Generic MCP
Connect a custom or third-party MCP server to Konrado.AI with the generic server integration
Use Generic MCP when you already have an MCP-compatible server or a third-party data source exposed through an MCP endpoint.
Estimated time: 10 minutes
Before You Start
- Access to Konrado.AI Integrations.
- A reachable MCP server URL, for example
https://yourdomain.com/mcp. - An API key or token accepted by the MCP server.
- A short integration name that helps your team identify the connection.
What This Connects
Konrado.AI calls the tools and resources exposed by your MCP server. The exact data depends on your MCP implementation.
MCP-Compatible
MCP means Model Context Protocol. An MCP-compatible server exposes a standard endpoint that lets Konrado.AI discover and call tools or read resources through the MCP protocol.
If your data source is a regular internal API, place an MCP server in front of it, then connect that MCP endpoint here.
Connect the MCP Server
In Konrado.AI, open Integrations > Applications, click Add Integration, and choose Generic MCP.

Fill in the form:
| Field | What to enter |
|---|---|
| Integration Name | A label for this connection, for example EU Node |
| Server URL | The MCP endpoint URL, for example https://yourdomain.com/mcp |
| API Key | The token sent to your MCP server |
The API key is sent as a bearer-style authentication token unless your MCP endpoint documents a custom header format. If your endpoint expects Basic auth or a custom header, configure the MCP server to accept Konrado.AI's token format or place a small proxy in front of it.
Click Test Connection. If the test passes, click Save.
Verify
A successful test confirms that Konrado.AI can reach the MCP endpoint, authenticate, and discover the expected MCP capabilities.
Troubleshooting
| Issue | What to check |
|---|---|
| Test Connection fails | Confirm the Server URL is reachable from Konrado.AI and points to the MCP endpoint |
| Authentication fails | Confirm the API Key matches the token expected by the MCP server |
| Data is missing | Verify the MCP server exposes the expected tools or resources |
| Private target is not reachable | Use Remote Agent when the MCP server is not publicly reachable |
Remove
Delete the Generic MCP integration in Konrado.AI and revoke the token on your MCP server.