ChatGPT

How to let ChatGPT watch a video

ChatGPT reaches Playhead over MCP once developer mode is switched on: Settings → Connectors → Advanced → Developer mode, then add https://mcp.tryplayhead.com/mcp as a connector.

Free tier: 60 credits a month, no card. One credit is about a minute of video.

What this changes

ChatGPT will summarise a YouTube video from its transcript and its metadata without ever seeing a frame, which is exactly the failure this fixes. A transcript cannot tell you what is on screen, when a cut lands, or what the text in the lower third says. With the connector attached, the answer comes from the pixels and carries the second it came from.

Setting it up

  1. 01

    Turn on developer mode

    Settings → Connectors → Advanced → Developer mode. Custom MCP connectors are hidden until this is on.

  2. 02

    Add the connector

    Paste https://mcp.tryplayhead.com/mcp and sign in with your Playhead account.

  3. 03

    Ask about a video

    Paste the link with your question. The tools appear in the composer once the connector is attached.

Things worth asking it

Each of these produces an answer with a second attached, because the timestamp is burned into the frame the model is looking at — it cannot lose track of when something happened.

  • Watch this YouTube link and list every scene change with its timestamp.
  • What is written on screen in the first five seconds of this TikTok?
  • Does this ad show the product before the 3-second mark?

How closely it looks is a choice

Six levels, picked by the agent from the question. Surveying an hour and reading a single frame of animation are different jobs, and charging the same for both would make one of them pointless.

LevelRateAnswers
overview0.5–2 fpsWhat is this video, start to end
standard5 fpsWhat happens when — cuts, overlays, entrances
detail5 fpsObjects and text blocks, tiles four times larger
fine10 fpsType, logos, subtle transitions
motion30 fpsAnimation, easing, dropped frames
singleone frameReading text at full resolution

Questions

Why does ChatGPT need a connector at all — it can open links?
It can read a page, including a transcript. It cannot decode a video stream or look at a frame. Everything visual — a cut, a caption, a gesture, a product shot — is invisible without something that turns the video into images.
Is developer mode risky?
It is the setting that allows custom MCP connectors at all. The scopes you grant are shown before you approve them, and Playhead's are read-only: analysing videos and reading your own plan.

Or somewhere else

One account, every client. The connection is attached to you rather than to a machine, so adding a second one costs nothing.

Stop describing the video to your agent.