Copilot gets stuck in a cancelling loop after failing to find MCP config

[tarikbrown-msft]

Bug description

When Copilot CLI attempts to locate an MCP configuration (e.g., via /mcp slash command or related logic) and fails, it can enter a repeated/cancelling loop in the terminal. After searching various default locations for the MCP config, if nothing is found, the tool sits in a 'Cancelling' state and fails to exit gracefully—even after the operation is manually aborted.

Steps to reproduce

1. Run a Copilot CLI command that depends on an MCP config in an environment where no such config is available.
2. Observe Copilot searching for config locations (VS Code workspace, user settings, roaming profiles, etc.).
3. If the config is not found, abort or let the operation fail.
4. See that the CLI enters a persistent 'Cancelling' state and does not exit cleanly.

Expected behavior

• Copilot CLI should recognize when no MCP config can be found and exit gracefully, reporting a clear error message.
• Cancelling should actually terminate the operation with no loop.

Actual behavior

• Operation appears to be cancelled by the user, but CLI gets stuck in a 'Cancelling' loop and does not return to a usable prompt.

Screenshot

Environment

• Copilot CLI version: (see bottom-left of screenshot, e.g., v1.0.16)
• OS: Windows
• Shell: PowerShell

Additional context

The terminal output shows:

• Copilot searching for MCP config locations.
• Operation aborted by user.
• "Operation cancelled by user."
• Persistent "Cancelling" loop.

This behavior blocks further CLI usage until the terminal session is reset.

Please fix CLI error handling for missing MCP config so it exits gracefully.