GitHub Copilot Adds Claude Agent Preview and Organization Agents to JetBrains IDEs

GitHub’s latest update to Copilot for JetBrains IDEs, which shipped on June 22, adds Claude as an agent provider in public preview. JetBrains users can now pick Claude from the agent picker in the Copilot Chat panel without switching tools.

Claude as Agent Provider

To use it, install the Claude Code CLI, then go to Settings > Tools > GitHub Copilot > Chat and point it at your Claude Code CLI path. Once configured, the agent picker in Copilot Chat shows Claude as an option alongside the default Copilot agent and the Copilot CLI agent.

One thing to know up front: the Claude agent currently runs in bypass permissions mode. That means all file edits and tool calls are automatically approved. GitHub says configurable permissions are coming in a future release. If you’re not comfortable with that tradeoff, it’s worth waiting.

For Copilot Business and Enterprise subscribers, an admin needs to enable the Editor preview features policy before users can access this.

Organization and Enterprise Agents

Admins can now publish curated agents to everyone in their organization from a central management console. When a developer opens the agent picker in their JetBrains IDE, they’ll see these organization-published agents alongside the built-in ones. No per-user setup required.

This is the JetBrains equivalent of what GitHub already shipped for VS Code, where organization agents have been available for a bit longer. Teams that have built custom agents for specific workflows — a deployment agent, a code review agent, a test writer — can now distribute them without asking everyone to configure them manually.

CLI Session Controls

The Copilot CLI inside JetBrains gets message queuing in this update. You can now send follow-up messages, steer a request mid-execution, or stop and start fresh while an agent session is still running. Previously you had to wait for the agent to finish before sending anything else.

There’s also a new “steer” option that lets you redirect an in-flight session without stopping it — useful when the agent has started going in a direction you want to course-correct without starting over.

Agent Debug Logs and Model Picker Changes

The debug panel now includes a summary view of agent activity. Instead of raw logs, you get a consolidated overview of what the agent did, which tools it called, and where time was spent. Full logs are still available.

The model picker gets a /models slash command you can run in chat to see available models, plus a recently-used section so you’re not scrolling through the full list to get back to what you had before.

Per-Turn AI Credits Indicator

Each conversation turn in Copilot Chat now shows an estimate of how many AI credits it consumed. Copilot moved to usage-based billing on June 1, so this visibility makes it easier to understand what’s driving your monthly credit usage before you hit a budget alert.

The June 22 update is rolling out to all Copilot subscribers with JetBrains IDE support. The Claude agent provider is in public preview, so expect rough edges.

Source: GitHub Changelog — New features and Claude as agent provider preview in JetBrains IDEs