用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/LYL1015/JarvisHub --skill flova-gameplay-demo-trailer命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
Use when the user asks JarvisHub AI Chat or a `webHero` workflow to generate a website, landing page, homepage, brand-style web design, preview-first webpage workflow, or wants section screenshots, webpage assets, and final HTML to be produced as one staged flow.
把 assets/demo 提炼成运行时可用的视觉连续性方法论:先锁资产/角色/场景/镜头语义,再做扩镜与视频。(素材生成场景:角色一致性、视觉锚点锁定、单素材生成前的锚点规范与一次只改一个变量。)
基于用户提供的设计稿图片还原真实网页。该 skill 可用于任意前端项目,但强依赖本地已启动且已授权的 JarvisHub vision 工作流;视觉模型由服务端默认配置决定,支持外部传入 vision prompt,且不可静默降级。
基于 SOC 职业分类
正在显示 SKILL.md
| name | flova-gameplay-demo-trailer |
| description | Use when 用户要生成游戏 Demo、伪实机演示、玩法预告、开放世界概念片、HUD/UI 关键帧、角色 Boss 战或游戏短片 trailer。 |
用于游戏 Demo 与玩法预告:为原创游戏概念生成角色、场景、UI/HUD、三段式玩法演示或短 trailer。
imageUrl / videoUrl,不能把提交态当完成态。critic sub-agent:只读取真实媒体并评审,不生成、不补素材。blocked 项,除非本轮工具列表明确暴露对应能力。Use this skill for:
Do not use it to reproduce commercial game names, official logos, UI, map icons, brand assets, exact trailer shots, or protected characters. Use original worlds and visual grammar.
Follow the JarvisHub Execution Model. Do not assume exact image/video model choices or UI lock capabilities. If exact UI text/position cannot be preserved by video generation, keep UI in still keyframes or add UI in assembly/post.
Collect:
If the user uploads a person photo to make a custom protagonist, use it only as identity/style reference and ask whether they want stylized game character conversion.
| Mode | Use For | Core Rule |
|---|---|---|
three-part_game_demo | 45s demo with onboarding, narrative, combat | A/B/C 15s segments |
open_world_slice | city/exploration/crime/chase/gameplay feel | original open-world grammar, no official IP |
single_gameplay_moment | 5-15s simple action | one readable action and camera |
Pause after spec, storyboard, character/scene assets, UI keyframes, first clip/batch, and final assembly.
Default 45s:
Video_A_onboarding 15s: title/start screen, character selection, loading, first playable scene.Video_B_narrative 15s: NPC interaction, objective reveal, world movement.Video_C_combat 15s: core combat, boss move, ultimate skill, victory/settlement UI.Each segment should be internally coherent and not depend on random montage.
Required shot fields:
segment_id,gameplay_function,start_keyframe,end_keyframe_if_needed,character_refs,scene_refs,ui_layer_plan,camera_type,player_action,npc/enemy_action,audio/SFX,continuity.Use original visual grammar:
Do not mention protected game titles in final prompts unless the user specifically asks for reference language and the system policy allows it. Prefer:
original open-world crime-action game trailer concept
Single-take slice:
Multi-cut event:
Treat UI as a controlled design asset.
Plan:
If exact UI must be readable:
UI lock checklist:
Character sheet:
Scene asset:
Boss/enemy:
Use only what fits the gameplay:
Do not combine too many camera grammars in a single short segment.
Use original game terms and visible actions.
Video prompt structure:
[camera type] <<<image_1>>> [character/action], <<<image_2>>> [scene], [UI plan if any], [enemy/environment motion], [SFX/audio], no subtitles, no random text, no official logos, no real brands.
For single gameplay slice:
one continuous gameplay-like shot, one simple readable action, no cuts, no montage
For multi-cut event:
multi-shot sequence with 4 quick cuts, same characters, same location, same time of day, same movement direction
For UI-heavy segments:
keep HUD positions stable; if text cannot stay readable, reserve clean overlay zones for post-production UI.
Plan separate layers when available:
When clips are later assembled under dedicated audio, add no music to video prompts to avoid embedded music conflicts.
Check:
Fix one UI frame, one prompt, or one segment at a time.