Connect ngram over MCP

Use this when

Use this guide when you want an AI assistant to use ngram tools, a connector asks for the ngram MCP server URL, authentication is unclear, tools do not appear after connecting, or ChatGPT can read ngram data but cannot create a video. MCP is for supported external AI clients; use the REST API or an automation connector when your system does not support remote MCP servers.

What this does not do

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

Before you start

Sign in to ngram and to the MCP client you intend to use. Confirm that the client supports remote HTTP MCP servers and whether it offers its own connector sign-in or accepts an Authorization header. Never paste an ngram key into chat, a prompt, a shared document, or a screenshot.

Overview

The Model Context Protocol is an open standard for connecting AI assistants to external tools. ngram exposes every public endpoint as an MCP tool with a typed schema the agent reads at connect time, so there is no glue code to write. Use the hosted HTTP endpoint for setup.

Connect to the hosted server (recommended)

Use https://mcp.ngram.com as the remote MCP server URL. Claude connector flows handle auth in the connector UI. In ChatGPT, full create-video actions currently require the web app on Business, Enterprise, or Edu, plus developer-mode and workspace app access; Pro is limited to read/fetch MCP actions. Header-capable clients can authenticate with Authorization: Bearer ngs_…. Every signed-in ngram user can create that bearer credential from Settings → API Keys. Connector flows that handle authentication in their own UI do not require you to paste a key.

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

Do not paste this JSON into a connector UI that only asks for a server URL. In those clients, add ngram as a custom or remote connector using https://mcp.ngram.com and follow the auth flow the client presents. The hosted endpoint is the supported default for normal setup.

Connect and verify the server

1. Open the client's connectors, integrations, or MCP settings and choose its action for adding a custom remote server.
2. Enter https://mcp.ngram.com as the server URL. Use the connector's sign-in flow when it offers one.
3. For a header-capable client only, create a key at ngram Settings → API Keys, store it in the client's secure credential field, and send it as an Authorization bearer header.
4. Finish the connection and review the tools the client discovered. Ask a read-only question such as “How many ngram credits do I have left?” before starting a render.

What you should see

The client should show ngram as connected and expose account, catalog, video, and image tools permitted by that client's plan and workspace policy. A credit question should return the authenticated ngram account's plan and balance. Creating a video starts a server-side job; the client can check its status and return the final link when it completes.

Tool catalog

Every ngram endpoint maps to a discrete tool. The agent reads the typed schema for each at connect time, so it can pick the right one without a docs lookup mid-conversation.

Account

• get_account — return the authenticated ngram account.
• get_credits_remaining — return the credit balance and plan.

Catalog

• list_voices, list_styles, list_aspect_ratios, list_durations — the options a render can use.
• list_animation_modes, list_scenarios, list_brand_kits — resolve modes, presets and your brand kits.

Videos

• create_video_from_text and create_video_from_url — start a render from a prompt or a source URL.
• prepare_video then create_video — preview the resolved settings, then confirm and generate.
• get_video_status, list_videos, cancel_video — track, list and stop renders.

Images

• create_image, edit_image, get_image_status, list_images — generate and edit images the same way.

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. Confirm that the server URL is exactly https://mcp.ngram.com and that you did not paste the JSON configuration into a URL-only field.
2. Check the signed-in ngram account, the MCP client's plan, developer-mode setting, and workspace app permissions. In ChatGPT, Pro can read or fetch through MCP but cannot start an ngram video render.
3. Disconnect and reconnect ngram once, then repeat the read-only credit check. Do not repeatedly replace keys or retry a create action while the connection state is uncertain.

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.