Thesis Creator
Stars-OC/thesis-creator
Walks Chinese undergraduates through writing a graduation thesis from topic to Word export, with text-similarity reduction and AI-text rate rewriting and checks.
Reduces the AI-text detection rate of academic papers by scanning for templated phrasing and making small deterministic edits that keep citations, data and terminology.
SKILL.md written in Chinese; this summary is our English description.
$ npx skills add xiaofenggan01/aigc-reduce --skill aigc-reduce -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install xiaofenggan01/aigc-reduce aigc-reduce --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
Claude Code skills documentation · loads skills from .claude/skills/
Install the "aigc-reduce" agent skill from https://github.com/xiaofenggan01/aigc-reduce/tree/main into .claude/skills/aigc-reduce/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "aigc-reduce", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add xiaofenggan01/aigc-reduce --skill aigc-reduce -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install xiaofenggan01/aigc-reduce aigc-reduce --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "aigc-reduce" agent skill from https://github.com/xiaofenggan01/aigc-reduce/tree/main into .agents/skills/aigc-reduce/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "aigc-reduce", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add xiaofenggan01/aigc-reduce --skill aigc-reduce -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install xiaofenggan01/aigc-reduce aigc-reduce --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "aigc-reduce" agent skill from https://github.com/xiaofenggan01/aigc-reduce/tree/main into .cursor/skills/aigc-reduce/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "aigc-reduce", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add xiaofenggan01/aigc-reduce --skill aigc-reduce -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install xiaofenggan01/aigc-reduce aigc-reduce --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "aigc-reduce" agent skill from https://github.com/xiaofenggan01/aigc-reduce/tree/main into .gemini/skills/aigc-reduce/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "aigc-reduce", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install xiaofenggan01/aigc-reduce aigc-reduceInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add xiaofenggan01/aigc-reduce --skill aigc-reduce -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "aigc-reduce" agent skill from https://github.com/xiaofenggan01/aigc-reduce/tree/main into .github/skills/aigc-reduce/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "aigc-reduce", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add xiaofenggan01/aigc-reduce --skill aigc-reduce -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install xiaofenggan01/aigc-reduce aigc-reduce --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "aigc-reduce" agent skill from https://github.com/xiaofenggan01/aigc-reduce/tree/main into .opencode/skills/aigc-reduce/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "aigc-reduce", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
aigc-reduceReduces the AI-text detection rate of academic papers by scanning for templated phrasing and making small deterministic edits that keep citations, data and terminology.
The skill is written in Chinese for Claude Code and Codex and is built around the idea that the agent should diagnose and plan rather than rewrite. A scanning script, `scripts/aigc_scan.py`, finds AI-style patterns, and the agent then applies targeted replacements instead of regenerating text, because a full model rewrite is said to add more AI traces. Its notes refer to how detectors such as CNKI, Wanfang, PaperPass and PaperPure work.
Rules require marking protected spans first, such as reference numbers, figure and formula labels, data and percentages, technical terms and direct quotes, and leaving them untouched. Edits aim past 40% of the text, but only through restructuring sentences and removing stock phrases, never through slang, filler or invented facts, and academic register wins whenever the two goals conflict. Work runs in rounds, starting with removing AI traces by word, sentence and paragraph edits and then adding written academic features. Word files are edited with python-docx.
4 steps, taken from the step headings in SKILL.md.
Read from SKILL.md and the folder at commit 3cc20dc. It shows what the files ask for, not the result of running them.
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Ships 1 file in scripts/ (Python), which the agent can run.
Shell commands in SKILL.md call:
python3From the folder's file list and the shell code blocks in SKILL.md.
Links to these hosts (documentation or services it may open):
matrix.tencent.comFrom URLs in SKILL.md, links to its own repository left out.
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
AIGC Detection Rate Reducer loads about 1.5k tokens when it runs, and up to ~12k if it reads all its reference files. Until then it costs about 63 tokens; SKILL.md has 290 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.
The full file from xiaofenggan01/aigc-reduce at commit 3cc20dc, republished under its MIT licence (© xiaofenggan01). 290 words, ~1,487 tokens.
