GitHub Copilot now resolves addressed code review comments automatically. It also suggests commit messages when applying code suggestions. Reviews can use shell tools to validate changes. Lite reviews now combine findings from multiple agents. Auto model selection adds efficiency, balance and intelligence tiers. Usage metrics for VS Code Agents are now generally available. Admins can request higher Copilot budgets when credits run out. Copilot suggests values for repository custom properties. The Copilot app now integrates Sentry for crash fixes. VS Code Agents can run in local Dev Containers. Inactive sessions can be auto-marked as Done. Pull requests can be created from Agent Host sessions.
What changed
Copilot code review left all comments open and did not suggest commit messages. Auto model selection had no cost-quality trade-off. Admins could not request higher Copilot budgets. Sentry integration was not available in the Copilot app. VS Code Agents required manual session cleanup.
What it unlocks
Resolving addressed code review comments automatically. Choosing cost, quality and speed tiers for auto model selection. Requesting higher Copilot budgets when credits run out. Fixing production crashes directly from Sentry reports in the Copilot app. Running agents in local Dev Containers. Auto-marking inactive sessions as Done. Creating pull requests from Agent Host sessions.
- three auto model selection tiers
- daily active users, session counts, user message totals
What you need to act on it
- VS Code, Copilot CLI or Copilot app for auto model selection
- Copilot Business or Enterprise with usage-based billing for budget requests
- Docker and a supported Dev Container configuration for local agents
- Copilot app for Sentry integration
- Copilot policy enabled for custom property suggestions
Sources