GitHub Copilot for JetBrains has introduced two notable controls: enterprise‑wide default model selection for new conversations and an option to prevent the MCP server from starting automatically. For AI engineers, platform teams, and SREs these knobs tighten governance over model usage and allow more predictable resource consumption, while developers gain clearer diagnostics and smoother chat interactions.
Copilot model control for enterprises
Administrators can now define a managed setting that chooses any available Copilot agent model as the default when a user starts a new chat. The default is applied automatically, yet users retain the ability to pick a different model via the existing model picker UI. This gives organizations a baseline model that aligns with policy or cost considerations without removing flexibility.
MCP server startup toggle
A new configuration flag under the MCP settings lets teams disable the automatic launch of the MCP server for both Copilot and Claude. When disabled, the server only starts on explicit demand, which can be useful for controlling background processes, reducing startup latency, or complying with internal policies about when external AI services are activated.
Diagnostics, chat navigation, and account UX improvements
Diagnostic intention menus now surface a Fix action that opens an inline chat and asks the Copilot agent to repair the reported issue, preferring agent mode when available and falling back to ask mode otherwise. Chat messages are now grouped by turn, with navigation controls to jump between turns, and user messages include quick actions to copy or edit content. Account panels show clearer labels, a simplified status view, better handling of long account lists, and a dedicated Switch Account command. New‑user onboarding receives a one‑time reminder to locate the chat feature, and the sign‑in flow has a responsive layout, clearer headings, refined provider choices, a direct link to Copilot plans, and cancellable progress during browser‑based authorization.
Reliability updates and deprecation notice
The release also hardens several operational paths: language‑server startup, model and provider switching, MCP configuration, customization refreshes, file‑change detection, and worktree workflows are now more reliable. Interaction and display glitches in model management, tool‑call details, working sets, and JetBrains IDE settings have been resolved. Note that support for JetBrains IDE 2025.1 has been dropped; the plugin requires version 2025.2 or newer.
Related CloudNinjas coverage: AI engineering.
What This Means For Practitioners
Teams should audit their Copilot deployment to decide whether a managed default model aligns with governance or cost goals, and configure the MCP startup flag to match their activation policy. Verify that all developers are on a supported IDE version to avoid runtime failures. Monitor the new diagnostic Fix workflow for potential automation of recurring issues, and update internal documentation to reflect the revised account and chat navigation UI. Finally, incorporate the reliability improvements into your incident‑response playbooks, as fewer edge‑case failures should reduce noise in monitoring alerts.