.claude/skills/aigc-reduce/SKILL.md (or your agent's skills folder). This skill also uses 12 other files; get the full folder from GitHub.你是一位学术论文 AIGC 降重专家。你的角色是诊断器和规划器——扫描文本中的 AI 痕迹,规划确定性替换方案,然后逐项执行。
本 skill 同时面向 Claude Code 和 Codex。Claude Code 的原有安装路径、触发语义和文档定位保持不变;Codex 通过 YAML frontmatter、相对路径脚本调用和 agents/openai.yaml 获得兼容支持。
调用脚本时优先使用当前环境可用路径:
# Claude Code 默认安装路径
python3 ~/.claude/skills/aigc-reduce/scripts/aigc_scan.py <file.txt>
# 通用方式:先进入本 skill 根目录
cd <aigc-reduce-skill-root>
python3 scripts/aigc_scan.py <file.txt>用 AI 重写 AI 文本会叠加 AI 指纹(PaperPure 实测:重写后检测率升至 100%)。所有改写必须通过确定性字符串替换完成——直接操作原文的特定片段,不经过 LLM 的 token 采样。
修改比例与检测规避率直接相关:轻度修改(<20%)仅 15-25% 规避;中度(20-40%)约 35-50%;深度修改(>40%)可达 60-80%。如果修改率不足 40%,继续补充操作。
修改率的贡献只能来自两类手段:
明令禁止用以下方式凑修改率:加入口语化/网络用语/情绪词、把学术表达改成大白话、堆砌破折号或碎句、无意义的同义词轮换(换了词但语义和语体都没变)。这些操作能提高字面修改率,但会破坏学术语体,属于"为降重而降重"。
当"达到 40%"与"保持学术语体"冲突时,语体优先。 宁可修改率停在 38%,也不能靠口语化把它推到 45%。语体规范是硬底线,修改率是优化目标。
每次操作只改一小处,保留原文大部分 token 不变。相邻段落的句式结构、开头方式和长度必须不同。
"去 AI 味"和"保持学术语体"是两个独立目标。去 AI 的默认方向会漂向口语化——因为口语色彩是降低检测率的强信号——必须主动约束。降重后出现网络用语、情绪词、口语表达,就算过了机审也过不了人审。
Register Anti-goals(降重时绝对不做的事):
正确方向:去 AI 味 = 打破模板化的句式结构(语序、句式、论证组织),让文本回到具体、规范的书面学术语体,而不是变短、变口语。正向标准与"降重不足 / 过度口语 / 恰当"三级对照见 references/positive-style-academic.md,口语化负面清单见 references/replacement-tables.md。
降重分三轮执行,每轮有不同目标。单轮替换解决不了的问题,多轮迭代可以。
划禁改片段 → 扫描定位 → 词级替换 → 句级重构 → 段落调整
改写前先划 protected spans:按 references/protected-spans.md 圈出参考文献引用/编号、图表编号/公式、数据与百分比、专业术语/定理、直接引语。这些片段在后续所有轮次中原样保留,不参与改写,更不能为降重编造。
运行扫描脚本:
python3 ~/.claude/skills/aigc-reduce/scripts/aigc_scan.py <file.txt>
# 或在本 skill 根目录执行:
python3 scripts/aigc_scan.py <file.txt>| 操作层级 | 贡献 | 主要手段 |
|---|---|---|
| 词级替换 | 10-15% | 模板词、AI 高频词、动词的系统性替换 |
| 句级重构 | 15-20% | 语序重组、长句拆分、被动改主动、主语更换 |
| 段落调整 | 10-15% | 对称段长打破、段间过渡 |
详细替换表见 references/replacement-tables.md。替换表中有 中文 AI 高频词清单,这些词在 AI 生成文本中出现频率远高于人类文本,必须全部替换。
对 docx 文件通过 python-docx 定位段落并替换文本。每处替换使用不同的替代词。
核心洞察:只去除 AI 痕迹是不够的,干净但节奏均匀的文本仍像 AI。但"注入的特征"必须是书面学术特征,不是口语特征。注入方向错了,就变成 issue 里被吐槽的"不像论文,成了说明文"。
本轮 register 底线(先读铁律 4 的 Anti-goals,再做本轮):下面每一条注入,都以"保持学术书面语体"为前提;任何一条一旦滑向口语,宁可不注入。
在第一轮修改率达标的基础上,对每个段落做以下检查和补充:
节奏工程(提高突发性,但保持书面完整句):
审慎推断语气(学术书面,非口语):
具体化操作细节(仅当原文本身包含、可据实补全时):
references/protected-spans.md)声音校准(如果用户提供了参考文本):
审视修改后的文本,回答:"这段话为什么看起来还像 AI 写的?"
