Integrations

MCP Server

Connect AI agents like Claude and Cursor to Demoship through the Model Context Protocol.

Demoship exposes an MCP (Model Context Protocol) server that lets AI agents and coding assistants interact with your demos programmatically.

What is MCP?

The Model Context Protocol is an open standard for connecting AI models to external tools and data sources. Demoship's MCP server lets AI agents create, edit, and manage demos through natural language.

Connecting an AI agent

Claude Desktop

Add the Demoship server to your Claude Desktop config:

{
  "mcpServers": {
    "demoship": {
      "command": "npx",
      "args": ["-y", "@demoship/mcp-server"],
      "env": {
        "DEMOSHIP_API_KEY": "your-api-key"
      }
    }
  }
}

Cursor

In Cursor settings, add a new MCP server with the same command and args shown above.

Other MCP-compatible agents

Any agent that supports MCP can connect using the @demoship/mcp-server package. The server communicates over stdio.

Authentication

  1. Generate an API key in Settings > API > Keys.
  2. Set the DEMOSHIP_API_KEY environment variable.
  3. The MCP server authenticates automatically on startup.

The API key must have at least demos:read and demos:write scopes. See API Authentication for scope details.

Available tools

The MCP server exposes these tools to connected agents:

ToolDescription
list_demosList all demos in the workspace
get_demoGet demo details including steps and configuration
create_demoCreate a new demo from a URL or specification
update_demoUpdate demo properties, steps, or annotations
get_analyticsRetrieve engagement data for a demo
create_share_linkGenerate a share link with optional access controls
list_viewersList viewers and their engagement data

Example usage

Once connected, you can ask your AI agent things like:

  • "Create a demo of our onboarding flow at https://app.example.com/onboarding"
  • "Show me the analytics for our pricing demo"
  • "Generate a share link for the product tour with a 30-day expiration"

Note: The MCP Server integration requires an Ultra or Full plan. The server respects the same permission scopes as the REST API.

On this page