste the command in a terminal, in the project folder.
  2. 02Restart Claude Code, then run /mcp.
  3. 03youtubescribe should read connected.

Nothing to install. Add the block to your client's config, restart it, and the ten tools appear.

mcp.jsonstreamable http
{
"mcpServers": {
"youtubescribe": {
"type": "http",
"url": "http://mcp.localhost:3000/v1",
"headers": {
"Authorization": "Bearer yts_live_..."
}
}
}
}
One key for both.

The REST key is the MCP key. Same balance, same credit costs, one dashboard.

Tools describe their cost.

Each schema carries its credit price, so a model reaches for the free tool before the expensive one.

Failures still do not bill.

A video with no captions returns a named error to the agent and draws down nothing.

Agent sessionMCPyoutubescribe connected

Summarise TED's three newest uploads and pull the sharpest line from each.

01latest_uploadschannel: "@MrBeast"
Free
02summarize_video3 video ids
36 credits
03get_transcriptvideo: "k9Rm_SZjKs4"
1 credits

All three circle the same argument, that attention is the scarce input now. The medicine talk puts it most plainly:

08:12 We automated the reading. We never automated the judgement.
3 tool calls · 37 credits · 5.2s
Ten tools, one server
get_transcript

Full caption track, timestamps optional

1 credit
summarize_video

TL;DR plus timestamped key points

12 credits
search_youtube

Videos, channels or playlists by query

1 / page
ask_video

A question answered with citations

8 / message
search_channel

One creator's catalogue, by topic

1 / page
get_comments

Comments, with optional sentiment

1 / page
list_channel_videos

Every upload from a channel, in order

1 / page
generate_images

Story frames keyed to the transcript

20 / image
get_playlist

Whole courses and lecture series

1 / page
latest_uploads

Newest uploads, safe to poll

Free
Clients it runs in
Claude DesktopClaude DesktopClaude CodeClaude CodeChatGPTCursorCursorWindsurfWindsurfCopilotCopilotZedZedContinuen8nn8nLangGraphLangGraph
Models it works with
ClaudeGPTGeminiGrokLlamaDeepSeekMistralQwen

The server is model-agnostic. Anything that speaks MCP can call it, and the tool schemas carry their own credit cost so a model can pick the cheap route first.

Three minutes, start to first response.

01

Get a key

Sign up and copy it from the dashboard. 50 credits are on the account before you have entered a card.

02

Call an endpoint

One header, one query string. Every response is JSON with the same envelope, so error handling is written once.

03

Ship it

Works from anything that speaks HTTP, your own code, n8n, Make, Zapier, a cron job on a Pi.

Node
const res = await fetch(
"https://api.localhost/v1/ai/summary?video=" + id,
{ headers: { Authorization: "Bearer " + process.env.YTS_KEY } }
);
const { tldr, points } = await res.json();
for (const p of points) {
console.log(p.index, p.text, p.t_ms);
}
Errors that name the cause

no_captions, private_video, region_blocked, live_in_progress. Never a bare 500, and never a bill.

Retries handled upstream

Transient YouTube failures are retried inside the request. You see one response, not a backoff loop.

Stateless by default

Transcript text is passed through, not stored. Turn on the library if you want it kept.

Pricing

One balance. Works in the app or through the key.

Credits bought on Pro or Studio spend either way. A transcript call costs one credit, an AI summary costs twelve, and the free endpoints never draw down the balance.

50 free credits to start, no card.
Free
$0

50 credits, once. Enough to wire it up and see the shape of the data.

50 credits to startThat is 4 video summaries, 2 thumbnails or 6 chat answers.
Top up any time, credits never expire
250 for $2.49750 for $5.992,000 for $13.99
  • Every endpoint unlocked
  • No card, no trial clock
  • 60 requests per minute
Pro
$4.99/month

Billed monthly. Cancel any time.

1,000 credits a monthThat is 83 video summaries, 50 thumbnails or 125 chat answers.
Top up any time, credits never expire
250 for $2.49750 for $5.992,000 for $13.99
  • VTT, Markdown and JSON exports
  • Bulk playlist and channel export
  • Saved history and collections

Secure checkout by Stripe

Studio
$12.99/month

Billed monthly. Cancel any time.

3,000 credits a monthThat is 250 video summaries, 150 thumbnails or 375 chat answers.
Top up any time, credits never expire
250 for $2.49750 for $5.992,000 for $13.99
  • Everything in Pro
  • The best rate we do, $0.0043 a credit
  • Same price whether you use text, images or voice
Credit cost per endpoint
One credit
Get a transcript1 credit
Search YouTube1 / page
Search inside a channel1 / page
List channel videos1 / page
Extract a playlist1 / page
Export comments1 / page
AI, priced by what it costs to run
Chat with a video8 / message
AI summary12 credits
Chapter a long video12 credits
Story image20 / image
Free, poll these as often as you like
Track new uploadsFree
Resolve a channel handleFree
Thumbnail setFree
Revenue estimateFree
Only successful requests bill

Failures do not bill. A video with no caption track, a private video, an upstream timeout, none of them draw down your balance.

The same balance works in the app and through the key

Credits bought on Pro or Studio spend either way, in the browser or through your key. There is no separate API subscription and no separate top-up rate.

Higher volume

Past roughly 250K credits a month the rate comes down and the rate limit is negotiable. Tell us the shape of the workload.

Estimate

What it costs at your volume.

Drag to your monthly credit volume. The figures follow the monthly cycle you picked above.

5,000

Roughly 5,000 transcripts, or 416 AI summaries, at this volume.

Total per month$26.98
Studio plan
$12.99
1 pack of 2,000 credits
$13.99
Credits covered
5,000 credits
Effective per 1,000 credits
$5.40
Questions

The things people ask first.

Credits are weighted by what the call costs to run. A transcript is 1 credit, an AI summary is 12, a story image is 20, and endpoints like tracking new uploads are free. Failures never bill, so no captions, a private video or an upstream timeout do not draw down the balance.

Start on 50 credits.

  • No card
  • Every endpoint
  • Failures do not bill
api.localhost/v1/Get an API key