Custom Connectors

The Docusign MCP (Model Context Protocol) server enables AI-powered tools to interact with Docusign’s agreement management workflows directly from within an AI assistant session.

Docusign provides native connector support for selected frontier model platforms like Claude, ChatGPT, etc. Refer Connectors page for more details.

Using Docusign MCP Server with Custom and Unsupported AI Clients

Organizations using alternative LLMs including locally hosted models, proprietary enterprise solutions, or custom AI tooling can still integrate with the Docusign MCP server through an MCP-compatible client supported by their platform. This guide is intended for developers who:

• Are using an LLM or AI assistant platform not yet officially supported by Docusign’s MCP integration.
• Are building a custom AI agent or orchestration layer that needs to call Docusign MCP tools.
• Need to set up MCP client configuration manually in their environment

Prerequisites

Before connecting to the Docusign MCP server, ensure you have the following ready:

• Docusign Developer Account: You can create one for free at the Developer Center.
• Integration Key: See Configure your app to add an Integration Key.
• Client Secret: See Configure your app to create a secret key.
• Redirect URI: Refer Register a Redirect URI in your Docusign App section below.
• OAuth Access Token: If you are using an MCP host configuration .json file to configure OAuth access, generate a token using one of the following authentication flows:
  • Authorization Code Grant (ACG): Default token lifespan is 8 hours. You must refresh your token manually if automatic refresh is not supported by your AI tool. See Generating an ACG Access Token.
  • JSON Web Token (JWT): Default token lifespan is 1 hour. See Generating a JWT Access Token.

Register a Redirect URI in your Docusign App

If your AI tool provides a redirect URI while configuring the MCP server, then register the redirect URI in your Docusign app.

1. Sign in to the Docusign Admin.
2. Go to Apps and Keys and select your app.
3. Under Authentication, add the redirect URI you intend to use (for local development this is typically http://localhost:3000/callback or a value specified by the MCP server’s auth flow)
4. Save your changes.

Method 1: Configuring Parameters using UI

If your AI tool has the MCP configuration to be added via UI, use the following list of values to connect the Docusign MCP server. Refer to your tool’s documentation for the detailed steps.

Method 2: Setting Up the MCP Server via MCP Host File

If your AI tool cannot be configured via UI, locate your AI tool or agent framework’s MCP host configuration file (typically a JSON file such as mcp.json, claude_desktop_config.json, or your platform’s equivalent). The exact filename and path vary by platform. Refer to your tool’s documentation for the exact location.

Docusign provides a hosted MCP server at https://mcp.docusign.com/mcp (production). Use this configuration for all MCP-compatible tools without any local installation and replace your token value

{
  "mcpServers": {
    "DocusignMCP": {
      "httpUrl": "https://mcp.docusign.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_AUTH_TOKEN"
      }
    }
  }
}

Authentication is handled via OAuth at the connection level. Your AI tool will redirect you through the Docusign OAuth consent flow on first use. No environment variables are required.

Reload or Restart Your AI Tool

After saving the configuration file, restart or reload your AI client so it picks up the new MCP server definition. Ensure you stop all the services, threads, and instances of the AI client and restart your AI service.

Your AI tool should now list Docusign MCP as an available MCP server. How the restart is performed depends on your platform:

• Embedded agent runtimes: Reload the agent configuration or restart the process
• IDE plugins with MCP support: Use the plugin’s reload/refresh command for MCP servers.