Custom Remote MCP Server Connector

Overview

The Custom Remote MCP Server connector enables you to integrate proprietary systems and custom data sources with AI Team using the Model Context Protocol (MCP). By implementing an MCP-compliant server, you can expose any system or data source to AI teammates—whether it’s a legacy system, internal database, custom API, or third-party service not natively supported.

The connector automatically discovers and makes available whatever tools your MCP server exposes. This allows AI teammates to interact with your custom systems through natural language, executing operations and retrieving data based on the capabilities you’ve implemented in your MCP server.

Add the Custom Remote MCP Server Connector

To add the Custom Remote MCP Server connector, you configure the connection details for your MCP-compliant server, including the server URL and authentication credentials. The connector automatically discovers available tools from your MCP server once the connection is established.

Prerequisites

Before configuring the connector, ensure you have:

• An MCP-compliant server deployed and accessible
• The server URL and authentication credentials (bearer token or OAuth client ID/secret)
• Network access from AI Team to your MCP server

Configuration Steps

1. Navigate to AI Team > Connectors in the Edge Delta application
2. Find Custom Remote MCP Server connector
3. Click Configure

4. Configure the connection settings (see General Options below)
5. Click Save

Once connected, the General tab shows your connection status and the Tools tab displays tools automatically discovered from your MCP server.

General Options

Server URL

HTTPS endpoint of your MCP-compliant server. Must be a fully qualified URL.

Example: https://api.company.com/v1/mcp

Token

Bearer token for authentication. Automatically attached to requests in the Authorization header with the Bearer prefix. Use this for simple token-based authentication, or use OAuth credentials for more secure token rotation.

OAuth Client ID

OAuth 2.0 client ID for authentication. Used together with OAuth Client Secret for dynamic token management.

OAuth Client Secret

OAuth 2.0 client secret corresponding to the client ID. Stored securely and used to obtain and refresh access tokens.

Additional Headers

Custom HTTP headers included with every request to your MCP server. Useful for API versions, tenant IDs, regional routing, or custom authentication schemes.

Format: Header-Name: Header-Value (one per line)

Example:

X-API-Version: v2
X-Tenant-ID: company-abc

Tools

Tools are automatically discovered from your MCP server after connection. The Tools tab displays all available operations based on what your MCP server exposes. Tool availability depends entirely on your MCP server implementation.

How to Use the Custom Remote MCP Server Connector

The Custom Remote MCP Server connector integrates with AI Team, enabling AI teammates to leverage your custom tools based on natural language queries. Once connected, AI teammates can invoke MCP tools without requiring users to know the underlying technical details.

Use Case: Custom System Integration

When you have proprietary internal systems, legacy databases, or specialized APIs not natively supported by Edge Delta, you can create an MCP server that exposes their capabilities as tools. AI teammates can then query these systems through natural language.

For example, if you build an MCP server that connects to your internal CRM, AI teammates can search customer records, retrieve order history, or check support tickets by translating user questions into the appropriate MCP tool calls.

Use Case: Legacy System Access

Organizations with legacy mainframe systems or older databases can create MCP adapters that bridge modern AI capabilities with legacy infrastructure. The MCP server translates Model Context Protocol calls into whatever format the legacy system requires.

For example, an MCP server that wraps mainframe access can let AI teammates query batch job status, transaction logs, or inventory data without requiring specialized mainframe knowledge from users.

Use Case: Third-Party Service Integration

For vendor-specific platforms or third-party services without native connectors, build an MCP server that acts as an adapter between the MCP protocol and the vendor’s API.

For example, an MCP server for a specialized industry platform can let AI teammates check service health, monitor API rate limits, or trigger data synchronization operations through natural language requests.

Troubleshooting

Connection errors: Verify the server URL is correct and accessible from AI Team’s network. Check that your MCP server is running and that firewall rules allow traffic.

Authentication errors: Confirm bearer token or OAuth credentials are correct. For OAuth, verify the client ID and secret match your MCP server’s configuration.

Tools not appearing: Ensure your MCP server exposes tools through its discovery endpoint. Tool names are case-sensitive.

Tool execution errors: Check that tool parameters match the expected schema. Review your MCP server logs for details about failures.

Next Steps

• Learn about creating custom teammates that can use your MCP server tools
• Explore other connectors to integrate additional services

For additional help, visit AI Team Support.