mcpField
Y
YouTube Reader

YouTube Reader

Remote
github.comMedia & Design0Updated 1w ago

About YouTube Reader

YouTube Reader is a server for AI agents that need video content as text. It returns video metadata and a timestamped transcript through HTTP or a single MCP tool; transcript retrieval may be rate-limited by YouTube.

Key features

  1. 01Return title and channel metadata
  2. 02Fetch timestamped transcripts
  3. 03Accept common YouTube URL forms
  4. 04Offer HTTP and MCP interfaces

Use cases

  1. 01Review spoken content without watching video
  2. 02Locate a passage using transcript timestamps

MCP support

Runs as an MCP server with one tool, `youtube_reader(url)`, which returns the same report as the HTTP API. Transcript calls are priced per call through x402, according to the README.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://web-production-0cbb7.up.railway.app/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

Details

From the project README

YouTube Reader

Turn any YouTube video into text. For AI agents that can't watch video.

Give it a video link. Get back the title, channel, and the full spoken transcript with timestamps. One call, clean JSON. No API key needed.

Quick example

You send:

POST /read
{ "url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ" }

You get:

{
  "video_id": "dQw4w9WgXcQ",
  "title": "Rick Astley - Never Gonna Give You Up (Official Video) (4K Remaster)",
  "channel": "Rick Astley",
  "channel_url": "https://www.youtube.com/@RickAstleyYT",
  "thumbnail": "https://i.ytimg.com/vi/dQw4w9WgXcQ/hqdefault.jpg",
  "transcript_available": true,
  "language": "English",
  "segment_count": 61,
  "duration_seconds": 213,
  "full_text": "We're no strangers to love...",
  "segments": [
    { "start": 18.5, "text": "We're no strangers to love" },
    { "start": 22.1, "text": "You know the rules and so do I" }
  ]
}

Run it

python3 -m venv .venv
.venv/bin/pip install -r requirements.txt
.venv/bin/python server.py        # http://127.0.0.1:8081
  • POST /read with {"url": "<youtube link>"} → full report (JSON)
  • GET /health → {"ok": true}

Accepts watch?v=, youtu.be/, /shorts/, /embed/, /live/ links and bare video IDs.

Use as an MCP server (Claude Desktop, Cursor, Windsurf...)

.venv/bin/pip install fastmcp
.venv/bin/python mcp_server.py

One tool: youtube_reader(url) — same report as the HTTP API.

How transcripts are fetched (in order)

  1. youtube-transcript-api — direct, free
  2. Invidious instances — fallback, free

Known limitation: YouTube rate-limits automated requests from datacenter IPs, so transcript fetches can intermittently fail from cloud servers. Metadata (title/channel via official oEmbed) is unaffected. Professional fix: add a residential proxy (~$3 per 1,000 transcripts) as step 3.

Deploy (Railway, free tier)

  1. Push this folder to GitHub
  2. https://railway.app → New Project → Deploy from repo
  3. It reads requirements.txt + Procfile automatically. Done.
  4. Your API is live at the Railway URL: POST https://<your-app>.up.railway.app/read

Pricing

  • Metadata (title/channel): free
  • Transcript via MCP (youtube_reader tool): $0.02 per call, paid in USDC on Base via x402 — no signup, no API key. See PAYMENTS.md.
  • Plain HTTP API (POST /read): free while in beta
  • Heavy users: $19/month tier planned (Stripe, phase 2)

Monetization (MCPize)

Per-call pricing is configured in mcpize.yaml and the MCPize dashboard (Payments tab). Full setup guide, wallet instructions, testnet flow, and caveats: PAYMENTS.md.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.