Skip to main content
Automations and developer

Connect ngram over MCP

Connect ngram to a compatible AI assistant, check your account, then turn a brief or public page into a video you can preview and export.

Written By Anish Muppalaneni

Last updated 9 days ago

Read before this

  1. Choose how to automate ngram and connect integrations

Use this when

Use this guide when you have a brief or a public source and want to turn it into a video from an AI assistant. For example, give a compatible MCP client a release note, review the resulting preview and export the approved MP4 for your team. If you’re connecting a script that speaks HTTP, start with the ngram API guide instead.

What this does not do

Connecting ngram does not add MCP support to an incompatible client or override the client's plan, workspace-app or tool-approval rules. It also does not make a secret safe to paste into a conversation.

Need to sign in while ngram captures a product workflow? Start that tutorial in ngram Studio. An MCP client can request a quote, start the approved video and check its result, but it cannot open Live Preview or hand you control of the browser. See Make a walkthrough with Browser Handoff.

Before you start

Have an ngram account, available credits and a client that supports remote HTTP MCP. Check that your client or workspace lets you add a connector and use its creation actions. Use OAuth sign-in when offered. If your client only accepts a bearer token, you’ll need an API key. Creating videos with a key requires Basic or higher.

Overview

Model Context Protocol (MCP) lets a compatible assistant ask ngram to run actions such as checking your account, preparing a video, or opening an existing project. The actions available depend on the server URL and your client.

Connect to the hosted server (recommended)

For a remote MCP client, use https://mcp.ngram.com. For a dedicated connector, follow the Claude or ChatGPT guide for the correct setup and permissions.

Connect and verify the server

  1. Open your client’s MCP or connector settings and choose the option to add a remote server.

  2. Paste https://mcp.ngram.com as the server URL.

  3. If OAuth is available, choose it and sign in to ngram when your client prompts you.

If your client requires an API key

If your client accepts a bearer token and has no sign-in flow, open Settings → API Keys and create a key. Use the client’s secure credentials field or its HTTP header configuration. The exact configuration format depends on the client.

{
  "mcpServers": {
    "ngram": {
      "url": "https://mcp.ngram.com",
      "headers": {
        "Authorization": "Bearer ngs_your_key"
      }
    }
  }
}

Keep the key in the client’s credentials field. If your client asks for a server URL and opens a sign-in screen, use that flow.

What you should see

Open a conversation with ngram enabled and ask, “Which ngram account is connected, and how many credits are available?”

The client shows ngram as connected. The account check returns the account you signed in with and its available credits. If it shows a different account, reconnect with the intended account before confirming a video.

Check credits before creating

Ask for your credit balance before starting a render. API-key video creation requires Basic or higher. Supported connected-app sign-in can create on Free with credits: ngram-flash remains available, while premium models have the monthly Free allowance and 60-second limit. A Free connected-app video file includes an ngram watermark. Your client’s permissions also apply.

Make a video, review the preview and export an MP4

Use the full profile at the root server or the Claude connector for this workflow. The narrated explanation connection for ChatGPT has separate creation, status and cancellation actions. See the ChatGPT workflow if you use that connection.

  1. Ask the assistant to make a video from your source. Try “Use this public release note to make a short explainer for project leads. Show what changed and what they should do next.” Add the release-note URL to the request.

  2. Check the proposed model, format, estimated duration and credit cost. Confirm creation when the quote matches what you want. Review the plan if the assistant asks you to.

  3. Wait for the video to finish. If you need an update, ask for the status of the same video. The final credit cost can differ from the estimate when the actual duration differs.

  4. Play the preview in chat if your client displays it. If the player is unavailable, open the project link in ngram. Check the key moment before approving the cut.

  5. Ask for an MP4 export of the reviewed version. In a client that shows Export MP4, select it, send the export request if it appears in the composer, and allow the action if asked.

  6. Wait for the export to finish. Open the download link, save the MP4 and play the part you reviewed. If you revised the video after exporting, request a new export for that version.

For a worked example of reviewing a timestamp, asking for a revision and checking the exported version, see How to preview and export a video with the ngram MCP server.

Tool catalog

These are examples from the full profile. Check the tools discovered by your connected client for the actions available to you.

Account

  • get_account shows the connected account and its credits.

Catalog

  • get_creation_options shows available creation and delivery settings.

  • browse_styles and browse_voices help you choose a style and voice.

Videos

  • prepare_video checks the settings and estimated cost. create_video starts a confirmed request.

  • get_video checks status and returns the preview or project link. cancel_video cancels a running job.

  • export_video exports the chosen version as an MP4.

Images

  • create_image starts a new image, edit_image edits an image, and get_image checks the result.

Example prompts

Once connected, just ask in plain language — the agent picks the right tool:

  • "Make a 30-second product teaser for our pricing page and tell me when it's done."

  • "How many credits do I have left?"

  • "List my last five videos and their status."

If authentication or tools do not work

  1. If sign-in does not open, confirm your client supports remote HTTP MCP and OAuth. Check the server URL for your client and reconnect once.

  2. If ngram connects but you cannot create, check your available credits and the client’s plan or workspace permissions. If you connected with an API key, check that your ngram plan is Basic or higher.

  3. If playback is missing, open the returned project link. If the assistant loses the video while it is processing, ask it to check the same job before starting another.

  4. If the preview is ready but you have no MP4, request an export on the full profile and wait for the export to finish.

Contact Support when

Contact Support when one reconnect still leaves ngram unauthorized, no tools are discovered, a read-only tool returns the wrong account, or an eligible create action is missing. Include the client name and version, account plan, workspace type, server URL, approximate time, and the visible error with all keys and personal data removed. Never send an API key or authorization header.

MCP server FAQ

Which MCP clients are supported?

Use a client that supports remote HTTP MCP and a compatible sign-in or bearer credential. ngram has specific setup guides for Claude and eligible ChatGPT workspaces.

Which endpoint should I use?

Use https://mcp.ngram.com for a general remote client. For Claude use /claude, and for the ChatGPT narrated explanation connection use /openai.

Can I create through a connector on Free?

Through supported connected-app sign-in, Free accounts can create with credits and the current model limits. Using an API key for video creation requires Basic or higher. Ask ngram to check your plan and credit balance first.

How do I authenticate?

Choose OAuth if your client offers it and sign in to ngram when prompted. Use an API key only when your client uses a bearer token instead of OAuth.

What tools does the server expose?

The full profile can check your account, find creative options, create videos and images, check results and export an MP4. The ChatGPT narrated explanation connection has a smaller set of actions.

Are there rate limits?

Yes — the same public-API limits apply. Up to three video jobs can be processing per account at once; the agent surfaces a rate-limit response when you exceed it.

Why are ngram tools not showing after I added the server?

Check the server URL for your client, finish signing in, and confirm the client supports remote HTTP MCP. Reconnect once if the tools do not appear.

Why can ChatGPT see ngram but not create a video?

ChatGPT creation actions currently require ChatGPT web on Business, Enterprise, or Edu with the required workspace app permissions. Pro MCP access is limited to read and fetch actions.

Should I paste my ngram API key into the conversation?

No. Connector-managed flows handle sign-in in their own UI. A header-capable custom client should store the key in its secure credential or header field, never in a prompt, message, or screenshot.

Keep going

Ready to connect ngram?

Add the server URL in your client and sign in to ngram when prompted. Use an API key if your client requires a bearer token.

Review connection steps


Read this guide in Markdown