AI summaries

Summarize YouTube videos with Claude or ChatGPT

For a YouTube video summarizer workflow, transcript1 returns the transcript with timestamps and optional speaker labels. Your AI assistant writes the summary. Connect through MCP, use the agent skill or pass the transcript from the API to your own model.

  • transcript1 supplies the transcript
  • Your assistant writes the summary
  • 1 credit per audio minute, rounded up

How it works

Connect your assistant
Add transcript1 as a remote MCP connector in Claude or ChatGPT and sign in with OAuth.
Give it the YouTube URL
The assistant requests the transcript through MCP. transcript1 uses available captions first, otherwise AI speech-to-text.
Ask for the summary you need
Once your assistant has read the transcript, ask it to summarize the video. transcript1 does not write the summary.
Keep timestamps and speakers
Use timestamps to check the summary against the recording. Request diarize=true for Speaker 1 and Speaker 2 labels; this uses speech-to-text instead of captions.
Use your own agent
The agent skill provides instructions in SKILL.md. Your code can also read a transcript from the API and pass it to a model.

MCP

Give Claude or ChatGPT the transcript

Connect to https://api.transcript1.com/mcp and ask your assistant to transcribe the YouTube link. Long transcripts are paged; the assistant needs to read the remaining pages before summarizing the whole video.

Tool calls
transcribe(platform="youtube", url="https://www.youtube.com/watch?v=dQw4w9WgXcQ")

get_transcription(id="tr_8fK2Lm0PqRsTuVwXyZaBcD", from_seconds=900)

Your model

Pass timestamped text to your summarizer

Read the completed transcript as text through the API and give it to your model with your summary instructions. Ask for timestamps beside key points so you can check them against the source.

cURL
curl "https://api.transcript1.com/v1/transcriptions/tr_8fK2Lm0PqRsTuVwXyZaBcD?format=text&timestamps=true" \
  -H "Authorization: Bearer $TRANSCRIPT1_API_KEY"

Agent skill

Let a skills-aware agent request the transcript

The transcript1 agent skill explains how to request and read transcriptions. Your agent uses that text as the input for its own summary.

Skill index and file
https://transcript1.com/.well-known/agent-skills/index.json
https://transcript1.com/.well-known/agent-skills/transcript1/SKILL.md

Start with 60 free minutes

60 free minutes after you verify your email. No card required.

Get your API key

Frequently asked questions

Something missing? Ask support

Is transcript1 a YouTube video summarizer?

transcript1 does not summarize videos. It returns the transcript with timestamps and optional speaker labels. Your AI assistant writes the summary using that text.

Can Claude summarize YouTube videos with transcript1?

Connect transcript1 to Claude over MCP, let Claude request and read the transcript, then ask Claude to summarize it. ChatGPT can use the same MCP endpoint.

Can I use the transcript with my own AI model?

Yes. Fetch text or JSON through the REST API and pass it to your model. Agents that support skills can also read the transcript1 agent skill.

What if the YouTube video has no captions?

With source=auto, transcript1 transcribes the audio when no usable caption track exists. The assistant can use that transcript to write its summary.

What does transcript1 charge for this workflow?

The transcript costs 1 credit per audio minute, rounded up. New accounts get 60 free minutes once after email verification, with no card required. Failed and canceled transcriptions are not charged. This allowance covers transcription, not your AI assistant's service.