Connect an AI assistant to Recordarama

Recordarama has an MCP server. An AI assistant that supports the Model Context Protocol can use it to find your videos, read what was said in them, and tidy up their titles and summaries, with your permission.

Server address

What you can ask an assistant to do

“Find my video about the onboarding checklist and give me its link.”
“Summarize what I said in yesterday’s recording.”
“Write a follow-up email from the transcript of my last video.”
“Give my untitled recordings clearer titles.”

How to connect

In your AI assistant, add a custom connector (it may be called an MCP server or an integration) and paste the server address above.
The assistant sends you to Recordarama. Sign in if you are not already.
You see which app is asking and what it wants to do. Choose Allow.
You are sent back to the assistant, and it is connected. There is nothing to copy or paste.

You need a Recordarama account. Sign in or see how to get one.

What a connected assistant can and cannot do

It works as you. It sees your own videos and nothing belonging to anyone else. (A site admin's connection can see every video, as the admin can.)
You see what it is asking for before you allow it: reading only, or reading and changing titles and summaries. It gets nothing you did not approve.
It cannot record, upload, delete videos, change your account or plan, or see your password or payment details.
You can disconnect it at any time on your Account page, under “Connected apps”. It stops working at once.

Tools

Tool	What it does	Takes	Permission
`list_videos`	Lists your videos, newest first, with title, summary, length, views and share link.	Optional: search (text to look for in titles and summaries), limit (1 to 100, default 25).	`videos.read`
`get_video`	Gives the details of one video, including its download link and whether it has subtitles, an AI voice or a trim.	id (the 12-character video id from list_videos).	`videos.read`
`get_transcript`	Returns what was said in a video, as plain text and line by line with timings. Available once the video has subtitles or an AI-written title.	id.	`videos.read`
`update_video`	Changes a video's title, its summary, or both. Only the fields given are changed.	id, and title and/or summary.	`videos.write`

For developers

Transport: Streamable HTTP. Send one JSON-RPC 2.0 message per POST to the server address; replies are JSON, not a stream.
Sign-in: OAuth 2.1 with PKCE (S256, required) and dynamic client registration. A request without a token gets a 401 whose WWW-Authenticate header points to the discovery document.
Discovery: https://recordarama.com/.well-known/oauth-protected-resource and https://recordarama.com/.well-known/oauth-authorization-server
Registration: POST https://recordarama.com/oauth/register/ with a JSON body containing client_name and redirect_uris. Redirect addresses must use https, or http on localhost.
Scopes: videos.read and videos.write. Asking for none requests both.
Tokens: access tokens last one hour. Refresh tokens last 60 days and are replaced each time they are used.
Scripts: for a tool that cannot open a browser, a signed-in person can create a long-lived token on their Account page and send it as Authorization: Bearer <token>.

Recordarama homePrivacy policyTerms of service