GitHub adds model selection and MCP launch controls to Copilot for JetBrains

On October 10, 2026, GitHub added two team settings to Copilot for JetBrains: enterprise administrators can choose a default model for new conversations, and users can disable automatic MCP server startup. The update lets teams set a starting configuration and control when connected tools become active.

In its October 10 announcement, GitHub said an administrator can assign any available Copilot agent model as the default for new conversations. Users can still explicitly choose a different model.

The default model does not restrict choice

The assigned model becomes the starting point for new conversations. This lets a team standardize Copilot’s initial setup while allowing developers to switch to another model in chat.

A separate MCP setting lets users disable automatic startup of servers for Copilot and Claude. These servers give models access to connected tools; teams can decide whether they should start automatically or only when a user chooses to launch them.

What else changed

GitHub also added a Fix action to the diagnostics menu. It opens inline chat and prompts Copilot to fix the detected problem. The action uses agent mode when available, or ask mode otherwise. The release also includes changes to conversation navigation, account management, and reliability, including when switching models and configuring MCP.

In the same update, GitHub ended support for JetBrains IDE 2025.1: the plugin now requires version 2025.2 or later. Teams updating Copilot should check IDE compatibility, choose a starting model for new conversations, and set policies for automatic MCP server startup.