Azure Machine Learning - Remote

Establishes and manages persistent WebSocket and VS Code Server connections to Azure Machine Learning Compute Instances via command-palette-driven authentication flow. Uses Azure identity tokens obtained through the parent Azure Machine Learning extension to authenticate connections, maintaining session state across VS Code restarts. Implements automatic server lifecycle management on the remote compute instance with manual kill-switch commands for troubleshooting hung connections.

Executes Python scripts on remote Compute Instance with automatic workspace context injection, allowing scripts to access mounted fileshares, datasets, and workspace metadata without explicit path configuration. Implements a run-and-capture pattern that streams stdout/stderr back to VS Code terminal, providing real-time execution feedback. Scripts execute with the Compute Instance's Python environment and installed packages, inheriting all dependencies configured in the instance's conda/pip environment.

Executes Jupyter notebooks on remote Compute Instance by proxying kernel communication through the established VS Code Server connection. Implements cell-by-cell execution with output streaming back to VS Code's notebook UI, maintaining kernel state across multiple cell executions. Automatically discovers and connects to Jupyter kernels available on the Compute Instance, supporting both default Python kernels and custom conda environments configured on the instance.

Enables interactive debugging of Python code executing on remote Compute Instance by proxying debugger protocol (likely pdb or debugpy) through the VS Code Server connection. Implements breakpoint setting, step-through execution, variable inspection, and call stack navigation in VS Code's debug UI, with all debugging state maintained on the remote instance. Supports both script debugging and notebook cell debugging with automatic debugger attachment.

Provides shell terminal access to the remote Compute Instance through VS Code's integrated terminal, executing arbitrary commands (bash, PowerShell, etc.) on the instance. Implements bidirectional I/O streaming between VS Code terminal and remote shell, supporting interactive commands, environment variable access, and file operations. Terminal inherits Compute Instance's environment configuration, including PATH, conda environments, and mounted fileshares.

Enables git operations (clone, pull, push, branch management) on remote Compute Instance through VS Code's source control UI, with automatic integration to workspace-mounted repositories. Implements git command proxying through the remote shell, supporting both HTTPS and SSH-based authentication. Provides visual diff and merge conflict resolution in VS Code's UI while maintaining repository state on the Compute Instance.

Provides read/write access to the remote Compute Instance's filesystem through VS Code's file explorer, enabling browsing, opening, editing, and deleting files on the instance. Implements file synchronization between local VS Code editor and remote filesystem, with automatic conflict detection if files are modified externally. Supports access to mounted Azure fileshares and datasets through the Compute Instance's filesystem mount points.

Integrates with Azure Machine Learning Studio web UI through click-out links and 'Edit in VS Code' buttons, enabling one-click connection to Compute Instances from Notebook and Compute tabs. Implements deep linking from Azure ML Studio to VS Code with automatic connection establishment, eliminating manual workspace/instance selection. Provides inline VS Code launch button on Compute Instance cards in Azure ML Studio UI.

Provides manual kill-switch commands to terminate hung or unresponsive VS Code Server and WebSocket servers on remote Compute Instance, enabling recovery from connection failures without restarting the entire instance. Implements server process termination through Azure ML backend APIs, with automatic server restart on next connection attempt. Supports troubleshooting scenarios where servers become unresponsive due to network issues or resource exhaustion.

Collects usage telemetry data on extension interactions (connection events, command execution, errors) and sends to Microsoft servers for product improvement analysis. Implements opt-out mechanism through VS Code's standard `telemetry.enableTelemetry` setting, allowing users to disable telemetry collection globally. Respects user privacy preferences without requiring extension-specific configuration.