zy-zmc

zy-zmc / tianming-skill

Public

天命 · AI长篇小说协同创作Skill — 模块化提示词工程系统 (Claude Skill / Prompt Engineering for Novel Writing)

github.comzy-zmctianming-novel-ai-writer ai-writing chinese-novel claude-skill creative-writing long-form-writing
35
7
69% credibility
Found May 24, 2026 at 35 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
PowerShell
AI Summary

TianMing is a creative writing tool designed to help authors collaborate with AI when writing long novels. The system provides a structured framework of 30+ modular instruction files that guide AI behavior to maintain consistency across hundreds of chapters. Writers prepare by filling in templates about their story's world, characters, and preferred writing style, then use simple text commands to generate story outlines, chapter plans, and full chapter content. The system includes built-in safeguards to prevent plot holes, ensure character consistency, and balance dramatic intensity with breathing room for readers. A supporting Python script helps quantify how intense each chapter is by scoring factors like character deaths, plot reveals, and major events. The entire project is open source and licensed for non-commercial use, with an optional desktop application available for those who prefer not to work with prompt files directly.

How It Works

1
📚 You discover a powerful writing partner

You learn about TianMing, an AI skill that helps write long novels while keeping your story consistent and well-structured.

2
📁 You set up your story's foundation

You download five template files and fill them with your story's world, characters, writing style, and key events.

3
🧙 You activate your AI writing assistant

You type 'initialize' and your AI companion wakes up, ready to help you create your epic novel together.

4
🗺️ You plan your entire story together

You use commands like '天命：大纲' and '天命：规划' to map out your story's grand blueprint, from major arcs to chapter outlines.

5
You write your chapters your way
📝
Blueprint first

Generate a chapter skeleton for review, then refine before expanding to full text

⚡
Direct to draft

Start with a complete chapter draft, then polish as needed

6
🏥 You keep your story healthy

You run '天命：体检' periodically to check your world consistency and make sure all plot threads stay connected.

📖 Your novel comes alive

You've created a complete, coherent long novel with consistent characters, rich plots, and satisfying payoffs for all your story threads.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 35 to 35 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is tianming-skill?

Tianming is a modular prompt engineering system for writing long novels with Claude. It takes a 995-line monolithic prompt and restructures it into a command-based workflow with over 30 protocol files, letting you generate chapter outlines, drafts, and final text through simple text commands like 「天命：大纲」 or 「天命：正文」. The system includes Python utilities for conflict scoring and PowerShell scripts for reference validation, targeting Chinese-language novel authors who want engineering-grade control over AI-assisted storytelling.

Why is it gaining traction?

The hook is token efficiency: by loading protocols on-demand through a routing table, Tianming claims 50-70% reduction in token consumption compared to dumping everything upfront. It enforces consistency across chapters through a "world stone" knowledge base that tracks characters, lore, and foreshadowing tiers. The separation between knowledge templates and protocol logic makes it portable across different Claude platforms. For writers who have been working with raw long prompts, this modular approach offers cleaner iteration and reuse.

Who should use this?

Chinese-speaking novel writers who are comfortable with structured prompts and want to scale beyond one-shot generation. If you have already tried feeding Claude a giant system prompt and found it expensive or inconsistent across chapters, this gives you the scaffolding to fix that. Developers interested in prompt engineering patterns will also find the reference-linter and conflict-score tools worth studying.

Verdict

At 35 stars, this is a niche, early-stage tool with a specific audience. The documentation is thorough and the architecture is sound, but maturity signals are weak. The 0.7% credibility score reflects the limited community validation. Try it if you write long-form Chinese fiction and want CLI-driven consistency; wait for more traction if you need robust community support or English-language resources.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.