Weavely’s MCP Server allows AI clients like Claude, Cursor, or others to integrate directly with Weavely’s form-generation API—using standard Model Context Protocol (MCP) tools. This guide walks you through what the Weavely MCP is, how it works, and how to configure it for any compatible AI agent.

​
What is an MCP Server?

MCP (Model Context Protocol) is a standard that allows AI models to call tools—like APIs or custom functions—by passing structured arguments. Instead of hardcoding every integration, you provide a reference to an MCP server that exposes tools dynamically. The LLM then learns what tools are available and when to use them, all without manual plumbing.

​
What Can You Do with Weavely MCP?

The Weavely MCP server currently exposes a create-form tool that lets your AI client instantly generate sophisticated, production-grade forms from a user prompt. For example:
“Create a post-purchase NPS survey for new customers using behavioral science principles.”
Behind the scenes, the MCP tool calls: 
POST https://api.weavely.ai/v1/forms/generate
and returns the url of the generated form.

​
Basic Setup

Here’s a sample configuration you can drop into your mcp.json or equivalent config file for any compatible AI client: 
{
  "mcpServers": {
    "weavely-mcp-sse": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.weavely.ai/sse"
      ]
    }
  }
}
This will load the tools served by https://mcp.weavely.ai/sse via the mcp-remote runner. Note: npx should be available in your terminal. If you’re on Windows and using a local agent, you may need to prefix commands with cmd /c.

​
Tools

create-form This MCP server currently exposes one tool:
NameDescriptionInputs
create-formCreate a new Weavely form from promptprompt (required), name (optional)

​
How It Works

Once loaded, the LLM sees create-form as a callable action and can invoke it whenever relevant in conversation. Example usage in Claude or Cursor:
“Create a bug reporting form for engineering users with required fields for steps to reproduce.”
The LLM will detect this intent, call the create-form tool, and return a full JSON definition of the form.