MCP Server requests redirect to /session_timeout.do and health endpoint fails because the MCP service routing was not appliedIssue With the MCP Server plugin (sn_mcp_server) installed, requests to the MCP server endpoint do not reach the MCP server. Symptoms - Requests to /sncapps/mcp-server/mcp/ return a 302 redirect to /session_timeout.do. - An MCP client using OAuth cannot connect to the instance MCP server. - /sncapps/mcp-server/health fails. - The MCP service records (System Services) and a Service Endpoint (for example, https://mcps-prod-default) exist on the instance. Facts - The System Service (sys_service) record for mcp-server is present, with a Next Dish Ping value. - The mcp-server service has an associated Service Endpoint (sys_service_endpoint) record. - The health endpoint fails until the service routing is applied. Release No specific release Cause The MCP service and Service Endpoint records were present, but the backend routing for the MCP server had not been applied. As a result, requests to the MCP server endpoint did not reach the MCP server. Resolution Restart the service automation for the mcp-server System Service so the Service Endpoint is regenerated and the routing is reapplied. 1. Navigate to System Services (sys_service) and open the mcp-server record. 2. Delete the associated Service Endpoint record. Delete the Service Endpoint, not the System Service record. 3. Clear the Next Dish Ping value on the mcp-server record and save. 4. Select Refresh Connections. 5. Allow up to one hour for the workflow to complete. 6. Confirm that Next Dish Ping repopulates and a new Service Endpoint record is created on the mcp-server record. 7. Open https://<instance>.service-now.com/sncapps/mcp-server/health and confirm the health status is returned. 8. Retest the MCP client connection to /sncapps/mcp-server/mcp/.