Issue: Task Tool Fails When Launching Sub-Agents

Description

When attempting to use the task tool to launch any sub-agent (e.g., code-reviewer), the tool consistently fails with the following error message:

{"task_response": {"content": "Tool execution aborted", "name": "task"}}

And the underlying error observed is:

undefined is not an object (evaluating 'Q.model')

Steps to Reproduce

1. Attempt to call the default_api.task() function with any valid subagent_type and prompt.
   • Example:

     print(default_api.task(description="Review code", prompt="Review this script", subagent_type="code-reviewer"))

Expected Behavior

The task tool should successfully launch the specified sub-agent without internal errors.

Actual Behavior

The task tool aborts execution, and an "undefined is not an object (evaluating 'Q.model')" error occurs internally, preventing the sub-agent from running.

Impact

This issue prevents the use of any sub-agent functionality, limiting the capabilities of the model for complex tasks that rely on these specialized agents.