novel-quality-check

What This Skill Does

The novel-quality-check skill is a comprehensive editorial suite designed to elevate AI-generated fiction to professional, human-like quality. It acts as a rigorous quality assurance agent that strictly adheres to the principle of "Human-Write Feel," ensuring text is nuanced, avoids robotic patterns, and maintains logical consistency. The tool employs a sophisticated twelve-step verification process, covering everything from forbidden word scanning and template sentence replacement to complex logical, character, and plot consistency checks. It features a built-in Python script (freq_check.py) for exhaustive word frequency analysis, ensuring that your narrative remains crisp and repetitive-free. Whether you are dealing with plot holes, rhythm issues, or common AI syntax traps, this skill provides actionable, line-by-line feedback to refine your work into high-quality literature.

Installation

To integrate this skill into your OpenClaw agent, use the following command in your terminal: clawhub install openclaw/skills/skills/chn012cjus/web-novel-publishing-readiness-and-quality-check-skill Ensure your project directory contains the necessary Python environment to support the embedded freq_check.py script for accurate word frequency scanning.

Use Cases

• Pre-Publishing Review: Conduct a final scrub of a chapter for prohibited words and repetitive patterns before uploading.
• Logical Consistency Auditing: Use it to ensure that character motivations, world-building rules, and plot progression align with your established outline.
• Style Improvement: Transform dry, machine-like descriptive sentences into engaging, perspective-driven prose that feels authored by a human.
• Cross-Chapter Synchronization: Check for disjointed plot points or repetitive descriptions across consecutive chapters to maintain narrative flow.

Example Prompts

1. "检查正文：请针对我刚刚粘贴的第三章内容进行质量报告，重点关注套路句识别和人设一致性检查。"
2. "执行违禁词扫描和逻辑漏洞检查，并对照我的大纲对比这一章的角色动机。"
3. "请对这一章进行自检，重点排查比喻词的使用和主观描述，将其转为客观描写并提供修改建议。"

Tips & Limitations

• Human-First Approach: Always prioritize the 'Human-Write Feel'. If the AI output feels too smooth or mechanical, use the C-档 (High Template Style) modification workflow.
• Frequency Scanning: Never rely on the Python script alone; it only detects frequency. You must manually perform the line-by-line forbidden word and sentence structure scan.
• Clean Corrections: When suggesting edits, ensure the revision itself is free from the forbidden patterns; don't swap one banned phrase for another.
• Constraint: This skill is compute-intensive; for very long manuscripts, consider checking chapter-by-chapter rather than submitting the entire book at once.