ALBEDO-TABAI/deep-novel-system

最近记录的来源活动: 2026年4月24日 04:20

SkillsMP 收录数据更新: 2026年10月29日 18:30

已收集 skills: 1

GitHub 星标: 20
GitHub Forks: 2

用 AI 助手安装

复制这段 Prompt，发给你正在使用的 AI 助手。

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/ALBEDO-TABAI/deep-novel-system.

仓库 README

仓库 README 来源：README.md

查看 README 源文件

展开完整 README

# Deep Novel System <p align="center"> <a href="README.md">English</a> | <a href="README_zh.md">中文</a> </p> <p align="center"> <img src="https://img.shields.io/badge/Type-Claude%20Skill-8A2BE2" alt="Type"> <img src="https://img.shields.io/badge/Tested%20on-Antigravity-blue" alt="Tested"> <img src="https://img.shields.io/badge/License-MIT-yellow" alt="License"> </p> A high-quality long-form novel writing skill designed for **Claude**, specializing in complex narratives, sensory descriptions, cross-chapter memory management, and feedback loops. Supports multiple genres and mature themes when authorized by the user. ## 🎯 About This is a **Claude Skill** following the [Agent Skills](https://github.com/anthropics/agent-skills) open standard. Skills are reusable AI agent capability modules that extend AI's professional abilities through structured instructions and workflows. ### Recommended Environment This skill has been tested on **Antigravity** with excellent results. Recommended configurations: | IDE / Client | Recommended Model | |-------------|-------------------| | **Antigravity** (Google DeepMind) | Gemini 3 Pro | | **Cursor** | Claude 4.5 Opus | | **Claude Code** | Claude 4.5 Opus | Also tested with solid writing quality: **GLM-5.1**, **MiMo-V2.5-Pro** — works in any Skills-compatible client. > 💡 **Tip**: Using more powerful models yields better long-form narrative coherence and detailed descriptions. ## ✨ Features - 📖 **Structured Writing Workflow** - Complete creation process from outline to chapters, with truncation detection and recovery - 🧠 **Memory Management System** - Cross-chapter state tracking with YAML-schema'd memory blocks - 🔄 **Feedback Loop Mechanism** - Versioned rules with archive flow for long-running projects - 📁 **Project Templates** - Idempotent init script (default: skip-existing; supports `--dry-run` / `--force`) - 🗂️ **Project State Machine** - `state.yaml` tracks current chapter/section so multi-session work stays consistent - 🎭 **Multi-genre Support** - Suitable for horror, mature/erotic, light novels, deep narratives, etc. ## 🚀 Quick Start ### Installation Clone this repository to your `.agent/skills/` directory: ```bash cd /path/to/your/project/.agent/skills/ git clone https://github.com/ALBEDO-TABAI/deep-novel-system.git ``` ### Usage In your AI conversation, say: > "I want to start writing a novel" or > "Continue writing `<project path>`" ## 📂 Directory Structure ``` deep-novel-system/ ├── SKILL.md # Main skill entry file ├── README.md # This file (English) ├── README_zh.md # Chinese version ├── LICENSE # MIT License ├── assets/ │ └── template/ # Project template (copied by init_novel.py) │ ├── config/instruction.md │ ├── drafts/ │ ├── memory/ │ ├── plans/{chapters,setting}/ │ ├── references/ │ ├── feedback/{requirements.md,archive/} │ └── state.yaml ├── docs/ # Skill's own rule docs (do NOT use as style references) │ ├── workflow.md # Long novel writing process │ ├── memory_management.md │ ├── reference_usage.md │ ├── feedback_loop.md │ └── onboarding.md # First-time interactive setup flow └── scripts/ └── init_novel.py # Project initialization script ``` > Note: `references/` only lives **inside each user project** (copied from the template) — it holds your own style samples and is not part of the skill repo itself. ## 📚 Workflows ### 1. Long Novel Writing Process See [docs/workflow.md](docs/workflow.md) **Key Stages**: 1. **Input Analysis** - Read previous chapters, outlines, and requirements 2. **Drafting** - Write content in blocks (e.g., `CH01 SEC01`), with truncation detection and append-style continuation 3. **Review & Polish** - Check against style guidelines 4. **State Update** - Advance `state.yaml` ### 2. Memory Management See [docs/memory_management.md](docs/memory_management.md) - Update `memory/` files after major events - Each block uses YAML front-matter (chapter / section / irreversible_changes / ...) - Future scripts can validate consistency from these structured fields ### 3. Feedback Loop See [docs/feedback_loop.md](docs/feedback_loop.md) - Each rule carries `id` / `added` / `last_hit` metadata - Periodic archive flow keeps `requirements.md` lean for long-running projects ## 🛠️ Project Initialization Two ways to bootstrap a project: ### 1. Interactive (recommended) Just tell the AI: > "I want to start writing a novel" It will run a short Q&A (genre, main characters, core conflict, target length, special requirements) and pre-fill `plans/global_plan.md`, `plans/setting/{characters,world}.md`, and `plans/chapters/ch01_plan.md` from your answers. See [docs/onboarding.md](docs/onboarding.md) for the full question sequence. ### 2. Script (manual / advanced) If you'd rather create the skeleton yourself and fill it in later, run: ```bash python scripts/init_novel.py <target_directory> # default: create missing files only python scripts/init_novel.py <target_directory> --dry-run # preview only python scripts/init_novel.py <target_directory> --force # overwrite existing files ``` The script only copies `assets/template/` into `<target_directory>` — you still need to fill in the outline, character cards, world setting, and style references before the AI can start writing. ## 🤝 Compatibility This skill follows the [Agent Skills](https://github.com/anthropics/agent-skills) open standard and is theoretically compatible with all AI agents supporting this standard: - **Antigravity** (Google DeepMind) ✅ Tested - Claude Code - Cursor - GitHub Copilot - Other Agent Skills compatible clients ## 📄 License [MIT License](LICENSE) ## 🙏 Acknowledgments Inspired by the [obra/superpowers](https://github.com/obra/superpowers) framework.

正在加载仓库 README…

这个仓库中的 skills

作家与作者

1 个职业分类 · 已分类 100%

skill

职业分类

描述

更新

deep-novel-system

作家与作者

一个用于撰写高质量长篇小说的综合系统，擅长复杂叙事、感官描写、跨章节记忆管理与反馈循环。支持多种类型与成人主题（在用户授权范围内）。当用户想开始新小说、继续现有小说或需要长篇结构化叙事指导时使用此技能。

2026-04-24

已展示 1 / 1 个已收集 Skill。