Editorial illustration for Anthropic Debuts Claude Code: Cloud Coding Tool Enters Browser Beta
Anthropic's Claude Code: Browser Tool Transforms Coding
Anthropic's latest product is a web portal for telling its AI to write code. This is not a subtle market entry. Claude Code, now in a closed beta, is a direct shot across the bow of GitHub Copilot and the rest of the AI coding crew.
The pitch is simple. You point it at a GitHub repo from your browser, describe a bug or a feature, and it tries to fix it. No terminal, no local environment.
The work happens in Anthropic's cloud, in a sandbox. The company says you can watch it work and steer it, like a very expensive backseat driver.
Anthropic has launched Claude Code on the web, a browser-based tool that allows users to assign and run coding tasks directly in the cloud. The feature is currently in beta as a research preview for Pro and Max users. Claude Code on the web is available at claude.com/code.
The company’s documentation details rate limits, security configurations, and sandboxing architecture. According to Anthropic, the new interface enables developers to delegate coding tasks directly from their browser without opening a terminal. Users can connect GitHub repositories, describe the required changes, and Claude will handle implementation through Anthropic-managed infrastructure.
Each task runs in an isolated cloud environment with real-time progress tracking. “You can actively steer Claude to adjust course as it’s working through tasks,” the company said. The system also supports multiple sessions running in parallel across repositories, with automatic pull request (PR) creation and change summaries.
The web tool is designed to complement existing Claude Code workflows, with particular use cases in bug fixes, backend updates, and answering questions about repository structure. Anthropic said that Claude Code applies test-driven development to verify backend changes. A mobile version of Claude Code is also available through Anthropic’s iOS app as part of the preview.
It's a research preview. Access is limited to paying Pro and Max tier users, a classic velvet rope strategy. The documentation talks about rate limits and sandboxing architecture, which is corporate speak for "we know this could go very wrong and are trying to contain it."
Anthropic is framing this as a complement to existing tools. The real test is whether a developer will trust a cloud sandbox to make changes to a live repository. The promise of automated PRs and parallel sessions is compelling for tedious refactoring work. The threat of an AI misunderstanding a codebase and creating a cascading architectural mess is equally real.
For now, it's a controlled experiment. The success metric won't be buzz. It will be whether any engineering team, six months from now, relies on it for their actual deploys.
Further Reading
Common Questions Answered
What unique features does Claude Code offer for developers?
Claude Code is a browser-based tool that allows developers to assign and run coding tasks directly in the cloud. The tool aims to provide a more integrated and seamless coding experience by embedding directly into web browsers, addressing previous limitations of traditional AI coding assistants.
How can developers access the Claude Code beta version?
Claude Code is currently available as a research preview exclusively for Anthropic Pro and Max users through claude.com/code. The beta version provides developers with a cloud-based coding environment and includes detailed documentation on rate limits, security configurations, and sandboxing architecture.
What is Anthropic's strategic approach with the Claude Code launch?
Anthropic is carefully testing the waters by launching Claude Code as a limited beta for Pro and Max users, signaling a strategic push into developer tools. The browser-based tool represents the company's effort to simplify task delegation and create a more intuitive AI-powered coding assistance platform.