趋势解读：New features and Claude as agent provider preview，聚焦 Agent 工作流自动化

This update adds support for organization and enterprise agents from GitHub, lets you queue and steer messages in Copilot CLI sessions, introduces a new agent debug logs summary view, and brings Claude as agent provider into public preview in GitHub Copilot for JetBrains IDEs. It also includes model picker enhancements, a per-turn AI credits indicator, and a set of user experience and reliability improvements. Cloud agent is now generally available. You can now use custom agents defined at the GitHub organization and enterprise level directly inside JetBrains IDEs. Administrators can publish a curated set of agents that are automatically available to everyone in the organization or enterprise. This makes it easier to share, standardize, and govern agent workflows across your team. Administrators define agents in GitHub: An organization or enterprise admin creates custom agents and publishes them so they are available to members. Once published, these agents are automatically distributed to all eligible users. Open the agent picker in Copilot Chat: In your JetBrains IDE, open the Copilot Chat panel and click the agent picker. Select an organization or enterprise agent: Available agents are grouped and you can choose the one that fits your task. Start working: The selected agent runs with the configuration defined by your admin, helping ensure consistent behavior and standards across your team. For more details, see Preparing to use custom agents in your organization and Preparing to use custom agents in your enterprise . When working on longer tasks in Copilot CLI sessions, you had to wait for a response to complete or previously cancel it. Now you can send follow-up messages while a request is still running. While a request is in progress, the Send button changes to a dropdown with three options: Add to Queue : Queue a message to be processed after the current response completes. Steer with Message : Tell the current request to yield once the active tool execution finishes, then process your new message immediately. Use this to redirect the agent when it’s heading in the wrong direction.