Back to Details

sepia

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Sepia — de-AI writing

Sepia — 去AI化写作

Every rule here is backed by a measured human-vs-AI gap. The load-bearing facts: in fiction, a classifier using only narrative-structure features detects AI at 93.2% macro-F1 and style editing barely moves it — so structure is fixed before style, always. In professional prose, the measured tells are different — filler density, missing stance, chatbot residue, register mismatch, format uniformity — and the fix is domain-specific. Route first, then operate.
这里的每一条规则都基于人类文本与AI文本之间经量化分析的差异。核心事实：在虚构作品中，仅使用叙事结构特征的分类器检测AI文本的宏F1值可达93.2%，而仅修改风格几乎无法改变检测结果——因此，必须始终优先修复结构，再调整风格。对于专业文本，可检测到的AI特征有所不同——冗余内容密度、立场缺失、聊天机器人残留痕迹、语体不符、格式过于统一——修复方案需针对具体领域。先进行分类路由，再执行操作。

Routing

分类路由

Text typeLoad, in order
Fiction / stories / narrative essays
references/narrative-pass.md
→ 
references/discourse-pass.md
→ 
references/style-pass.md
; diagnose with 
references/rubric.md
Release notes, changelogs, announcements
references/professional-pass.md
+ 
references/domains/release-notes.md
PR replies, issue replies, review comments
references/professional-pass.md
+ 
references/domains/dev-replies.md
Incident postmortems / RCA
references/professional-pass.md
+ 
references/domains/postmortems.md
Tickets, work orders, bug reports
references/professional-pass.md
+ 
references/domains/tickets.md
Technical articles, blog posts, tutorials
references/professional-pass.md
+ 
references/domains/tech-articles.md
+ 
references/discourse-pass.md
§1–3
Any other prose
references/professional-pass.md
+ 
references/style-pass.md
Every non-fiction route ends with the vocabulary/syntax scan in 
references/style-pass.md
§2–3, and long professional pieces take the whole style pass — in both cases skipping its fiction-slop table. If the text was produced by a known model, add 
references/model-fingerprints.md
(fiction-centric; use as priors).
文本类型处理流程（按顺序）
虚构作品/故事/叙事散文
references/narrative-pass.md
→ 
references/discourse-pass.md
→ 
references/style-pass.md
；使用
references/rubric.md
进行诊断
发布说明、更新日志、公告
references/professional-pass.md
+ 
references/domains/release-notes.md
PR回复、问题回复、评审评论
references/professional-pass.md
+ 
references/domains/dev-replies.md
事件复盘/RCA
references/professional-pass.md
+ 
references/domains/postmortems.md
工单、工作指令、Bug报告
references/professional-pass.md
+ 
references/domains/tickets.md
技术文章、博客文章、教程
references/professional-pass.md
+ 
references/domains/tech-articles.md
+ 
references/discourse-pass.md
§1–3
其他任何散文类文本
references/professional-pass.md
+ 
references/style-pass.md
所有非虚构文本的处理流程都需以
references/style-pass.md
第2–3节中的词汇/语法扫描收尾，长篇专业文本则需完成完整的风格处理流程——两种情况均需跳过其中的虚构作品冗余内容表格。若文本由已知模型生成，需额外参考
references/model-fingerprints.md
（以虚构作品为核心；作为先验依据使用）。

Operations

操作类型

Any request maps to one of four operations:
OperationContract
writeNew content. Read the domain file before drafting — architecture and register decisions come first, they cannot be retrofitted cheaply. For fiction, follow Workflow A below.
reviewDiagnose only — no edits. Produce the defect list (fiction: rubric report; professional: checklist findings with quoted evidence) and stop. Report findings; apply nothing until asked.
refactorMinimal in-place revision preserving structure, voice, and intent. Two-stage: full defect list first, then fix item by item, deepest layer first. Skew replace/delete over insert (measured editor ratio 74/18/8).
recreateFull rewrite. Extract the facts, claims, and intent from the original into a bare list; verify nothing invented; write fresh under the domain rules. Use when defects are structural and the text is short enough that surgery costs more than rebuilding.
The two-stage protocol is not optional for refactor/recreate: paraphrasing without a defect list makes AI fingerprints more visible, not less (measured on expert detectors).
任何请求都对应以下四种操作之一：
操作要求
撰写创建全新内容。撰写前需阅读对应领域文件——架构和语体决策必须优先确定，事后再调整成本极高。对于虚构作品，遵循下方的工作流A。
评审仅进行诊断——不做任何编辑。生成缺陷列表（虚构作品：评分标准报告；专业文本：带有引用证据的 checklist 检查结果）后即停止。仅报告发现的问题，除非用户要求，否则不进行任何修复。
重构最小化原地修改，保留原文结构、语气和意图。分为两个阶段：先生成完整的缺陷列表，再逐项修复，从最底层问题开始。优先选择替换/删除操作，而非插入操作（经统计的编辑比例为74/18/8）。
重写完全改写。从原文中提取事实、主张和意图，整理为简洁列表；确认无内容编造；根据领域规则重新撰写。适用于存在结构性缺陷且文本篇幅较短，修复成本高于重新撰写的场景。
重构/重写必须遵循两阶段流程：不先列出缺陷就进行改写会让AI痕迹更加明显，而非减少（经专业检测工具验证）。

