Claude guide

Get a video transcript with Claude

Pull the timed transcript of a YouTube video so the assistant can quote and summarise it.

humansagents

1. Connect looot to Claude

  1. In Claude, open Customize, then Connectors.
  2. Click "+ Add", then "Add custom connector".
  3. Name it looot and paste the remote MCP server URL https://api.looot.ai/mcp. Click Continue.
  4. Review the authentication settings Claude detects and click Continue. Under Authentication pick "Sign in now" or "Sign in when needed". Under OAuth client, Anthropic recommends "Use Claude's published identity". Click Add.
  5. looot's sign-in page opens: sign in with your looot account and allow access. Runs are paid from your prepaid looot balance.

Anthropic's help center says custom connectors work on Free, Pro, Max, Ultra, Team and Enterprise, and a Free account can add one custom connector. On Team and Enterprise an Owner or Primary Owner adds it under Organization settings, Connectors, and each member then clicks Connect under Customize, Connectors.

In Claude Code instead: Add looot from your terminal with one command:

Command
claude mcp add --transport http looot https://api.looot.ai/mcp

Then type /mcp inside Claude Code and follow the browser sign-in.

2. Ask Claude

In a new chat, click the "+" button at the lower left, choose Connectors and switch looot on for this conversation. Then paste the prompt.

Prompt
Use looot to get the transcript of [YouTube URL]. For a YouTube link, run the endpoint scrapecreators-youtube-video-transcript. For another platform, search looot for the job video.transcript.get and pick the endpoint made for that platform. Then list the main points with the timestamp where each starts.

3. Jobs it calls and what they cost

4. A real run

We ran this task through the looot MCP on 2026-11-10. Asked: youtube.com/watch?v=jNQXAC9IVRw (Me at the zoo, 19 seconds). Total charged: $0.00188.

  • First try, through the job route: two endpoints for other platforms refused the link. Charged $0.
  • Second try, on the YouTube endpoint: 6 timed lines in English, from 0:01 to 0:16.
  • First line: "All right, so here we are, in front of the elephants". Caption tracks: English and German.
  • video.transcript.get on scrapecreators-facebook-adlibrary-ad-transcript (Scrape Creators, then AnyAPI (TikTok endpoint)): missed, charged $0. Run id run_4b9ac84f81634401b83e6360bfb63840.
  • video.transcript.get on scrapecreators-youtube-video-transcript (Scrape Creators): answered, charged $0.00188. Run id run_6eae4539e6f7455baecd414ffc398e9d.

Good to know

  • On 2026-11-10 the job route sent a YouTube link to endpoints made for Facebook ads and TikTok, which refused it at no charge. Naming the YouTube endpoint avoids that until the route is fixed.

The same task in ChatGPTThe prompt on its ownAll Claude guides

Give your agent every data provider, one key.

Start your workspace, top up, and paste one prompt into the agent you already use.