This topic explains how administrators can leverage the Skope IT application to monitor Model Context Protocol (MCP) communications in real-time. You will learn how to gain granular visibility into the entire interaction lifecycle—from initial handshakes and protocol versioning to specific tool invocations and resource requests—to ensure organizational compliance and detect potential threats. Furthermore, it explains how to analyze detailed event logs and initialization data to verify successful executions while keeping sensitive information protected.
An administrator can create an RTP Policy to block a specific MCP Server. Create a new policy. In the application select an MCP Server. For example “Notion MCP”. Set the Profile & Action to Block. Give it a name and save. Apply the changes.
The list of MCP Servers shown in the application drop down are remote MCP Servers listed in the MCP Servers section of the App Catalog.
Policy Configuration Interface: The “Create Policy” screen. The Source dropdown is set to “Application” and the Application is set to “Notion MCP”. Not selecting any activity applies the policy to all activities. “Profile & Action” field is set to Block.
To block a specific activity of an MCP Servers, in the Destination section select the specific Activity. The supported list of MCP activities are:
- CallToolRequest
- CallToolResult
- CreateMessageRequest
- CreateMessageResult
- ElicitResult
- GetPromptRequest
- GetPromptResult
- InitializeRequest
- InitializeResult
- ListPromptsResult
- ListResourcesResult
- ListResourceTemplatesResult
- ListToolsResult
- ReadResourceRequest
- ReadResourceResult
Then go to VS Code and start the Notion MCP server. The user will get a response that this is a non-compliant action.
However if a different server e.g., Github Copilot MCP server is started it doesn’t get blocked. You can see this in the application events where Notion MCP server was blocked but not Github Copilot.
