Claude Code can now build its own cloud development environment: install dependencies, configure networking, and even securely manage API keys.

Beating AI News Flash: Anthropic has updated Claude Code's cloud development capabilities. Now, when Claude is tasked with running a GitHub project, it will automatically check for missing dependencies, identify which websites need to be accessed and which APIs need to be called, and then configure the runtime environment so it can be used directly next time.

For example, if Claude is asked to run a weather app, it will discover that the project lacks Python dependencies, network permissions for the weather API, and an API key. It will then configure installation scripts and network access rules, and request the key from the user. In the past, these steps typically required developers to complete them manually.

Key management has also been isolated. API keys provided by users are stored outside the sandbox. When Claude initiates a request, a proxy server is responsible for adding the key. This way, the model can call external services without seeing the plaintext key, and there is no need to write the key into code or environment variables.

Configured environments can be reused for subsequent cloud sessions. If the initialization script completes execution within about 5 minutes, the system will also cache the installed tools and dependencies. The next time a cloud session is started, there is no need to re-download and reinstall these components.