SuperAPI AI Directory

novel-memory

Writing & research

View repository

Updated 10-16 01:56

About this skill

Novel Memory System

You have been injected with the novel-memory skill. This skill guides how to use the workspace file system as persistent memory for novel creation.


File Structure

novel/{书名}/
├── plan.md              # 总控大纲(由 novel-worldbuilding 生成)
├── 设定/
│   ├── characters.md    # 人物档案(全部角色)
│   ├── world.md         # 世界观规则
│   ├── foreshadowing.md # 伏笔台账
│   ├── timeline.md      # 时间线
│   └── relations.md     # 人物关系网络
├── 正文/
│   ├── 第1章_标题.md
│   └── ...
└── summary/
    ├── chapter_01.md    # 章节摘要
    └── ...

All paths are based on the workspace root directory.Files are memory, directories are categories.


Memory Categories

1. Static Settings (long-term valid)

Stored in the novel/{书名}/设定/ directory:

File Content
characters.md Character profiles: name, age, appearance, personality, background, motivation, abilities
world.md Worldview: power system, technology level, political system, geographical environment
relations.md Relationship network: relationship types between characters, factions, power hierarchy

2. Dynamic Logs (advanced per chapter)

File Content
summary/chapter_XX.md Per-chapter summary: core events, character status changes, new/recovered foreshadowing
设定/foreshadowing.md Foreshadowing ledger: ID, description, planted chapter, expected recovery chapter, status
设定/timeline.md Timeline: in-story time progression, time span of each chapter
设定/characters.md Current character status (append updates, do not overwrite history)

Operational Guidelines

Reading (mandatory before each chapter starts)

  1. Read the previous chapter summary:summary/chapter_XX.md
  2. Read current character status:设定/characters.md the latest status section at the end
  3. Read unrecovered foreshadowing:设定/foreshadowing.md entries with status "unrecovered" in
  4. Read timeline:设定/timeline.md obtain the current story time

Query Technique: when you need to search for specific settings, use grep in the 设定/ directory to search for keywords.

Writing (mandatory after each chapter is completed)

  1. Create this chapter's summary file:summary/chapter_XX.md
  2. Update character status: append change records in 设定/characters.md Append change record in
  3. Update foreshadowing ledger:设定/foreshadowing.md mark additions/recoveries in
  4. Advance timeline:设定/timeline.md record this chapter's time in

Initialization (when starting a new book)

  1. Create the novel/{书名}/设定/ directory and all initial files
  2. Fill in initial character profiles, worldview, empty foreshadowing ledger, timeline starting point
  3. Create the novel/{书名}/正文/ and summary/ directories

Operational Principles

  1. Record immediately: when defining new entities/rules, immediately write them into the corresponding file
  2. Update in sync: when settings evolve, update them promptly and keep files consistent with the main text
  3. Avoid redundancy: only record core, long-term valid information
  4. Prevent forgetting: rely on files rather than memory; at chapter 100, reading the files allows you to recall the settings from chapter 1
  5. Append first: use appending rather than overwriting for character status changes, preserving the change history

Foreshadowing Ledger Format

设定/foreshadowing.md Format of each foreshadowing entry in

### F001: {伏笔描述}
- 埋设:第 X 章,{具体场景描述}
- 预计回收:第 Y 章
- 状态:未回收 / 已回收(第 Z 章)
- 关联角色:{角色名}

Query Process

需要核实设定 → 读取 设定/ 下对应文件 → grep 关键词定位 → 获取信息
需要前文回顾 → 读取 summary/ 下近几章摘要 → 了解上下文
需要全局检查 → 遍历 设定/ 全部文件 → 交叉验证一致性