MCP & API

Typecut MCP for any agent

Connect Typecut to your workflow and create finished, motion-designed videos straight from a conversation.

ClaudeClaude CodeCursorChatGPTYour agent

Connected in three steps

2

Add the Typecut connector

Name it Typecut and paste the URL:

3

Authenticate

Paste an API key from Settings → API keys (Creator plan and up) as the bearer token — then just ask Claude to create a video.

Tools

Just describe it — Typecut builds it

What your agent can do, in the words you'd actually use.

Create a project

Create an 8-minute documentary on the Bronze Age collapse — calm narrator, editorial captions, archival imagery.

Run a template

Run the History template on this topic outline and give me a draft to review.

Submit raw NJS

Render this NJS composition and send me the share link when it’s viewable.

Check on projects

List my projects and show the generation trace for any that failed.

Find assets

Search open and archival sources for 1960s Tokyo street footage with clear licensing.

Render & share

Render the final export in 4K and send me the share link when it’s done.

The pipeline's engines, one connection

SeedreamGPT ImageFluxGemini TTSElevenLabsZebra Motion Engine
Under the hood

A complete studio inside your agent

Server-authoritative projects

Agents create real projects through the same service layer the studio uses — nothing is a one-shot render you can’t touch afterwards.

Scene-level control

Every video is a composition, not a blob of pixels. Regenerate a scene, restyle the captions, swap an asset — without starting over.

Finished, not raw

Script, voiceover, motion design, music, captions, and the final render — the agent gets back a publishable video, not clips to assemble.

Same credits everywhere

No separate API pricing. Your plan credits flow through the app, the API, and MCP alike.

Read the API reference

Got any questions left?

How does Typecut connect to AI agents?
Over MCP (Model Context Protocol), the open standard that gives agents access to external tools. Once connected, your agent can create projects, check generation status, search assets, and fetch share links in a single session.
Do I need an API key?
Yes, for now — create one in Settings → API keys (Creator plan and up) and use it as the connector’s bearer token. OAuth sign-in is on the roadmap so agents can connect without managing keys.
How does pricing work?
There is no separate API pricing. Generations triggered by an agent draw from the same plan credits as the studio, at the same rates.
Which agents are supported?
Any MCP-capable client: Claude and Claude Code, Cursor, ChatGPT, or your own agent framework over the standard Streamable HTTP protocol.
How long does generation take?
Scene drafts land in minutes; full renders depend on length and resolution. Everything runs asynchronously — your agent polls project status and delivers results when they’re ready.
Can I drive Typecut without MCP?
Yes — the same capabilities are live today as a REST API under /v1: projects, workspaces, raw-NJS compositions, and asset discovery. See the API reference for the full surface.

From concept to content in one conversation

Start creating