AI 코딩 도구,
이번엔 뭐가 바뀌었나

• ! Windows 환경에서 Local harness의 Windows sandboxing을 더 이상 지원하지 않으며, 대신 Agent Host 세션을 사용하도록 변경한다 # 보안·권한
  원문

  [Chat] Windows sandboxing for Local harness is no longer supported: Windows sandboxing for terminal commands in Local agent sessions has been deprecated. For sandboxed command execution on Windows, use an Agent Host session. See the Agent Host sandboxing documentation for details.

• ~ Local harness에 대한 Windows sandboxing 지원을 중단한다 # 보안·권한
  원문

  [Deprecated features and settings] New deprecations in this release: Windows sandboxing for Local harness is no longer supported

• + Copilot harness를 새 세션의 기본 harness로 점진적으로 도입한다 # 에이전트
  원문

  [Agents] Copilot harness: The Copilot harness is powered by the Copilot SDK, a shared foundation that helps align behavior, capabilities, and improvements across Copilot products. In VS Code, the Copilot harness runs sessions through the Agent Host, which lets sessions continue in the background and be accessed from both the VS Code editor window and the Agents window. The Copilot harness has been available from the harness picker in the chat input. Based on overwhelmingly positive and valuable feedback, we're progressively rolling it out as the default harness for new sessions in VS Code. When you start a new session, Copilot in the harness picker indicates that the session will use the Copilot harness. Thank you to everyone who has shared feedback. We're actively using it to shape the experience. One example is reintroducing Ask as a built-in agent in the Copilot harness. Your feedback showed

• + 에이전트와 협업을 위한 인터랙티브 인터페이스인 Canvases를 프리뷰로 제공한다 # 에이전트
  원문

  [Agents] Canvases (Preview): Setting: chat.canvases.enabled Canvases are interactive interfaces to collaborate with an agent. A canvas opens beside the chat, where you and the agent can inspect and update the same artifact instead of coordinating solely through prompts. For example, use a Kanban canvas to plan and track implementation work. You can create and move cards directly, while the agent can break a feature into tasks, update their status, or add newly discovered work. Canvases work in both the VS Code editor window and the Agents window, and canvas extensions are also supported in the GitHub Copilot app. Canvases are available in Preview in the VS Code desktop app for local sessions that use the Copilot harness. Enable chat.canvases.enabled , and then start a new session with the Copilot harness. Every canvas is provided by a canvas extension. To get started, you have two options: Crea

• + Copilot 및 BYOK 제공자의 모델 선택기를 재설계하여 모델 선택과 전환을 용이하게 한다 # UI
  원문

  [Chat] Redesigned model picker: Copilot and bring-your-own-key (BYOK) providers offer many models, each with strengths for different tasks. The redesigned model picker makes it easier to choose and switch models as your work changes. The model picker uses a tabbed interface to separate Copilot and BYOK models. When you hover over a model in the list, you can further configure its thinking level and context window size.

• ~ Python Environments Extension에서 환경 탐색, 인터프리터 선택, 터미널 활성화 관련 문제를 수정한다 # 에이전트
  원문

  [Languages] Python: This release includes fixes for environment discovery, interpreter selection, and terminal activation in the Python Environments Extension. Environment discovery: Python installations managed by uv now appear under Global. Selecting an interpreter by path adds its virtual environment to the environment list, and deleting a virtual environment outside VS Code updates the selection. Project-specific environments: Automatic selection now prefers a project's own virtual environment over nested environments. Conda environment selection and Poetry package operations respect project scope, and background refreshes no longer overwrite newer interpreter selections. Interpreter configuration: Fixed resolution of ${workspaceFolder}, relative paths, and commands such as python3 in python.defaultInterpreterPath. The project-view environment selector now includes Enter Interpreter Path. Term

• + Enterprise 환경에서 정책 기반의 채팅을 Copilot Agent Host로 자동 이동시키고 정책을 강제하는 기능을 추가한다 # 보안·권한
  원문

  [Enterprise] Automatically move policy-managed chats to Copilot Agent Host: Enterprise managed settings let administrators govern agent permissions and sandboxing. When your organization configures a nonempty permissions object or sets sandbox.enabled to true, VS Code now requires the Copilot harness in Agent Host for full chat sessions, to be able to enforce these policies. Opening an existing, unarchived Local chat automatically moves it to Copilot, preserving the transcript and unsent draft, including attachments. Nothing is sent automatically. The new chat uses its normal model and permission defaults under current organization policy. VS Code archives the original chat only after the new chat opens successfully. If you switch away while the move is in progress, the original stays unchanged and the unopened destination is removed. While these policies apply, Claude, Codex, and terminal chat are

• + Windows에서 숫자형 정책이 DWORD 및 QWORD 값을 지원하도록 개선한다 # 보안·권한
  원문

  [Enterprise] Windows numeric policies support DWORD values: On Windows, policies that accept numeric values support DWORDs and QWORDs.

