RP Growth Memory

English | 中文

中文

如果你希望几十万字的小说或 RP 在压缩后，角色仍保有由经历形成的核心，而不是被固定成一张角色卡，这套方法与它的 Codex Skill 适合你。

它不把人物写成“某人是理性的”“某人永远不会这样做”。它会先还原历史中的事件、选择、关系、代价与信息权限，再整理出人物在不同阶段如何因为经历而改变。目标不是设定一致性，而是成长一致性：人物可以继续变化，只要这种变化能从已有历史自然发展出来。

它会做什么

• 将聊天记录、小说章节、散乱笔记或已有摘要统一整理为可续写的长期 RP 记忆。
• 保留时间线、重大转折、能力或势力变化、关系发展、未解决线索与信息权限。
• 通过“经历 -> 选择 -> 后果 -> 新的可能性”理解人物，而不是输出固定性格标签。
• 合并没有新增意义的重复日常，保留第一次建立人物互动的场景，以及后来真正改变关系或选择的场景。
• 默认整理为统一结构：时间跨度、[剧情记录整理]、阶段与事件、【关键关系变化】、【当前场景】、【当前角色状态】、【阶段人物轨迹补充】。

在任何 AI 中使用

复制 核心方法、成长轨迹方法 与 连续性检查表 到 AI 的项目指令、系统提示词或当前对话中，再同时提供完整历史记录。它们本质上是平台无关的 Markdown 工作流；不同平台的差别只在于能否自动加载这些文件。

在 Codex 中安装

将本仓库克隆到 Codex skills 目录：

git clone https://github.com/syydmtl/ai-write-growth-memory.git "$CODEX_HOME/skills/rp-growth-memory"

若未设置 CODEX_HOME，通常可使用 ~/.codex/skills/rp-growth-memory。随后在任务中使用 $rp-growth-memory。

使用示例

使用 $rp-growth-memory，把这份 RP 聊天记录整理成长期记忆。
保留人物成长、关系变化、信息权限和当前场景；不要把角色写成固定人设。

English

If you want a novel or roleplay history to remain alive after hundreds of thousands of words have been compressed, rather than reducing its characters to static character cards, this method and its Codex Skill are for you.

It does not define people as “rational,” “kind,” or destined to behave one way forever. It reconstructs events, choices, relationships, costs, and information boundaries first, then captures how those experiences changed the characters across each phase. The goal is not setting consistency. It is growth consistency: characters may continue to change when that change can naturally grow from what they have lived through.

What It Does

• Turns chat logs, novel chapters, scattered notes, or existing summaries into continuation-ready long-term RP memory.
• Preserves chronology, turning points, ability or faction changes, relationship development, unresolved threads, and information permissions.
• Frames character development as experience -> choice -> consequence -> new possibility, not as fixed personality labels.
• Merges repetitive routine only when it adds no new meaning, while retaining scenes that establish or change relationships and decisions.
• Normalizes output into a consistent structure for reliable future continuation.

Use With Any AI

Copy the core method, trajectory method, and continuity checklist into project instructions, a system prompt, or the active conversation, then provide the complete source history. They are platform-neutral Markdown workflows; platforms differ only in whether they can load the files automatically.

Install for Codex

git clone https://github.com/syydmtl/ai-write-growth-memory.git "$CODEX_HOME/skills/rp-growth-memory"

If CODEX_HOME is not set, use ~/.codex/skills/rp-growth-memory. Invoke the skill with $rp-growth-memory.

Example

Use $rp-growth-memory to turn this RP log into long-term memory.
Keep character growth, relationship changes, information permissions, and the current scene. Do not turn people into fixed character cards.

Repository Layout

core/GROWTH_MEMORY_METHOD.md     # Platform-neutral workflow and output format
core/trajectory-method.md        # Evidence-based growth extraction
core/continuity-checklist.md     # Final continuity checks
SKILL.md                         # Codex adapter
agents/openai.yaml               # Codex UI metadata