逐项排查(详细说明见 references/ai-patterns.md):
references/replacement-tables.md 负面清单,命中即改回书面学术表达)references/protected-spans.md)对剩余痕迹做确定性替换。如果发现普遍性问题,回到第一轮或第二轮补充操作。特别地:若审计发现任何口语化表达,必须无条件改回书面语,这优先于修改率目标。
再次运行扫描脚本,对比各维度指标变化。验证修改率是否 >40%。可在免费平台验证:腾讯朱雀 https://matrix.tencent.com/ai-detect/ai_gen_txt
原文(AI 痕迹明显):
SPI和KC含量的增加导致凝胶的L*显著降低(P<0.05),归因于两方面,首先,由于SPI本身呈黄绿色、KC呈浅黄色,浓度升高使凝胶内有色物质总量增加。此外,蛋白质分子的聚集及交联度增加会增强对光线的吸收,降低反射光线强度。上述结果一致表明,蛋白与多糖浓度的变化显著影响复合凝胶的色泽。
第一轮后(去除 AI 痕迹):
SPI和KC用量增大的同时,L*值出现了明显下降(P<0.05)。主要有两方面原因:一是SPI本身偏黄绿色、KC偏浅黄色,浓度上来后凝胶体系中的色素总量随之增多;二是蛋白分子聚集和交联程度提高后对入射光的吸收更加强烈,反射光量因此减少[20]。把以上几组数据放在一起看,蛋白和多糖的用量变化对凝胶外观颜色产生了不可忽视的影响。
第二轮后(注入书面学术特征 — 注意节奏变化与审慎推断语气,全程保持书面语体):
SPI和KC用量增大的同时,L*值出现了明显下降(P<0.05)。主要有两方面原因:一是SPI本身偏黄绿色、KC偏浅黄色,浓度上来后凝胶体系中的色素总量随之增多。二是蛋白分子聚集和交联程度提高后对入射光的吸收更加强烈,反射光量因此减少[20]。综合这几组数据可以看到,蛋白和多糖的用量变化确实改变了凝胶的外观颜色,不过两者各自的贡献比例仍有待进一步拆解。
⚠️ 反面示范(第二轮做错的样子,切勿模仿):把上文改成"浓度一上来颜色就崩了""这个趋势我们测下来很稳,基本没跑偏""说白了就是料越多越暗"——修改率是够了,但已经不是论文语体。
第三轮后(Anti-AI 审计 — 检查无剩余问题):
(如果第二轮后没有发现剩余 AI 痕迹,第三轮无需修改。)
references/detection-principles.md 的"护栏"段。references/positive-style-academic.md — 降重正向标准 + "降重不足 / 过度口语 / 恰当"三级对照(守语体的核心)references/protected-spans.md — 改写前必须圈出的禁改片段(引用、公式、数据、术语),禁止为降重编造事实references/replacement-tables.md — 词级/句级/段落级替换表 + 中文 AI 高频词清单 + 口语化负面清单references/ai-patterns.md — 10 种深度 AI 痕迹识别模式references/detection-principles.md — 主流检测器技术原理与弱点evals/benchmark.md — 双向 benchmark(SF 该改 / SNF 不该误杀·不该口语化)scripts/aigc_scan.py — 自动化 AI 特征扫描脚本(9 维度,含口语化与破折号密度)© xiaofenggan01, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
SKILL.md and 12 other files (scripts, references) in the repository root of xiaofenggan01/aigc-reduce.
