aigc-reducer

降低论文AIGC检测率的专业化处理流水线。当用户需要降低AI生成文本检测率、降低AIGC百分比、去除AI写作特征、规避Turnitin检测、论文去AI化、让论文更像人类写作时使用。本skill包含5个专业化模块：词汇清洗、句法节奏打乱、机械连词隐蔽、学术克制注入、困惑度重构。

Security Scan

The skill's declared purpose (rewriting text to reduce AIGC detection) matches its instructions and resources; it requires no credentials or installs and contains no hidden endpoints, but it is explicitly designed to help evade detection and has ethical/legal risks.

Purpose & Capability

The name/description describe reducing AI-detection in academic text and the SKILL.md contains five text-transformation modules that directly implement that purpose (vocabulary replacement, sentence rhythm changes, removal of mechanical transitions, hedging language, and substitution of high-predictability phrases). There are no unrelated environment variables, binaries, or installs requested.

Instruction Scope

The instructions operate solely on user-provided text (500‑word paragraphs) and reference only the included jargon list; they do not request system files, environment variables, or external endpoints. However, the instructions are explicit about evading detectors (Turnitin etc.) and provide open-ended transformation rules that could be misused for academic dishonesty. This is a behavioral/ethical concern rather than a technical incoherence.

Install Mechanism

No install spec or code files beyond instructional markdown are present. Because this is instruction-only, nothing is written to disk and no external code is fetched—low technical installation risk.

Credentials

The skill requires no environment variables, credentials, or config paths. Requested capabilities are proportionate to a text-rewriting skill.

Persistence & Privilege

Flags show normal defaults (always:false, agent invocation allowed). The skill does not request permanent presence or modify other skill/system settings. Autonomous invocation is allowed by default but is not combined with other privilege or credential requests here.

Assessment

This skill is internally consistent: it only contains instructions and a wordlist for rewriting text to look less like AI output and does not request credentials or install code. However, it is explicitly designed to help evade AI-detection and could enable academic dishonesty or violate institutional policies (Turnitin/academic integrity) and platform terms. Before installing or using it, consider: (1) legal/ethical implications—do not use it to cheat or to hide misconduct; (2) provenance risk—the source is unknown and there is no homepage or maintainer information; (3) data sensitivity—do not submit confidential or personally identifying material to be rewritten; (4) alternatives—if your goal is legitimate editing (improving clarity, style, or humanizing tone), prefer tools and workflows that disclose edits and preserve attribution. If you proceed, audit outputs carefully for factual integrity and citation correctness, and confirm usage does not breach any policy or contract that applies to you.

AIGC检测率降低流水线

本skill提供5个专业化处理模块，系统性消除AI生成文本的机械化特征，注入人类写作特有的微妙语境和句法突发性。

流水线顺序

处理顺序：词汇清洗 → 句法节奏打乱 → 逻辑衔接专家 → 学术克制注入 → 困惑度重构

每次处理约500字的段落。

---

模块1：词汇精炼器 (AI黑话清洗)

清洗"AI专属黑话"，替换为精准的学术术语。

触发词清单：详见 references/jargon-list.md

处理流程

1. 识别AI常见触发词（深入探讨、多层面的画卷、促进、关键的、强调、阐明等）
2. 用精准、符合语境的学术术语替换
3. 保持核心论点和经验数据不变
4. 如有必要，直接重写整个分句

---

模块2：节奏变频师 (句法节奏打乱)

最大化文本的"突发性(Burstiness)"，打破AI单调的句式。

处理流程

1. 在冗长复杂的从句之间，穿插极短的爆发性短句（5-8词）
2. 将被动迂回的长句转化为主动直接陈述
3. 保持学术专业基调
4. 确保底层逻辑流完好

---

模块3：逻辑衔接专家 (机械连词隐蔽)

消灭机械过渡词，利用概念自然重叠创建隐形连接。

禁用连词：此外、而且、另外、因此、从而、Furthermore、Moreover、Additionally、Therefore、Thus

处理流程

1. 移除段落开头和机械位置的连接词
2. 通过"语义链"创造过渡：上一句句尾概念回荡在下一句主语中
3. 确保过渡逻辑顺理成章

---

模块4：语义精细化工具 (学术克制注入)

软化AI傲慢的绝对确定性，注入学术模糊限制语。

绝对化动词替换：

• 证明了 → 暗示了
• 确保了 → 表明了
• 保证了 → 可能有助于

处理流程

1. 将绝对化陈述替换为克制替代
2. 在宽泛主张中注入方法论警示语和边界条件
3. 保持学术自信但不傲慢

---

模块5：困惑度重构员 (困惑度与语境重构)

最大化文本"困惑度"，用罕见但精准的学术表达替换陈词滥调。

常见陈词滥调：快速发展、取得平衡、深入研究、显著提高

处理流程

1. 识别高预测性词汇搭配
2. 替换为罕见且精准的学术表达
3. 保持完美连贯性和学术严谨性
4. 确保词汇符合顶刊阅读习惯

---

使用建议

• 每次处理约500字段落
• 按顺序依次通过5个模块
• 保持原始论点和数据完整性
• 最终输出应呈现人类学者特有的审慎与自然节奏