This project offers an AI-powered pre-check tool for content creators on platforms like Douyin, Xiaohongshu, and WeChat Channels. It analyzes drafts for potential rule violations, explains reasoning with official citations, and suggests compliant revisions. The tool helps creators avoid penalties and discover engaging, yet compliant, ways to express topics, improving content quality before publication.

How It Works

The tool employs AI for semantic analysis, prioritizing context over simple keywords to understand nuances. It distinguishes itself by assessing rule violations based on usage (e.g., personal experience vs. sales claims). The process includes pre-check, AI judgment, and intent-preserving revisions. This approach leverages 38 real-world samples for calibration and anchors decisions to 72 official rule citations for transparency.

Quick Start & Requirements

Installation involves sending a command to AI assistants supporting "Agent Skills" or cloning the repo: git clone https://github.com/yuwen-cool/yuwen-publish-precheck.git ~/.cursor/skills/yuwen-publish-precheck. It requires only Python 3 and runs entirely locally. Initial setup includes a brief guided onboarding for context.

Highlighted Details

• Rule Calibration: Uses 38 real-world samples to calibrate judgment, distinguishing acceptable anecdotes from prohibited claims.
• Official Citations: All rules link to official sources, with 72 original text citations provided for verification.
• Myth Debunking: Clarifies common self-censorship myths, stating official stances.
• Scenario-Based Detection: Features 41 detection modes tailored to commercial attributes and industries.
• Personalized Rule Database: Users can contribute experiences to a local data directory, making the tool progressively more accurate and tailored.

Maintenance & Community

Community contributions are encouraged via CONTRIBUTING.md, and updates are tracked in CHANGELOG.md. Specific community channels are not detailed.

Licensing & Compatibility

Licensed under CC BY-NC 4.0. Commercial use (integration into products/services, reselling) requires author permission. Using the tool for one's own self-created content, regardless of commercial nature, is free and requires no authorization.

Limitations & Caveats

The tool does not guarantee 100% approval due to platform-specific, non-visible factors. It strictly avoids providing evasion techniques ("绕审"). A "can post" verdict indicates no blocking issues were found within the current check's scope.