• + 현재 채팅을 다른 폴더나 Git worktree로 이동하는 기능을 추가한다 # 에이전트
  원문

  [Agents] Move chats between folders and worktrees: Previously, if you started investigating a task in one folder and then wanted to implement it in a Git worktree, or continue in a different repository, you had to start a new chat and rebuild the context. In the Agents window, you can now ask the agent to move the current chat to another folder or into a new Git worktree, and it keeps the conversation history intact. Changing the workspace also works for chats that already have a workspace, and the agent can isolate work in a new worktree. Ask the agent with a natural-language prompt, for example: The agent finishes the current turn, moves the chat, and shows a notice when the workspace changes. Then it continues the task automatically in the new location. Edits and commands run in the new location while your original checkout stays unchanged. The move is available for agent sessions that use t

• + Copilot harness 세션에서 실행 중인 백그라운드 셸을 중단하는 기능을 추가한다 # 에이전트
  원문

  [Chat] Stop background shells: In Copilot harness sessions, the background shells pill above the chat input lists the commands that the agent is still running in the background. When one of those commands is no longer useful, for example a test run that you want to restart or a command that runs longer than expected, you can stop it instead of waiting for it to finish. Select the Background Shells pill, select a shell to open its details, and then select Stop Shell next to its command. Only that command ends, and the shell leaves the list. You can reach Stop Shell with the keyboard, and screen readers announce when the shell stops. Stopping a shell is only available for attached shells, which the agent runs as part of its current request. Detached shells run independently of the agent and can't be stopped from this interface. Shell details also show each shell's full description above its com

외 12개 항목은 원문에서 볼 수 있습니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

• + Copilot harness를 통해 새로운 에이전트 기능을 추가한다 # 에이전트
  원문

  [Agent loop] Copilot harness: The Copilot harness adds exciting new agent functionality to VS Code. It's powered by the Copilot SDK, so its behavior and capabilities are consistent with other Copilot products, including the standalone GitHub Copilot app and the Copilot CLI. To start using the Copilot harness, select it from the harness picker in the chat input. In this release, it might already be the default selection for you. You can continue working as you normally do. The harness runs in a dedicated agent host process based on the Agent Host Protocol (AHP), so you can connect to the same agent session from multiple VS Code windows. Learn how to work with the Copilot harness, or explore the architecture and workflows in the agent host blog post. If you have feedback or requests, please file an issue.

• ◆ Copilot 에이전트 호스트에서 `chat.agent.sandbox.enabled` 설정을 통해 샌드박싱을 지원한다 # 보안·권한
  원문

  [Agent security] Sandboxing in the Copilot agent host: Settings: chat.agent.sandbox.enabled Sandboxing gives you more control over what Copilot can access while working on a task. It limits how supported agent operations access files and network resources, and helps to reduce the impact of model mistakes, prompt injection, untrusted dependencies, and locally launched tool servers. Depending on the operation and platform, these restrictions use operating-system protections or checks within the agent host process. Sandboxing adds a layer of protection, but does not replace endpoint security or provide a standalone security boundary. Sandboxing is available on Windows, macOS, and Linux. On Windows, required operating-system updates must be installed. Refer to Check platform availability for more details. To configure sandboxing for all your sessions, enable the chat.agent.sandbox.enabled setting. You can

• + 여러 개의 GitHub Enterprise 인스턴스에 동시에 로그인할 수 있도록 `github-enterprise.uris` 설정을 추가한다 # 보안·권한
  원문

  [Authentication] Sign in to multiple GitHub Enterprise instances: Setting: github-enterprise.uris Many organizations use GitHub Copilot through a GHE.com account but keep their code on GitHub Enterprise Server (GHES). The github-enterprise.uri setting held a single instance, so you couldn't be signed in to both at once. For example, you couldn't use Copilot with your GHE.com account and review pull requests on GHES in the same VS Code window. The new github-enterprise.uris setting takes a list of GHE.com and GHES instances: The GitHub Enterprise authentication provider holds accounts from every instance in the list. Each account is labeled with its host, such as monalisa - octocat.ghe.com, so you can tell them apart in the Accounts menu. When more than one instance is configured, signing in asks which instance to use. The order of the entries doesn't select a default instance. To check or change which

