Skip to content

/game-build

Build a risk-matched whitebox or the approved production game for its target runtime. Turn GAME_DESIGN, and ART_DIRECTION when production begins, into a minimal BUILD_BRIEF and a runnable candidate that can be iterated with replayable evidence. Use for prototype the riskiest

From plugin
novel-to-game
9537 skills

Claude Code

$ npx -y skills add zenstory-ai/novel-to-game --skill game-build --agent claude-code

SKILL.md

game-build.SKILL.md
name: game-build
description: "Build a risk-matched whitebox or the approved production game by adapting to the host environment's available rendering stacks, 3D/2D generators, and modeling tools. Turn GAME_DESIGN and ART_DIRECTION into a concise BUILD_BRIEF and a runnable game candidate with deterministic self-checks. Use for prototype the riskiest design question, implement the approved game design, or turn this design into a running 2D, 2.5D, or 3D game. 游戏构建执行。结合本机与宿主环境实际可用的渲染栈与生成/建模工具(如 Three.js、混元 3D 等 3D 模型、Blender、2D/音视频工具),先针对头号设计风险搭建白盒原型,或将已批准的 GAME_DESIGN 与 ART_DIRECTION 实现为可运行游戏并配齐自检。用于验证核心设计风险、构建游戏原型或实现正式可玩版本。"

游戏构建执行

承接已批准的体验设计与美术方向,结合当前机器与会话的真实能力实现可玩版本(白盒阶段须有 `GAME_DESIGN.md`,正式生产阶段还须有 `ART_DIRECTION.md`;上游决策缺失时退回 `game-design` 补齐,不在 `BUILD_BRIEF.md` 里暗自重做设计)。先读 [build-brief-contract.md](references/build-brief-contract.md)。

产物语言由 `PRODUCT_BRIEF.md` 锁定;未锁定时跟随对话语言,不默认产出中文。

环境探测、技术选型与发挥空间

动手前先盘点本机与宿主环境实际能调动的运行时、渲染库、命令行、MCP 与生成服务,因地制宜选定最出效果的实现路线,给足技术与表现发挥空间,不生硬套用单一模板:

  • **渲染栈与运行环境**:在 `PRODUCT_BRIEF.md` 批准的目标平台内自由选用契合作品气质的技术栈。例如浏览器端做立体箱庭、棋盘沙盘、空间探索或光影演出时,可直接使用 **Three.js** 等主流 3D 渲染库搭建主光、天光、雾效与镜头调度;做俯视彩绘地图、卡牌或横版玩法时可用 2D Canvas / WebGL 渲染;桌面与移动端若本机已备齐 Godot、Unity 等引擎工具链则直接原生构建。目标工具链确实缺失时不得擅自改做网页版顶替,只走 `PRODUCT_BRIEF.md` 已批准的替代环境,并如实分记 `targetRuntime`、`testedRuntime` 与未覆盖特性(当前环境完全无法构建时,如实说明阻塞原因并仅交付 `BUILD_BRIEF.md`)。
  • **三维资产与空间表现组合**:做 3D 或 2.5D 项目时,先探测环境内能否调用 **混元 3D(Hunyuan3D)等图生 3D / 文生 3D 模型或接口**、本机是否装有 **Blender**(CLI、Python 脚本或 MCP)及网格转换工具。能调用 3D 生成模型时,适合以批准的角色或关键器物设定图生成带贴图网格(如 `.glb`),再按需配合 Blender 校准比例、脚底轴心、减面或动作绑定后接入 Three.js / 目标引擎;本机仅有 Blender 时可用脚本生成风格化低多边形或建筑模型;若外部建模与生成服务皆不可用,也可直接用 Three.js 程序化几何体、受光材质或分层景深立绘(2.5D 立体剧场)做出空间层次,不必退回单调的平面表单。
  • **图像、音视频与程序化混搭**:结合宿主可用的生图、视频生成与语音合成工具,把生成预算优先投向主角、核心地标、高潮 CG 与关键剧情过场;成套图标、UI 装饰、粒子动效与环境音效可灵活搭配矢量绘制、着色器及 Web Audio / 引擎合成器共同完成。所有成品资产落盘到项目工作区并接入真实画面(白盒临时占位须标明阶段属性,正式版不留未执行的提示词或坏链充数;调用外部工具只传当前资产所需描述与参考图,不外传原著全文,也不将密钥写入仓库)。

构建前声明 `buildStage: whitebox|production` 与 `buildPath: template|custom`:成熟模板能完整保留原著独有的动作、代价和世界反馈时可走 `template`;核心乐趣仰仗实时手感、空间走位、立体视角、物理碰撞或独特交互机制时走 `custom`,白盒阶段用最轻的几何体或占位素材把真实交互跑通,不把空间与动作玩法一律压成纯文字选项。

