Could you walk me through the logic of your run method? Specifically, how it executes the tool loop.

The run method starts by initializing counters for steps and failures. It then enters a loop that iterates up to a predefined step limit. In each iteration, it selects the next tool from the list (often cycling through the list). The tool’s execute function is called with the current input. If the call succeeds, the result is appended to the conversation history and the failure counter is reset. If an exception is raised, the failure counter increments and a fallback tool may be invoked. After each iteration, the history is trimmed to keep only the most recent 10 messages. The loop terminates either when the step limit is reached, the failure limit is exceeded, or all tools have been processed. Finally, the method returns the accumulated history.