GitHubLog in
MCP server for YouTube, TikTok & Instagram

MCP server for YouTube, TikTok & Instagram

Point any MCP-compatible AI client at TranscriptFetch so your assistant can fetch transcripts from YouTube, TikTok, Instagram and direct media files, and discover videos, mid-conversation.

Server URL
https://transcriptfetch.com/mcp
Streamable HTTP · OAuth or API key

New here? The MCP server overview covers what it does and what it costs. This page is the setup and tool reference.

Setup

Add the server to your client

Nothing to install. All seven tools. Recommended for most clients.
Settings → Connectors
Add custom connector

Name  TranscriptFetch
URL   https://transcriptfetch.com/mcp
Sign in to TranscriptFetch and approve access when Claude asks. No API key needed; custom connectors need a paid Claude plan.Connect in Claude
Tools

Seven tools, one key

Credit pricing
ToolWhat it doesCost
get_transcriptFull transcript for a YouTube ID/URL, TikTok, Instagram, or direct media file. Pass ai_fallback: true to transcribe audio when captions are unavailable.1 credit
search_videosSearch YouTube, TikTok or Instagram by keyword; returns titles, IDs, and URLs. On YouTube, filter by upload date, length or captions, or sort by views.1 credit
search_channelsHosted onlyFind YouTube channels, or playlists, by keyword. Each result's URL goes straight to the channel and playlist tools.1 credit
list_channel_videosVideos for a YouTube handle (@name), channel ID (UC…), or URL, or for a TikTok or Instagram profile. On YouTube, pick the Shorts or live tab, sort by popular or oldest, or search inside the channel.1 credit
list_channel_playlistsHosted onlyA YouTube channel's playlists, or its podcasts. Pass a playlist's URL to list_playlist_videos for its videos.1 credit
list_playlist_videosVideos in a YouTube or TikTok playlist, by ID or URL.1 credit
get_creditsThe caller’s remaining credit balance. The only free tool.Free
Billed on success only. Failed, blocked and empty results cost nothing. Hosted only: not in the local package yet.
Example call

What your agent sends and gets back

Tool callget_transcript
{
  "video": "dQw4w9WgXcQ"
}
Responsesource: captions1 credit · abridged
{
  "videoId": "dQw4w9WgXcQ",
  "platform": "youtube",
  "title": "Example video",
  "channel": "Example Channel",
  "duration": 212,
  "language": "en",
  "source": "captions",
  "segments": [
    {
      "start": 0,
      "duration": 3.5,
      "text": "We're no strangers to love"
    }
  ],
  "outcome": "ok",
  "creditsSpent": 1,
  "balance": 99
}
Good to know

Behaviour worth planning for

  • Videos without captionsget_transcript says whether transcribing the audio would work (aiFallback.available). The assistant then calls it again with ai_fallback: true, billed at 1 credit per started 5 minutes of audio on delivery. Short media comes back in that call, typically in about 30 seconds; longer media returns outcome: "processing" with a jobId, and the next call with the same video collects the text.
  • Hosted or localThe hosted server has all seven tools, signs in with OAuth or an API key, and waits inline for short AI transcription. The local package runs over stdio with an API key, needs Node 18 or later, and has five tools.
  • Running in DockerThere is no port to expose. Pass -i, otherwise stdin closes and the container exits immediately.
  • Shared creditsTool calls draw from the same credits and rate limits as the API and the dashboard, and are logged against the key or account you connect with.
Open source, MIT licensedThe local server: read the source, run it from a clone, or install it from npm.
GitHubtranscriptfetch-mcp
Set up your client
Claude MCP server setup

All three Claude paths on one page: the claude.ai one-click connector (OAuth, no key), Claude Code's one-liner, and the Claude Desktop mcp-remote bridge.

Cursor MCP server setup

The ~/.cursor/mcp.json entry, and the same block for VS Code, so the tools appear in Agent mode after a restart.

Codex MCP server setup

Codex configures MCP over stdio, so this is the mcp-remote bridge and the ~/.codex/config.toml it belongs in.

By platform
YouTube MCP server

Ask about any YouTube link in the chat. Captions where the video publishes them, AI Fallback Transcription where it does not.

TikTok MCP server

Short-form clips read as text, with timestamps, so an assistant can quote what was said rather than the caption.

Instagram MCP server

Reels and video posts transcribed on demand, returned in the same shape as every other source.