Anthropic announced the launch of Claude Code on the web in research preview, extending its popular coding assistant beyond the command line. The new interface allows developers to write, test, and manage code directly in their browser, combining Claude’s reasoning and automation capabilities with secure, cloud-based execution. Users can connect repositories, assign tasks, and track progress in real time—all without leaving their workspace.

• Browser-native experience: Users can connect GitHub repositories and assign tasks, with Claude Code now able to handle multiple instances concurrently while showing real-time progress.
• Secure isolation: Each coding session runs in a dedicated sandboxed environment, ensuring no cross-task file or network access outside its designated workspace.
• Multi-device access: Pro and Max subscribers can access Claude Code via the browser or iOS app, enabling cross-platform task management and automatic pull requests once work completes.

Why it matters: The shift from terminal-based to web-based coding marks a major accessibility step for Anthropic’s developer tools. It brings Claude Code closer to mainstream cloud IDEs, offering a lightweight and secure alternative for those who prefer guided, conversational programming over traditional scripting. For teams, it means developers can monitor and delegate coding workloads from anywhere, while less experienced users gain a low-friction way to experiment with automation. The move reinforces Anthropic’s strategy of blending reasoning models with real-world developer workflows—merging AI copilots and agentic tasking into one browser-native tool.

Full announcement: Claude Code on the Web – Anthropic.