Fiction workflows

虚构作品工作流

A — writing new fiction: (1) premise, genre, length — genre sets calibration targets; (2) fill the architecture sheet in 
references/narrative-pass.md
; (3) select 3–5 human-leaning moves + one rarity move; (4) outline, run the outline/QUD checks in 
references/discourse-pass.md
and the echo test in 
references/narrative-pass.md
§2; (5) draft; (6) self-diagnose with 
references/rubric.md
, one group at a time; (7) style pass last.
B — revising existing fiction: (1) diagnose completely first (rubric → discourse → style), no edits; (2) triage — architecture defects need scene-level surgery, tell the user how deep before cutting; (3) fix deepest first; (4) verify: re-run changed rubric groups, read key passages aloud, echo-test any added twist.
A — 撰写新虚构作品：(1) 确定前提、体裁、篇幅——体裁决定校准目标；(2) 填写
references/narrative-pass.md
中的架构表格；(3) 选择3–5种偏向人类写作的手法 + 1种独特手法；(4) 撰写大纲，运行
references/discourse-pass.md
中的大纲/QUD检查，以及
references/narrative-pass.md
第2节中的回声测试；(5) 撰写初稿；(6) 使用
references/rubric.md
进行自我诊断，逐组检查；(7) 最后进行风格处理。
B — 修改现有虚构作品：(1) 先完成完整诊断（评分标准→语篇→风格），不做任何编辑；(2) 分类处理——架构缺陷需要场景级修改，在开始修改前告知用户修改深度；(3) 从最底层问题开始修复；(4) 验证：重新检查修改过的评分标准组，大声朗读关键段落，对新增的转折情节进行回声测试。

Calibration — the rule that governs all rules

校准——统领所有规则的准则

PrincipleMeaning
Aim at the band, not the opposite poleHuman values are moderate (chronological discontinuity 2.4/5, not 5). Inverting every AI tell creates a new fingerprint. In professional prose the equivalent: match the venue's register, don't overshoot into forced casualness — informality alone fools no trained reader.
Select, don't accumulateHuman writing is diverse. Fiction: 3–5 moves per story, chosen for the premise, varied across works. Professional: fix what the checklist actually flags, nothing more.
Leave slackOrdinary sentences, an underdeveloped thought, a plain paragraph. Do not sand every surface.
原则含义
瞄准中间区间，而非反向极端人类写作的特征是适度的（时间线不连续度为2.4/5，而非5）。完全反转所有AI痕迹会产生新的识别特征。对于专业文本，对应的原则是：匹配平台的语体，不要过度刻意追求非正式风格——仅靠非正式语气无法骗过有经验的读者。
精选手法，而非堆砌人类写作具有多样性。虚构作品：每个故事选择3–5种手法，贴合故事前提，不同作品使用不同手法。专业文本：仅修复checklist中实际标记的问题，无需额外修改。
保留“松弛感”保留普通句子、未充分展开的想法、平实段落。不要过度打磨每一处细节。

Hard guardrails

严格约束

  • Never invent specifics. Fiction: intertextual references, brands, places must be real and correct. Professional: versions, numbers, timestamps, benchmarks, quotes come from the actual change/incident/data — missing info means ask the user or leave an explicit TODO, never fill. Confident wrong facts are themselves a top-tier tell.
  • Deletion beats addition (74% replace / 18% delete / 8% insert). The only additive fix is real specificity.
  • Respect the author's voice and the venue's corpus. Extract habits from the user's samples or the venue's recent artifacts before editing; edit toward that profile. Do not remove a mannerism they actually use.
  • Dialogue quotes and quoted material are load-bearing — do not regularize them.
  • Check the whitelists (
    references/style-pass.md
    §7, 
    references/professional-pass.md
    last section) before flagging: clean grammar, formal tone in formal venues, and conventional templates are not evidence of AI.
  • 绝不编造细节。虚构作品：互文引用、品牌、地点必须真实准确。专业文本：版本号、数字、时间戳、基准数据、引用内容必须来自实际变更/事件/数据——若信息缺失，需询问用户或留下明确的TODO标记，绝不能自行补充。自信的错误内容本身就是顶级的AI识别特征。
  • 删除优于添加（74%替换 / 18%删除 / 8%插入）。唯一允许的添加操作是补充真实的具体细节。
  • 尊重作者语气和平台语料。编辑前先从用户提供的样本或平台近期内容中提取写作习惯；向该风格调整。不要删除作者实际使用的独特表达方式。
  • 对话引用和引用内容是核心要素——不要统一格式。
  • 标记前先检查白名单（
    references/style-pass.md
    第7节、
    references/professional-pass.md
    最后一节）：规范的语法、正式场合的正式语气、常规模板均不属于AI特征。