From novel
基于全部设定续写小说，自动生成章节元数据。支持接续最新章节或指定位置续写。 Use when: "续写", "continue", "继续写", "下一章", "接着写"
npx claudepluginhub cabinary/claude-plugins --plugin novel## Context ### 大纲 !`if [ -f outline.md ]; then cat outline.md; else echo "（缺失：请先 /outline 创建大纲）"; fi` ### 风格设定 !`if [ -f style.md ]; then cat style.md; else echo "（缺失：请先 /style 定义风格）"; fi` ### 时间线 !`if [ -f timeline.md ]; then cat timeline.md; else echo "（暂无时间线）"; fi` ### 全部人物设定 !`if ls characters/*.md >/dev/null 2>&1; then for f in characters/*.md; do echo "=== $(basename "$f" .md) ==="; cat "$f"; echo; done; else echo "（暂无人物设定）"; fi` ### 全部事件设定 !`if ls events/*.md >/dev/null 2>&1; then for f in events/*.md; do echo "=== $(basename "$f" .md) ==="; cat "$f"; echo; done; else echo "（暂无事...
/continueAdvances the active vibe session by one step: checks status, runs worker and explainer agents, optionally explorer, and displays progress summary with changes, verifications, and next actions.
/continueScans docs/feature/ for feature progress across development waves, displays status summary with symbols and details, detects anomalies, and resumes next wave after confirmation.
/continueResumes prior session by scanning progress files and git state, summarizes work status with branch/phase details and next action, then loads context after confirmation.
/continueAutonomously executes pending tasks from orchestrator queue, delegating to agents like frontend-dev or backend-dev, with git checks, retries, and quality gates until completion or pause.
/continueContinues a previous Codex session using a thread ID and follow-up prompt to iterate on findings, request fixes, or drill deeper into analysis.
/continueAnalyzes existing Slidev presentation files, summarizes status on structure, content quality, visuals, and suggests priority next steps.
!if [ -f outline.md ]; then cat outline.md; else echo "（缺失：请先 /outline 创建大纲）"; fi
!if [ -f style.md ]; then cat style.md; else echo "（缺失：请先 /style 定义风格）"; fi
!if [ -f timeline.md ]; then cat timeline.md; else echo "（暂无时间线）"; fi
!if ls characters/*.md >/dev/null 2>&1; then for f in characters/*.md; do echo "=== $(basename "$f" .md) ==="; cat "$f"; echo; done; else echo "（暂无人物设定）"; fi
!if ls events/*.md >/dev/null 2>&1; then for f in events/*.md; do echo "=== $(basename "$f" .md) ==="; cat "$f"; echo; done; else echo "（暂无事件设定）"; fi
!if ls chapters/*.md >/dev/null 2>&1; then ls chapters/*.md | grep -v '\.meta\.md$' | sort; else echo "（暂无章节）"; fi
!latest=$(ls chapters/*.meta.md 2>/dev/null | sort | tail -1); if [ -n "$latest" ] && [ -f "$latest" ]; then echo "=== $(basename "$latest") ==="; cat "$latest"; else echo "（暂无元数据）"; fi
!if ls chapters/*.meta.md >/dev/null 2>&1; then for f in $(ls chapters/*.meta.md | sort); do echo "=== $(basename "$f" .meta.md) ==="; awk '/^## 伏笔/,/^## [^伏]/' "$f"; awk '/^## 悬念/,/^## [^悬]/' "$f"; echo; done; else echo "（暂无伏笔数据）"; fi
用户想要续写小说。这是本项目最核心的功能。
/character 基于大纲生成主要人物 批量生成，不阻塞根据最新 .meta.md 的 context_depth 决定加载策略：
context_depth 章的正文.meta.md 的 summary 字段第一章或 context_depth 未定义时，默认读最近 2 章全文。
动笔前必须在心中完成以下核对：
正文：
chapters/{序号}-{标题}.md，序号承接已有章节，三位数零填充# 第X章 标题---元数据（必须同步生成）：
保存为 chapters/{序号}-{标题}.meta.md：
---
summary: "一两句话概括本章核心内容"
context_depth: N
characters: [出场人物]
location: 主要场景
time: 故事内时间
mood: 氛围基调
---
## 伏笔
- [ ] 本章新埋设的伏笔
- [x] 本章回收的伏笔（第Y章埋设，本章回收）
（从前文元数据复制所有仍未回收的伏笔，保持 [ ]）
## 悬念/钩子
- 本章结尾的悬念
## 人物状态变化
- 人物名：本章中的状态/情感/认知变化
context_depth 规则：
回复末尾简述：
用户输入：$ARGUMENTS