Skip to main content
Glama
BulkTranscripts
by pratie

BulkTranscripts: YouTube transcript MCP server for Claude, ChatGPT and Cursor

YouTube transcripts for AI agents — one video, a whole channel, or a playlist — plus YouTube search and free new-upload tracking, through one hosted MCP server. No local process, no yt-dlp to keep fresh, no IP blocks to fight: extraction, proxy rotation and PO-token handling all happen server-side.

Install

Claude Code

claude mcp add --transport http bulktranscripts https://bulktranscripts.co/mcp

Run /mcp and pick bulktranscripts to sign in. This repo is also a Claude Code plugin (server + skill):

/plugin marketplace add pratie/bulktranscripts-mcp
/plugin install bulktranscripts@bulktranscripts

Claude (web and desktop)

Settings → Connectors → Add custom connector → URL https://bulktranscripts.co/mcp. Claude opens the Google sign-in; credits then follow your account.

ChatGPT

Install BulkTranscripts from ChatGPT's plugin directory and sign in with Google. It works on free ChatGPT plans too.

If your workspace blocks plugins, enable Developer mode under Settings → Connectors → Advanced, then Create a connector with the URL https://bulktranscripts.co/mcp and authentication OAuth.

Gemini CLI

gemini extensions install https://github.com/pratie/bulktranscripts-mcp

Sign in with Google when Gemini CLI opens the OAuth page on first use.

Cursor, VS Code, Windsurf, any client with a JSON config

{
  "mcpServers": {
    "bulktranscripts": {
      "type": "http",
      "url": "https://bulktranscripts.co/mcp"
    }
  }
}

OAuth-capable clients start the sign-in on the first request. Clients that cannot do OAuth send an API key instead — create one free at https://bulktranscripts.co/app?tab=mcp (Google sign-in, 30 free credits) or use the license key from a credit pack:

{
  "mcpServers": {
    "bulktranscripts": {
      "type": "http",
      "url": "https://bulktranscripts.co/mcp",
      "headers": { "Authorization": "Bearer ${BULKTRANSCRIPTS_API_KEY}" }
    }
  }
}

Clients that cannot send headers can append ?key=YOUR_KEY to the URL.

Related MCP server: fetchworks-mcp

What the agent can do

Tool

What it does

Cost

get_transcript

Full transcript of one YouTube (or TikTok) video as clean text with metadata; timestamped segments on request.

1 credit, free on repeat reads from your library

get_transcripts

Up to 20 videos in one call; videos without captions are reported per item and never fail the batch.

1 credit per new transcript

search_youtube

Search YouTube for videos, channels or playlists by keyword.

1 credit

search_channel

Search inside one channel's uploads for a topic.

1 credit

get_channel_videos

List up to 1,000 of a channel's videos (id, title, duration, URL).

1 credit

get_playlist_videos

List up to 1,000 videos of a public or unlisted playlist, in order.

1 credit

get_latest_videos

A channel's newest uploads with publish dates, from RSS.

Always free

start_bulk_extract

Fetch every transcript of a channel, playlist or video URL (up to 1,000) as a background job; returns a run_id at once.

Listing free, 1 credit per new transcript

get_bulk_status

Progress of a bulk job with check_again_in_seconds, so agents poll at the right pace.

Free

list_bulk_results

Paged per-video outcomes of a bulk job (id, title, channel, duration, word count, status). Text stays behind get_transcript, free once in the library.

Free

Every tool is read-only: nothing on YouTube or in your account is modified or deleted. Transcripts land in your library once and are free to re-read forever, in any format, so agents that revisit the same videos do not burn credits. get_latest_videos is free on purpose: poll it to detect new uploads, then spend a credit only on what is actually new. For anything larger than about 20 videos, start_bulk_extract runs the whole channel or playlist on the server while the agent polls get_bulk_status.

Example prompts

Summarise the last 10 videos from @veritasium and tell me which ones cover quantum mechanics.

Find talks about founder mode on the Y Combinator channel and pull the transcripts.

Watch @simonwillison for new uploads and transcribe anything about local models.

How auth works

An unauthenticated request to /mcp answers 401 with a WWW-Authenticate challenge pointing at /.well-known/oauth-protected-resource/mcp, which is how MCP clients know to start the OAuth flow (dynamic client registration is supported). The same Authorization header accepts an API key (bt_ak_…) or a license key, with or without the Bearer prefix.

Pricing

Signing in includes 30 free credits. After that, one-time packs: $4.99 / 200 credits, $14.99 / 1,200, $29 / 5,000. No subscription; credits never expire. Details at https://bulktranscripts.co/#pricing.

Repo layout

File

Used by

.mcp.json

Claude Code plugin config (hosted endpoint, OAuth)

.claude-plugin/plugin.json, .claude-plugin/marketplace.json

Claude Code plugin marketplace

plugin.json, mcp.json

Cursor / agent-plugins.org

.codex-plugin/plugin.json, .agents/plugins/marketplace.json, assets/icon.png

Codex plugin

skills/youtube-transcripts/

Agent skill for the MCP tools (the REST API version is at pratie/youtube-transcript-skill)

gemini-extension.json

Gemini CLI extension (hosted endpoint, OAuth)

server.json

Official MCP registry manifest

License

MIT — see LICENSE. This repo holds the manifests, plugin and skill for the hosted server; the extraction service itself is operated at bulktranscripts.co.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      B
      maintenance
      Provides AI agents with token-optimized access to YouTube data, including video details, transcripts, channel statistics, trending videos, and search.
      1,508 npm
      MIT
    • A
      license
      A
      quality
      D
      maintenance
      Enables AI agents to retrieve YouTube transcripts from individual videos, channels, and search results, supporting multiple output formats such as plain text, SRT, and VTT.
      3
      37 npm
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Enables AI agents to read transcripts of any public video, research channels, analyze performance, read comments, and manage multiple owned channels.
      16
      74 npm
      1
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables AI clients to pull full transcripts from YouTube videos, search videos and channels, browse a channel's complete upload history, search within a single channel, and extract every video in a playlist through one hosted remote endpoint. It supports API-key or OAuth 2.1 sign-in with no local installation, returning metadata and paginated results with stable error codes.
      10
      MIT