• + 비활성 에이전트 세션의 워크트리에서 저장 공간을 회수하는 기능을 추가한다 # 성능
  원문

  [Agent loop] Reclaim storage from inactive worktrees: Run Chat: Open Worktree Cleanup to see how much disk space inactive agent session worktrees use and remove the ones you no longer need. Filter sessions by how long they have been inactive, review each worktree's size, and select the worktrees to clean up. The cleanup editor also lets you configure automatic cleanup for sessions with merged pull requests and control whether VS Code suggests cleanup when session storage grows large. When chat.agentSessions.sessionStorageCleanupSuggestion.enabled is enabled, the Agents window notifies you when enough inactive worktrees accumulate or they use enough disk space to make cleanup worthwhile. Cleaning up marks the selected sessions as done and deletes their worktrees. You can restore a session later to recreate its worktree. Active, running, needs-input, and pinned sessions are protected from cleanup.

• + 백그라운드에서 실행되는 셸을 추적하는 기능을 추가한다 # 에이전트
  원문

  [Agent loop] Keep track of background shells: When the agent runs a command in the background, it keeps working on other steps while the command runs. In Copilot harness sessions, a Background Shells pill above the chat input lists the shells that are still running, in both the Agents window and the editor window. Select the pill to see each shell's command and how long it has been running, and select a shell to watch its output stream as the command runs, if output is available. A shell leaves the list when its command finishes.

• + 에이전트가 트리거하는 메시지에 대한 제어 기능을 강화한다 # 에이전트
  원문

  [Agent loop] More control for agent-triggered messages: When agents delegate work across conversations, they can adjust the delegated task as requirements change instead of waiting for the other session to finish. The send_message tool supports these actions between sessions and nested sessions on the same agent host: Steer an active conversation with a correction. Queue a follow-up as a separate turn. Replace a queued message while preserving its place in the queue. Cancel a queued message before processing begins. Queued messages run in order. After a message starts processing, it can no longer be replaced or cancelled.

• + Agents 윈도우에서 세션을 그리드 형태로 배치한다 # UI
  원문

  [Agent UX] Arrange sessions in a grid: Compare results, monitor long-running tasks, or work across related conversations without repeatedly switching between sessions. Like in the editor window, the two-dimensional session grid in the Agents window lets you drag sessions into horizontal or vertical splits and resize the panes. Maximize a session when you need to focus, then return to the grid to continue working across conversations.

• + 새 세션 제어 항목을 간소화하고 세션 옵션 트레이를 추가한다 # 에이전트
  원문

  [Agent UX] Simplified new-session controls (Experimental): Settings: sessions.chat.experimental.newSessionComposerLayout , sessions.chat.unifiedWorkspacePicker.enabled (Agents window only) The experimental new-session composer keeps your prompt at the center while making session choices easier to review. Workspace, repository, and harness controls now appear in a collapsible Session Options tray above the chat input. Getting-started tips now appear in a rounded notice below the chat input, so they no longer separate the session controls from your prompt. Collapse the tray to reduce visual clutter without changing your selections. The composer remembers your choice, adapts labels as the window narrows, and keeps the options expanded when screen reader optimized mode is active. When you use a custom chat background, the tray has an opaque surface to keep its controls readable.

• + 재로드 없이 로컬 외부 Copilot 세션을 이어받는다 # 에이전트
  원문

  [Agent environments] Continue local external Copilot sessions without reloading: New conversations from the Copilot CLI and GitHub Copilot app are now picked up automatically as external sessions as soon as you send the first request, even when the Agents window is already open. You can open and continue these conversations directly from the Agents window. Use the Created In and Created Externally filters to choose which conversations appear in the sessions list.

• + ChatGPT 앱 또는 Codex CLI에서 시작한 Codex 채팅을 VS Code에서 이어받는다 # 에이전트
  원문

  [Agent environments] Codex chat handoff between apps: Pick up a Codex chat started in the ChatGPT app or Codex CLI and continue it in VS Code's Agents window on the same machine, with your conversation history intact. You can switch back to the ChatGPT app or Codex CLI later and continue the same chat. To use your GitHub Copilot subscription in VS Code, choose a Copilot model in the model picker. For example, you can keep working with Copilot after reaching your ChatGPT usage limit. With Codex enabled and Created Externally sessions visible, chats created or updated in ChatGPT or Codex CLI appear within a few seconds without reloading VS Code. Only one application can send messages to the chat at a time. If ChatGPT or Codex CLI still has the chat open, a This chat is open in another app banner in VS Code explains why sending is unavailable. Fully quit ChatGPT or exit the Codex CLI session, then select Retr

