yuwen-cool GitHub avatar

yuwen-publish-precheck

yuwen-cool

A local AI-powered skill that pre-checks Chinese social media content (Douyin, Xiaohongshu, WeChat Channels) for policy violations, suggests rewrites, and builds a personal rule library from user feedback.

Stars

232

7-day growth

No data

Forks

22

Open issues

0

License

NOASSERTION

Last updated

2026-07-15

Why it is worth attention

It goes beyond simple keyword blacklists by using two-layer detection (surface keyword + semantic context) and is grounded in 72 official citations, actively debunks self-censorship myths, and stores all data locally without uploading.

Who it is for

  • Content creators posting on Douyin, Xiaohongshu, WeChat Channels
  • E-commerce livestream hosts and marketing professionals
  • Self-media operators who want to avoid compliance penalties
  • Writers or editors needing rapid pre-publication compliance checks

Use cases

  • Pre-publish review of a script for platform rules
  • Rewriting violating content while preserving intent and tone
  • Post-penalty root cause analysis and rule memorization
  • Building a personal, domain‑specific compliance rulebase via one‑sentence feedback

Strengths

  • Two‑stage detection (keyword hit → context judgment) avoids false positives from pure word lists
  • Every rule references official sources with 72 linked citations, enabling independent verification
  • Actively tells users when common avoidance tricks (e.g., using '赚米' for '赚钱') are unnecessary
  • Fully offline; user data stays in local `data/` directory and is never uploaded

Considerations

  • Does not guarantee 100% approval; final determination depends on unseen platform models and account status
  • Currently supports only three platforms (Douyin, Xiaohongshu, WeChat Channels); more in roadmap
  • Requires an AI agent environment (Cursor, Codex, Claude Code) and Python 3, not a standalone application

README quick start

安装

Description

发布前审|发抖音/小红书/视频号前先让 AI 审一遍:哪句踩线、依据哪条官方规则、给能直接用的改法。38 篇真实样本校准判定尺度,72 条官方原文引文可查证,你踩过的坑沉淀成本地规则库越用越准。不承诺过审,不教绕审。

Related repositories

Similar projects matched by category, topics, and programming language.

NousResearch
Featured
NousResearch GitHub avatar

hermes-agent

Hermes Agent is an open-source, self-improving AI agent with a built-in learning loop that creates skills from experience, runs across platforms, and supports any LLM provider.

AI & Machine Learning
215,644
TianhangZhuzth
Featured
TianhangZhuzth GitHub avatar

Fundamental-Ava

Runx402 is a protocol and application that enables autonomous AI agents to pay for API calls using native ETH on Robinhood Chain via the HTTP 402/x402 standard, eliminating the need for human intervention, API keys, or subscriptions.

AI & Machine Learning
477
William-Lu-stack
William-Lu-stack GitHub avatar

Flawless

Flawless is an AI-native SRE control plane for Kubernetes that connects alerts, evidence, topology, approval, remediation, and recovery verification into a closed-loop AgenticOps system.

Python
655