MCP tool call terminates after 60s even when timeout is set to 300s

[shubhamWaghmare-sap]

Describe the bug
When using an MCP Server with long-running tools, tool invocations in Amazon Q consistently time out after ~60 seconds.
Although Amazon Q allows configuring a higher timeout (e.g., 300 seconds) as shown below, increasing this value does not take effect. The request still times out at around 60 seconds.

The error we receive is:
MCP error -32001: Request timed out

Interestingly, reducing the timeout value (below 60 seconds) behaves as expected.

To Reproduce

1. Use any MCP server built using TypeScript or JAVA MCP SDK with a tool call running for more than 60 seconds (attached an example using TS SDK server-stdio.js)
2. Configure the MCP server in Amazon Q with timeout set to 300
3. Invoke the long running tool. (From the attached example use add tool)

Expected behavior
The configured timeout value should be honored.
For example:

• With a timeout of 300 seconds, the tool should be allowed to run up to 300 seconds without timing out.
• Users should be able to safely run long-running MCP tools without premature termination.

Your Environment:

• OS: Windows
• Eclipse Version: 2026-03 (4.39), Build id: I20260226-0420
• Amazon Q for Eclipse plugin version: 2.7.3.202605010210
• JVM version: 21
• MCP SDK used: JAVA MCP SDK
• MCP Transport: Streamable HTTP (localhost)