humanizer-academic：学术文本去 AI 痕迹并保持语域
humanizer-academic 是 SkillHub 社区 Skill（v1.0.0，无需 API Key），面向英中及混排学术/专业文本：在词汇、结构、统计三层削弱 AI 写作信号，保留 scholarly register，并加入作者立场、来源具体性、句段节奏等学术人文纹理，且不编造事实。附带脚本仅做检测；纯检测请求只返回 detector map。适合论文、摘要、综述、报告与工作论文；不适合 casual 口语化、虚构证据或强修辞创作。安装见 skillhub.cn/install，坐标 @user_15292d5a/yjkj-humanizer-academic，对话可用 $humanizer-academic 触发。
Read MoreAI文档检测：对 docx/pdf/txt/md 做 AI 生成痕迹审核
AI文档检测是 SkillHub 社区 Skill（@user_fcc443fb/ai-text-auditor，v1.0.0），面向 docx、pdf、txt、md 做 AI 生成痕迹检测，输出总体 AI 率评分与逐维度证据的结构化审核报告，标注 requires_api_key=false。文章说明其定位、与 humanizer-zh 的边界，以及通过 skillhub CLI 安装到 Cursor 等 Agent skills 目录的步骤与典型对话用法。
Read More论文润色降重一站式工具：本地论文 AI 率自查与润色指引
介绍 SkillHub 上的论文润色降重一站式工具（paper-polisher-pro）：面向学位论文与投稿稿的本地 AI 率自查、大模型指纹归因、句子级改写策略与术语/翻译腔/AIGC 合规检查。说明 SkillHub 评分 99492、content-creation 分类与安装路径，给出 skillhub install、pp_doctor、ai_detector、pp_workflow 等可复现命令，并区分与 paper-rewriter、机构查重的边界。
Read MoreAI率检测助手：在 Agent 里做中文短文本 AI 率自检
AI率检测助手（@user_34343d1f/ailv-aijiance）是 SkillHub 社区 Skill，面向 AI 检测、AI 率与机器写作痕迹等需求，在对话中按规则调用 aijiance.org 接口，无需 API Key。支持 detect.py 与 curl，返回概率与中文解读。限 300–1000 有效字，超长稿引导 ailv.run。本文说明安装路径、典型命令与注意事项。
Read MoreAI Detection Assistant: Agent Skill for Chinese Content — Free AI Rate Detection
AI Detection Assistant (@user_34343d1f/aijiance) is a Skill in the SkillHub community’s content-creation category, v1.0.1. It offers free AI-rate detection for Chinese content and requires no API key. It is suitable for self-checking papers, assignments, and self-media copy during Agent conversations in environments such as Cursor and Claude Code. Installation can be performed by using `skillhub install` and specifying the skills directory for each client, or by following `skillhub.md` to have the Agent install it. For content over 1,000 words, it is recommended to visit ailv.run. This article is compiled based on SkillHub metadata and official installation documentation, and does not fabricate detection values.
Read Moredsh-humanizer: Pre-Generation Injection of Human-Like Style Constraints + AI-Smell Detection Report
`dsh-humanizer` is a plugin for DeepSeek Harness (DSH) designed to address the frequent “AI flavor” in model outputs. It injects writing constraints through the system prompt to standardize outputs and provides a pure-code detection tool based on local regular expressions. It features zero token consumption and reproducible results, supporting automatic style injection and manual AI-flavor detection. Once installed, its detection logic can be reused in a Node.js environment. As an auxiliary reference tool, it effectively identifies 22 common patterns, such as filler phrases and summary-like tone, though its reliance on literal matching should be noted as a limitation.
Read More