Open the folder on GitHubat commit 3cc20dc
AIGC Detection Rate Reducer next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| AIGC Detection Rate Reducer this skillxiaofenggan01/aigc-reduce | 617 | — | ~1.5k | Automated safety check: Pass | MIT | |
| Thesis CreatorStars-OC/thesis-creator | 230 | — | ~2.8k | Automated safety check: Pass | MIT | |
| Aigc Detectorfree-revalution/AIGC-Detector-Pro | 141 | — | ~2.8k | Automated safety check: Pass | MIT | |
| User-Facing Text Cleanupguillaumemeyer/watermarks-remover | 24k | — | ~3.5k | Automated safety check: Pass | MIT | |
| Chinese Text Humanizerop7418/Humanizer-zh | 19k | — | ~2k | Automated safety check: Pass | MIT | |
| Natural Japanese Business Writingcoji/natural-japanese | 1.9k | — | ~2.1k | Automated safety check: Pass | MIT |
Stars-OC/thesis-creator
Walks Chinese undergraduates through writing a graduation thesis from topic to Word export, with text-similarity reduction and AI-text rate rewriting and checks.
free-revalution/AIGC-Detector-Pro
Academic paper AI content detection, rewriting, and thesis writing assistant.
guillaumemeyer/watermarks-remover
Audits prose for invisible Unicode characters and rewrites it while keeping facts, citations, code and required disclosures unchanged and the writer's voice intact.
op7418/Humanizer-zh
Edits Chinese articles, comments and documents to remove filler, repetition and template phrasing while keeping the facts, the level of certainty and the author's voice.
coji/natural-japanese
Writes and edits Japanese business documents so they read clearly and naturally, removes AI-sounding phrasing and can score how AI-like a text reads.
iflytek/skillhub
Audits and rewrites formulaic, AI-sounding prose while keeping facts, voice and format, using a local Python scorer and inspect-only, rewrite or embedded-gate modes.
Works with
Categories
Reduces the AI-text detection rate of academic papers by scanning for templated phrasing and making small deterministic edits that keep citations, data and terminology. The skill is written in Chinese for Claude Code and Codex and is built around the idea that the agent should diagnose and plan rather than rewrite.py`, finds AI-style patterns, and the agent then applies targeted replacements instead of regenerating text, because a full model rewrite is said to add more AI traces.
AIGC Detection Rate Reducer fits situations like: lowering the AI-detection score of a thesis chapter before submission; finding templated AI phrasing in a draft that needs revising; revising a Word document of a paper while keeping citations and data unchanged; checking that edits have kept an academic tone rather than a casual one.
Run `npx skills add xiaofenggan01/aigc-reduce --skill aigc-reduce -a claude-code`. Or copy the skill folder (the xiaofenggan01/aigc-reduce repository) into .claude/skills/aigc-reduce in your project. Claude Code loads it when a task matches its description.
Run `npx skills add xiaofenggan01/aigc-reduce --skill aigc-reduce -a codex`. Or copy the skill folder (the xiaofenggan01/aigc-reduce repository) into .agents/skills/aigc-reduce in your project. Codex loads it when a task matches its description.
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add xiaofenggan01/aigc-reduce --skill aigc-reduce -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/aigc-reduce, .gemini/skills/aigc-reduce, .github/skills/aigc-reduce and .opencode/skills/aigc-reduce in your project.
Going by SKILL.md and its folder, AIGC Detection Rate Reducer needs Python for the scripts in its folder and the command-line tools its instructions call (python3). Our summary lists: Python 3 to run `scripts/aigc_scan.py`; python-docx for Word files.
SKILL.md names 1 domain. As links in the text: matrix.tencent.com. This is read from the text; nothing was executed.
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
AIGC Detection Rate Reducer is published under the MIT licence (from the LICENSE file in the skill folder). It allows redistribution, so the full SKILL.md is shown on this page.
About 1.5k tokens (SKILL.md is roughly 5.9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 11k tokens, read only when the agent opens those files.
Skills that share tags, products or a category with AIGC Detection Rate Reducer: Thesis Creator (Stars-OC/thesis-creator, 230 stars), Aigc Detector (free-revalution/AIGC-Detector-Pro, 141 stars), User-Facing Text Cleanup (guillaumemeyer/watermarks-remover, 24k stars) and Chinese Text Humanizer (op7418/Humanizer-zh, 19k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
xiaofenggan01 (a GitHub user) maintains it in xiaofenggan01/aigc-reduce, which has 617 GitHub stars. The repository was last updated on July 8, 2026.
Source: xiaofenggan01/aigc-reduce on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.