Render videos from Claude, Cursor, ChatGPT
Ngram is a native MCP server. Your assistant calls it like any other tool — in chat, in your editor, in any MCP client.
- Connect Claude Desktop or Cursor in one config edit
- Authenticate once with an API key — every chat is hooked up
- Ask the assistant to make a video — it picks the right call
- Check credits, list videos, fetch status — all from your chat
Rendered the explainer. Here's the preview - it ran in about 38 seconds.
Reach every AI client in one config
The bridge between your assistant and ngram. The AI decides when to render.
Works in every assistant you use
Claude Desktop, Cursor, Cline, Notion AI — any MCP client connects with one config edit.
Safe to wire into production
Per-request key validation, HTTPS-only, errors scrubbed of context — ready for real workloads.
Every ngram action available to chat
Create videos, check status, list past renders, read credits — each capability exposed as a discrete tool.
Make videos straight from chat
Ask the assistant for a 30-second explainer. The right tool gets picked, the render kicks off, the result lands in chat.
See every tool call in PostHog
Path, method, user agent, duration, status — every request flows through PostHog for observability.
Run hosted or local
Point at the hosted mcp.ngram.com URL, or run the official npm package locally for stdio clients.
Three steps to chat-to-video
From zero to a rendering video in under five minutes.
Create your ngram key
Open Settings → API Keys in ngram and create a new key. It starts with ngs_.
Point your client at ngram
Edit your MCP client config (Claude Desktop, Cursor, Cline) and add ngram as a server with your key.
Ask the assistant to create
Open chat in your client and ask it to make a video, check credits, or list past renders. The AI handles the rest.
What will your assistant build?
Anywhere your AI workflow lives, ngram is one tool call away.
Agent-driven video creation
Let an AI agent decide when to render a video as part of a longer workflow - research, draft, render, deliver.
Developer workflows
Build a video into your dev assistant. Ship release notes, demos, or onboarding videos from the same chat you ship code.
Knowledge-base videos
Pair Notion AI with ngram to turn knowledge-base entries into watchable explainers.
Customer success automations
Have an agent watch for trigger conditions in your support tool, then render a personalized video walkthrough.
Status checks at the speed of chat
Ask Claude how many credits you have, what your last 5 videos rendered as, or the status of a single video by id.
Custom internal tools
Wire ngram into your team's internal AI tooling. Every tool becomes a video on demand.
Built for AI-native teams
AI-native teams
Already using Claude Desktop or Cursor every day. ngram becomes one more tool the assistant can reach for.
Developer advocates
Render release demos and tutorial clips from the same chat you write changelogs.
Solo founders
Skip the context switch. Stay in chat, get a video, ship.
Internal-tools builders
Plug ngram into custom internal AI agents using the same MCP your other internal tools already speak.
Research teams
Turn deep-research notes into shareable explainers without leaving your assistant.
Frequently asked questions
Under the hood
- Native Model Context Protocol server
- Per-request API key validation, HTTPS-only
- stdio + HTTP transport support
- Scrubbed error context
- PostHog event stream per tool call
- All ngram endpoints exposed as MCP tools