Menu
Build enterprise AI ad asset systems and batch AI UGC video production briefs. Use when a user asks for AI UGC ads, product demo videos, ad asset factories, prompt packs, or reusable video production workflows.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | ai-ugc-factory |
| description | Build enterprise AI ad asset systems and batch AI UGC video production briefs. Use when a user asks for AI UGC ads, product demo videos, ad asset factories, prompt packs, or reusable video production workflows. |
| category | media |
| symbolName | video.badge.waveform |
| seedVersion | 2 |
Use this skill to turn a product or service into a repeatable AI UGC ad production system, not a one-off prompt.
使用这个 skill，把一个产品或服务转成可重复运行的 AI UGC 广告生产系统，而不是一次性 prompt。
Produce a compact package:
输出一个轻量生产包：
Read the product website, source files, screenshots, and user notes. Separate verified facts from assumptions. Do not invent prices, discounts, certifications, bonuses, or performance claims.
读取产品官网、源文件、截图和用户说明。区分已验证事实和推测。不要编造价格、折扣、认证、赠品或效果承诺。
Rename or label assets by visual function:
按视觉功能重命名或标记素材：
hero-product-image｜主产品图packaging-detail｜包装/细节lifestyle-context｜生活场景feature-proof｜功能证明website-cta｜网页/CTAcreator-reference｜真人/创作者参考style-reference｜风格参考contact-asset｜联系/转化资产The manifest should say which assets must be actual model inputs. A path in a prompt is not proof that the model read the image.
素材清单必须说明哪些素材需要作为真实模型输入。只在 prompt 里写路径，不等于模型真的读取了图片。
Pick 1-3 lanes that fit the product:
根据产品选择 1–3 条视频生产路线：
Every 1-2 seconds should carry a job: hook, product proof, pain point, benefit, objection, credibility cue, visual reset, CTA.
每 1–2 秒画面都应承担一个明确任务：钩子、产品证明、痛点、利益点、异议、可信度、视觉重置或 CTA。
For every segment, state:
每个片段都要写清：
Before final video generation, confirm any user-owned style decisions: mascot/character identity, voiceover tone, rhythm, CTA/contact asset, and whether exact captions should be post-composited.
最终生成前，确认所有用户拥有的风格决策：角色/吉祥物、口播语气、节奏、CTA/联系资产，以及精确字幕是否需要后期合成。
Require per segment:
每个片段必须检查：
MODEL_INPUT_IMAGES_ACTUALLY_ATTACHEDPRODUCT_REFERENCE_WAS_IMAGE_INPUTPROMPT_ONLY_PRODUCT_REFERENCE_USEDPRODUCT_VISIBLE_AND_MATCHES_REAL_ASSETIf product fidelity or text quality fails in client-facing shots, request a narrow correction pass instead of accepting the render.
如果客户可见镜头中的产品一致性或文字质量失败，应要求窄范围返修，而不是直接接受成片。
User: “帮我为这个产品批量生产 AI UGC 广告视频，并沉淀成可复用 workflow。”
Response shape｜响应结构：