MCP server / YouTube
YouTube MCP server for transcripts, search, and channels
Paste a YouTube link into your AI and it reads the video. Search, list a whole channel or playlist, and pull every transcript, from one server.
Connect it to your client
Any MCP client connects the same way: point it at the server URL and send a TranscriptFetch API key as a bearer header (on claude.ai, OAuth replaces the key). Client-by-client walkthroughs: Claude, Cursor, Codex, or the configuration reference.
https://transcriptfetch.com/mcp
npx mcp-remote https://transcriptfetch.com/mcp \ --header "Authorization: Bearer tf_live_YOUR_KEY"
What it can do
YouTube gets the deepest coverage of any platform here. get_transcript takes a watch link, a youtu.be link, or a bare 11-character ID; search_videos finds videos by keyword; and list_channel_videos and list_playlist_videos, which exist for YouTube only, turn a channel or playlist into a video list your assistant can walk. Chained together, that is a research loop: find the videos, pull the words, work on the text.
You: What does this channel say about pricing strategy?
youtube.com/@somecreator
Agent: [list_channel_videos, then get_transcript x6]
Across six videos the recurring advice is... (with quotes)Why YouTube is the easy one
Most YouTube uploads already carry a caption track, auto-generated if the creator never wrote one, and often in several languages. So a YouTube fetch is usually a read, not a transcription job: fast, and billed as a single credit like everything else. The AI fallback only has to step in for the minority of videos where captions are disabled or missing.
Scale is the thing to plan for instead. An hour of talking is roughly 10,000 words, and a channel can hold hundreds of hours, which is more than an assistant wants in context at once. The timestamped segments are what make that workable: an agent can skim, quote the exact moment, and point back to 41:32 instead of paraphrasing an hour.
The failures left are narrow and honest ones: live streams while they are still live, age-restricted or private videos, and clips with no speech. Everything public and finished generally yields text.
Frequently asked questions
Which tools are YouTube-only?
search_videos, list_channel_videos and list_playlist_videos. Transcript fetching itself also covers TikTok, Instagram, podcasts and direct media files, but discovery tools exist for YouTube alone.
Does it work on videos without captions?
Yes. When no caption track exists the audio is transcribed automatically and the tool returns the result the same way, so your assistant does not need a fallback path.
Can my assistant read an entire channel?
Yes: list_channel_videos returns the channel's videos, then get_transcript fetches each one. One credit per successful transcript; the listing calls are free.
What does it cost?
100 free credits a month, one credit per successful fetch, failures never billed. Paid plans start at $5 a month for 500 credits.
Is this affiliated with YouTube or Google?
No. TranscriptFetch is an independent transcript API; the MCP server exposes it over Anthropic's open Model Context Protocol and is listed in the official MCP Registry as com.transcriptfetch.
Your AI, with YouTube's words in reach
Sign up, grab a key, paste one config block into your client. 100 free credits a month, failures never billed.