按项目实际用到的能力读取可选参考:

  • 3D 生成模型、Blender 与 Three.js 等空间资产联调:[blender-asset-workflow.md](references/blender-asset-workflow.md);
  • 2.5D 多朝向/多动作序列帧烘焙:[baked-sprite-workflow.md](references/baked-sprite-workflow.md);
  • 生成式自由输入、知情权限、事件回放或存档迁移:[playable-model-contract.md](references/playable-model-contract.md);
  • 角色配音:[tts-production-contract.md](references/tts-production-contract.md);
  • 动态视频、生成式音乐或音效:[generative-media-pipeline.md](references/generative-media-pipeline.md);选择后播放剧情过场(互动影游)再读 [cutscene-production.md](references/cutscene-production.md)。

构建循环

1. **实现状态与交互**:`whitebox` 只实现验证头号设计风险所需的真实输入、状态流转与核心反馈;`production` 完整实现 `GAME_DESIGN.md` 的状态、动作、分支与结局闭环。 2. **固定复现与排障**:保留能稳定复现问题的初态与输入序列(涉及随机时固定 `seed`),修改后重跑原失败路径及相邻分支反例;同一手感或表现问题连续两次修改未果时,先检索成熟开源做法理清机制再改。 3. **回写工具链与对齐设计**:在 `BUILD_BRIEF.md` 如实回写实际选用的渲染栈、工具链、版本号与运行命令(无法获取的版本填 `NOT_AVAILABLE: 原因`);实现与设计/美术文档存在偏差时,退回 `game-design` 修订或在 `BUILD_BRIEF.md` 记明冲突。

`whitebox` 执行完第 3 步后运行最窄风险检查并交回总入口组织试玩校准;`production` 继续执行:

4. **精修面向玩家的游戏内文案**:按 [player-copy-polish.md](references/player-copy-polish.md) 通盘审校代码与数据文件中的所有上屏文字,清零工程话术与套式 AI 腔。 5. **配齐规则自测与权威 `verify` 入口**:针对核心规则、数值门槛与分支汇流编写正反向自测断言并接入权威 `verify` 命令,运行冒烟自检清零启动报错、控制台阻断异常、资源加载失败与崩溃。 6. **实机视口校准后移交盲玩**:对配有批准目标图的界面与核心资产,在目标视口下截实机图与目标图并排比对布局、比例、色光与层级并修正(连续两轮无改善则停手记入 `limitations`,不篡改目标图迁就缺陷),最后移交 `game-qa` 做独立新玩家盲玩验收。

Read more
@zenstory-ai
Ships withnovel-to-game

Agent skills for source-grounded novel-to-game adaptation (小说改编游戏 / story to game): cited design docs, target-runtime builds, evidence-based QA. Three playable browser examples. For Claude Code, Codex, and Kimi Code.

Stars on GitHub

Momentum

stars in total
953
Star history of zenstory-ai/novel-to-game on GitHub, a chart by star-history.com

Other skills on novel-to-game.

game-concept
Auto-invoked

game-concept

Design game concepts from a novel. From SOURCE_BIBLE and PRODUCT_BRIEF, compare meaningful alternatives on the dimensions still…

game-qa
Auto-invoked

game-qa

Verify a game with evidence on its selected target runtime. Launch the actual build and prove real rendering, input, the core…

game-world-design
Auto-invoked

game-world-design

Design game experience, systems, and levels. Converge the chosen concept into one GAME_DESIGN defining the player promise, core…

novel-game-analyze
Auto-invoked

novel-game-analyze

Deconstruct a novel for game adaptation. Compress a raw novel, deconstruction library, or writing project into a SOURCE_BIBLE…

novel-to-game
Auto-invoked

novel-to-game

Turn a novel into a fully playable game on the selected target platform. Orchestrates the whole adaptation pipeline —…

More skills from other plugins.

brainstorming
Auto-invoked

brainstorming

You MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or…

@obra@obraInstall Skill
ask-matt

ask-matt

Ask which skill or flow fits your situation. A router over the skills in this repo.

skills12d309k
@mattpocock@mattpocockInstall Skill
accessibility
Auto-invoked

accessibility

Design, implement, and audit accessible UI to WCAG 2.2 Level AA across Web, iOS, and Android — semantic ARIA roles and labels,…

ecc13d293k
@affaan-m@affaan-mInstall Skill
compiler-commit

compiler-commit

Use when you want to verify compiler changes and commit with the correct convention. Runs tests, lint, and format, then commits…

react3d252k
@facebook@facebookInstall Skill

Questions about game-build.

Which coding agents can use game-build?
game-build installs in Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode and Windsurf with the skills CLI, using each agent's flag shown on this page. Flowy's auto-invocation, which fires a skill when your prompt matches, runs in Claude Code only.