Martini

Martini MCP

Official client integration and distribution repo for Martini's hosted Model Context Protocol server.

Martini is an AI video production tool for filmmakers. The MCP server lets compatible AI assistants browse and create Martini projects, scenes, subjects, assets, bins, uploads, and generation jobs.

Server

https://www.martini.film/mcp

Authentication is handled by Martini's OAuth flow. This repository does not include local binaries, API keys, or secrets.

Install

Gemini CLI

gemini extensions install https://github.com/Martini-Film/mcp --auto-update

Restart Gemini CLI, then verify the connection:

/mcp

You should see a martini MCP server.

Claude

In Claude, add Martini as a custom connector with this server URL:

https://www.martini.film/mcp

Claude Code

claude mcp add --transport http martini https://www.martini.film/mcp

Generic MCP Clients

Use a remote HTTP MCP server named martini:

{
  "mcpServers": {
    "martini": {
      "type": "http",
      "url": "https://www.martini.film/mcp"
    }
  }
}

Some clients use httpUrl instead of url:

{
  "mcpServers": {
    "martini": {
      "httpUrl": "https://www.martini.film/mcp"
    }
  }
}

Gemini CLI Commands

This extension includes prompt-wrapper commands:

• /martini:connect
• /martini:overview
• /martini:generate-plan
• /martini:upload-check
• /martini:project

Uninstall

gemini extensions uninstall mcp

Docs

• Client setup
• Gemini CLI extension
• OAuth and authentication
• Tool safety
• Where this repo can be used
• Release checklist
• Risk mitigation

Example Prompts

List my Martini projects.

Open my latest Martini project and summarize its scenes and assets.

Create a new Martini project for a 30-second product film about a ceramic coffee dripper.

In my current Martini project, create a scene breakdown for a moody tabletop coffee commercial.

Security Notes

• Review tool approval prompts before allowing write, upload, or generation actions.
• Do not paste Martini access tokens into chat.
• Treat signed asset URLs as sensitive while they are active.
• Use disposable projects when testing new automated workflows.

Support

Email [email protected].