自动增加屏幕和反复重生成可能消耗超出用户预期的付费额度
原文依据:5 处该 Skill 不仅按要求生成画面,还要求在认为流程需要时增加屏幕、生成额外细节图,并持续重生成到质量达标;这些扩展无需先取得用户对数量或预算的确认。
如果底层图像服务按次计费或受配额限制,一次模糊的“应用概念”请求可能触发多张图片和多轮重试,增加费用、耗尽额度或延长任务时间。
这是生效中的默认行为,而非示例:Skill 将图像生成积极度设为最高,并要求在其判断有助于流程时增加屏幕和细节图;对较弱结果还要求继续重生成。若所用图像服务按次计费或受额度限制,这些未预先设定数量上限、也未要求预算确认的额外生成可能增加费用或耗尽额度。源码未说明具体服务价格,因此只能确认这种条件性风险,不能断言已产生收费。用户可要求作者加入生成数量、重试次数和预算上限,并在超限前征求确认。
`(1 = repeated screen templates, 10 = clearly differentiated screen rhythm)`- IMAGE_GENERATION_EAGERNESS: 10 `(1 = minimal screens, 10 = generate as many screens and detail views as needed)`- SPACING_GENEROSITY: 9 查看另外 4 个位置
- If the user says "fintech", "health", or "productivity", increase trust, calmness, and structural clarity.- Do not be lazy with screen count.- If more screens would make the flow better, generate more screens.- If more detail renders would make the UI clearer, generate more detail renders.- Default toward richer art direction than standard AI mobile output.## 34. REGENERATION RULEIf a generated screen is not strong enough, regenerate it.Do not settle for the first mediocre render.Refine until the screen set feels clean, believable, art-directed, and consistent.9. lock an internal design bible for consistency10. generate the required screen images11. generate more screens if needed for a believable flow12. generate extra detail renders if needed13. keep the first screen especially clean