The Troubleshooting MCP Server is available during active AL debugging sessions in Visual Studio Code and becomes accessible when execution is paused at a breakpoint or runtime error. In this state, GitHub Copilot, or other coding agents, can use the server to analyze the current execution context and provide insights based on real runtime data rather than static code inspection.

Through the Troubleshooting MCP Server, coding agents can retrieve and analyze the full call stack to explain how execution reached a specific point, including scenarios involving deep or complex call chains across multiple application objects. This makes it easier to trace error origins, understand execution flow, and identify which entry points or tests triggered a failure.

The Troubleshooting MCP Server also enables inspection of variable state across different stack frames, allowing partners to understand why conditions evaluated as they did or how data was passed and transformed between procedures. Variable inspection supports both local and global variables, as well as complex types, helping partners diagnose subtle state‑related issues that are difficult to uncover with traditional single‑frame debugging.

In addition, the Troubleshooting MCP Server can retrieve source code for the relevant execution frame, including system or framework code, enabling analysis of implementation details without manually navigating files. Thus, coding agents can also place breakpoints programmatically based on ongoing analysis, supporting targeted investigations in areas that are otherwise hard to reach.

While the Troubleshooting MCP Server enhances debugging for complex and multilayered scenarios, it's designed to complement—not replace—traditional debugging tools. Interactive step‑through debugging and quick visual inspection remain well-suited for simple issues or for learning code flow, whereas the Troubleshooting MCP Server excels in advanced analysis, multiframe reasoning, and AI‑assisted diagnostics.