Driving Exam Question and Image Generator

Problem Addressed

Creating driving-exam questions often fails because the text, traffic rules, and illustration drift apart: a stem mentions yielding, overtaking, or a blind spot, while the image lacks a clear vehicle position; options exist, but the explanation still needs manual traffic-law grounding. This skill targets bulk driving-exam question generation, question-bank expansion, and driving-school teaching materials. Given parameters such as topic, question_type, and perspective, it returns the stem, options, correct answer, explanation, and one official-style illustration matched to the scene viewpoint.

How It Works and Boundaries

The workflow first performs compliance checks, then parses the exam point, question type, viewpoint, and difficulty, and decides whether a traffic-sign reference image is needed. It generates the question content, converts scene descriptions such as who proceeds first or where overtaking occurs into an absolute-coordinate layout, then assembles an image prompt from the reference image and viewpoint, calls the generation API, and polls the task result. Real calls require showapi_appkey; tasks are asynchronous and return a task_id for querying. Without a key, only a simulated preview is returned, not a real generated result. Note that topic and question_type are required, aspect_ratio supports only 1:1, 3:4, 4:3, 16:9, and 9:16, and polling results should be surfaced to the user as soon as each response arrives.