외 20개 항목은 원문에서 볼 수 있습니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

• + Copilot harness를 통해 새로운 에이전트 기능을 추가하고 여러 VS Code 창에서 동일한 에이전트 세션에 연결할 수 있게 한다 # 에이전트
  원문

  [Agents] Copilot harness: The Copilot harness adds exciting new agent functionality to VS Code while keeping the way you work familiar. It's powered by the Copilot SDK, so its behavior and capabilities are consistent with other Copilot products, including the standalone GitHub Copilot app and the Copilot CLI. To start using the Copilot harness, select it from the harness picker in the chat input. In this release, it might already be the default selection for you. You can continue working as you normally do. The harness runs in a dedicated agent host process based on the Agent Host Protocol (AHP), so you can connect to the same agent session from multiple VS Code windows. Learn how to work with the Copilot harness, or explore the architecture and workflows in the agent host blog post. If you have feedback or requests, please file an issue.

• + HydraFusion 모델 오케스트레이션 시스템을 도입하여 작업에 따라 모델과 워크플로우를 자동으로 선택하게 한다 # 에이전트
  원문

  [Agents] HydraFusion model orchestration (Research Preview): HydraFusion is an adaptive model orchestration system that chooses the models and workflow for each coding task. It can solve a task with one model, escalate to a stronger model, or have another model critique and revise the result. This approach aims to improve result quality while balancing speed and cost, without requiring you to coordinate models yourself. Eligible users with preview features enabled can select HydraFusion in the model picker.

• + Agents window에서 연결된 원격 에이전트 호스트로 작업을 위임하는 기능을 추가한다 # 에이전트
  원문

  [Agents] Delegate tasks to remote agent hosts (Experimental): Settings: chat.remoteAgentHostsEnabled , chat.remoteSessions.tools.enabled (Agents window only) Your agent can delegate work to connected remote agent hosts from the Agents window, without you selecting a host in a picker for each task. The new built-in tools let your agent: Discover hosts, models, resource capacities, and session load with list_agent_hosts. Start a session with create_remote_session. Specify a host, or let automatic placement match an operating system (Windows, Linux, or macOS), minimum memory capacity, logical CPU count, and optional model. Among matching hosts, placement favors the host with the fewest running sessions and pending creations. Check a remote session's status and latest response with get_remote_session. Send follow-up work or report results and questions back to the originating chat with send_remote_

• + 관리자가 AI 기능 사용 전 최소 버전 요구사항을 설명하도록 하며, 업데이트가 필요한 경우 적절한 안내를 제공한다 # 보안·권한
  원문

  [Enterprise] Explain minimum version requirements: Administrators can require developers to update VS Code before using AI features, for example, to adopt newer sandboxing protections. Instead of an Update Required dialog, VS Code now explains the requirement wherever AI features are unavailable: Chat: shows the required and installed versions, with an update action. Editor window: shows a banner even when Chat is closed. Other editor features stay available. Agents window: shows a blocking overlay, with an Open Editor Window action. When an update is available, VS Code offers the matching Check for Updates, Download Update, Install Update, or Restart to Update action. If your organization disables built-in updates by policy, VS Code instead asks you to contact your administrator. The notices clear once the requirement is met. How VS Code enforces the requirement is unchanged.

• + 멀티 폴더 세션 기능을 통해 여러 리포지토리나 격리된 워크트리에서 관련 작업을 한 세션에서 조정할 수 있게 한다 # 에이전트
  원문

  [Agents] Multi-folder sessions (Experimental): Multi-folder sessions let you coordinate related work across repositories or isolated worktrees in one session. Previously, every chat in a multi-chat session shared the same folder and checkout. In this release, each chat can use its own folder or worktree without changes leaking between chats. Each chat uses its folder for its terminal, tasks, changes, pull request, and Agent merge state. Chats that use the same folder share that state. In the sessions list, hover over a session to see a summary across all its folders, or hover over a nested chat to see details for only its folder. Example use cases include: Implement a feature across repositories: Ask the main chat to work in one repository and create a peer chat for a second repository. Compare approaches in separate worktrees: Ask the main chat to create peer chats that each use a fresh worktr

• + 세션 내에서 작업이 완료된 채팅을 마크하고 보관하여 세션의 나머지 부분을 활성 상태로 유지하게 한다 # 에이전트
  원문

  [Agents] Archive chats in a session: Keep the rest of a session active when one chat finishes its work. Because sessions can contain multiple chats, select Mark as Done to archive only the finished peer chat, not the entire session. To restore it with its title and transcript intact, select the Done filter in the sessions list.

