Select Language:
Here’s a simplified guide to help you troubleshoot and resolve the issue of your custom MCP tool not appearing in Amazon Connect AI Agent Designer.
First, make sure everything is set up correctly. You should have a Bedrock AgentCore Gateway created in the us-east-1 region, and it must be properly connected to your Amazon Connect instance. You also need to add a Lambda target to this gateway, along with an inline schema for your tool called “lookup_policy.” Additionally, ensure that you set the correct resource policies to allow the gateway to be invoked and create an MCP server integration using the AgentCore Gateway. Confirm that this integration is successfully associated with your Amazon Connect instance.
Once everything is set up, when you open Amazon Connect, go to AI Agent Designer, then to AI Agents, and try to add a tool. If your custom tool doesn’t show up here, your setup might be missing some key steps.
The problem most likely lies in the fact that the custom MCP tool, “lookup_policy,” is not being discovered. When you check in the browser’s network activity by opening the “Add Tool” menu, your system sends a POST request to “/mcp” requesting the list of available tools. If the custom tool isn’t listed in the response, it means it’s not being made available through the MCP namespace or gateway.
Looking at CloudTrail logs confirms that all setup operations have succeeded, such as creating the gateway, workload identity, resource policies, and the MCP integration. However, despite these success messages, the custom tool still doesn’t appear.
This issue indicates that third-party AgentCore Gateway MCP tools might not be enabled for your Amazon Connect account in the current region. The key reason is that the MCP server integration isn’t exposing your custom “lookup_policy” tool in the MCP tools/list response, even though the integration is connected to your Connect instance.
To fix this, verify that third-party MCP integration discovery is enabled for your account and region. You may need to check your configuration settings, ensure your custom schemas are correctly created and registered, and confirm that all necessary permissions are in place. After making sure the MCP discovery feature is activated, your custom “lookup_policy” tool should become visible in the AI Agent Designer, allowing you to use it as intended.
If after these checks your custom tools still don’t appear, consider reaching out to support or consulting the documentation to confirm that all integration and schema setup steps are correctly implemented.
ChatGPT
Perplexity
Gemini AI
Grok AI
