Anthropic 对 Claude Code 进行了重大改进，旨在提高其在复杂开发任务中的自主性和效率。主要更新包括：新的原生 VS Code 扩展，提供更丰富、图形化的界面，可以直接在 IDE 中进行实时差异对比，并补充了刷新的终端界面，改进了状态可见性和可搜索的提示历史记录。文章还介绍了 Claude Agent SDK（以前称为 Claude Code SDK），为开发者提供了构建自定义 AI 代理的核心工具，现在支持子代理和钩子，以实现更大的定制化。关键新增的检查点（Checkpoint）机制功能可以在每次更改之前自动保存代码状态，使用户能够自信地委派长时间运行的任务，并在需要时恢复到以前的版本。结合用于并行开发的子代理、用于自动操作的钩子以及用于非阻塞进程的后台任务，这些功能共同使 Claude Code 能够更可靠地处理复杂的重构和功能探索。这些进步由 Sonnet 4.5 驱动，使 Claude Code 成为一个更有能力和集成化的 AI 开发助手。
We’re introducing several upgrades to Claude Code: a native VS Code extension, version 2.0 of our terminal interface, and checkpoints for autonomous operation. Powered by Sonnet 4.5, Claude Code now handles longer, more complex development tasks in your terminal and IDE.
Claude Code on more surfaces
VS Code extension
We’re introducing a native VS Code extension in beta that brings Claude Code directly into your IDE. You can now see Claude’s changes in real-time through a dedicated sidebar panel with inline diffs. The extension provides a richer, graphical Claude Code experience for users who prefer to work in IDEs over terminals.
Enhanced terminal experience
We’ve also refreshed Claude Code’s terminal interface. The updated interface features improved status visibility and searchable prompt history (Ctrl+r), making it easier to reuse or edit previous prompts.
Claude Agent SDK
For teams who want to create custom agentic experiences, the Claude Agent SDK (formerly the Claude Code SDK) gives access to the same core tools, context management systems, and permissions frameworks that power Claude Code. We’ve also released SDK support for subagents and hooks, making it more customizable for building agents for your specific workflows.
Developers are already building agents for a broad range use cases with the SDK, including financial compliance agents, cybersecurity agents, and code debugging agents.
Execute long-running tasks with confidence
As Claude Code takes on increasingly complex tasks, we're releasing a checkpointing feature to help delegate tasks to Claude Code with confidence while maintaining control. Combined with recent feature releases, Claude Code is now more capable of handling sophisticated tasks.
Checkpoints
Complex development often involves exploration and iteration. Our new checkpoint system automatically saves your code state before each change, and you can instantly rewind to previous versions by tapping Esc twice or using the /rewind command. Checkpoints let you pursue more ambitious and wide-scale tasks knowing you can always return to a prior code state.
When you rewind to a checkpoint, you can choose to restore the code, the conversation, or both to the prior state. Checkpoints apply to Claude’s edits and not user edits or bash commands, and we recommend using them in combination with version control.
Subagents, hooks, and background tasks
Checkpoints are especially useful when combined with Claude Code’s latest features that power autonomous work:
- Subagents delegate specialized tasks—like spinning up a backend API while the main agent builds the frontend—allowing parallel development workflows
- Hooks automatically trigger actions at specific points, such as running your test suite after code changes or linting before commits
- Background tasks keep long-running processes like dev servers active without blocking Claude Code’s progress on other work
Together, these capabilities let you confidently delegate broad tasks like extensive refactors or feature exploration to Claude Code.
Getting started
These updates are available now for Claude Code users.
- Claude Sonnet 4.5 is the new default model in Claude Code. Run /model to switch models
- VS Code extension (beta): Download from the VS Code Extension Marketplace to get started
- Terminal updates, including the visual refresh and checkpoints, are available to all Claude Code users—just update your local installation
- Claude Agent SDK: See the docs to get started