• + 위임된 작업 시 세션 계획 또는 결과물 여부와 워크스페이스 및 워크트리 선택을 분리하여 결정하게 한다 # 에이전트
  원문

  [Agents] Choose the session relationship and workspace for delegated work: Make two separate decisions when you ask an agent to delegate work. First, decide whether the task belongs to the current session's plan or deliverable. Then specify a workspace only if the task needs repository files, and request a worktree only if file changes need isolation. Related work can still use a different workspace. Create a peer chat for related work: Ask the agent to create a peer chat in the current session. To reuse the current workspace and checkout, say so in your prompt. Because the peer chat shares the checkout, use this option for research, planning, or related work that doesn't need isolated file changes. Create an independent session for unrelated work: Ask the agent to create an independent session with no workspace when a separate deliverable doesn't need repository files. The new session doesn't

• + Agents window에서 새 세션 생성 시 선택적 환영 문구와 사용자 이름을 포함하는 기능을 추가한다 # 에이전트
  원문

  [Agents] Personalize new-session welcome phrases (Experimental): Settings: sessions.chat.experimental.welcomePhrases , sessions.chat.experimental.welcomeName , accessibility.verbosity.newSessionWelcome (Agents window only) Add some personality to the Agents window with an optional welcome heading when you create a new session in the Agents window. The heading shows one of five phrases, such as What are we building? or Let's ship something, and can append your first name from your GitHub profile. Welcome phrases are off by default. When enabled, you can use Set Welcome Name from the heading, its context menu, or the Command Palette to choose a different name. The name syncs across devices. Clear it to use your GitHub first name when available, or a generic phrase when your profile has no name. When screen reader optimized mode is active, VS Code announces the heading once when the composer appea

• + 세션 컴포저 레이아웃을 개선하여 제어 항목을 그룹화하고 단축키를 추가한다 # 에이전트
  원문

  [Agents] Improved session composer controls (Experimental): Settings: sessions.chat.experimental.newSessionComposerLayout , sessions.chat.unifiedWorkspacePicker.enabled (Agents window only) The experimental session composer layout reduces the need to move between separate parts of the Agents window. In a new session, the workspace, branch, worktree, and harness controls are grouped above the chat input. Enable both settings to use the layout. The unified workspace picker also has commands for moving directly to its controls. Use ⌘K ⌘F (Windows, Linux Ctrl+K Ctrl+F) to focus the workspace picker or ⌘K ⌘H (Windows, Linux Ctrl+K Ctrl+H) to focus the harness picker. Hovers show your configured keybindings, and each picker's context menu provides a Configure Keybinding action. These keyboard improvements require only the unified workspace picker, not the experimental composer layout. Starting with t

• + 에이전트 오케스트레이션의 한도를 높여 복잡한 워크플로우가 중단되지 않도록 개선한다 # 에이전트
  원문

  [Agents] More reliable agent orchestration (Experimental): Setting: chat.agentHost.agentOrchestrationLimits Coordination-heavy agent workflows are less likely to stop before the work is complete. This release raises the process-wide limits for agent-created sessions, chats, inter-session messages, and recursive session creation. For example, an agent can classify CI failures, delegate each fix to a separate session, and coordinate those sessions without exhausting the message limit before the work is complete. Reaching a limit blocks new orchestration actions but does not interrupt work that is already running. The setting defaults to on, and changes apply without restarting the Agent Host. Set it to off to remove the orchestration limits. Confirmation requirements and input validation continue to apply.

외 16개 항목은 원문에서 볼 수 있습니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

변경 항목 12개 · 요약이 만들어지면 여기에 항목별 요약이 나옵니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

변경 항목 12개 · 요약이 만들어지면 여기에 항목별 요약이 나옵니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

변경 항목 11개 · 요약이 만들어지면 여기에 항목별 요약이 나옵니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

변경 항목 15개 · 요약이 만들어지면 여기에 항목별 요약이 나옵니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

변경 항목 8개 · 요약이 만들어지면 여기에 항목별 요약이 나옵니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

변경 항목 7개 · 요약이 만들어지면 여기에 항목별 요약이 나옵니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

변경 항목 6개 · 요약이 만들어지면 여기에 항목별 요약이 나옵니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

변경 항목 11개 · 요약이 만들어지면 여기에 항목별 요약이 나옵니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.

변경 항목 8개 · 요약이 만들어지면 여기에 항목별 요약이 나옵니다.

AI가 만든 요약입니다. 원문과 다르면 오류 제보해 주세요.