Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add mguozhen/multi-platform-publisher --skill lieflatgit clone --depth 1 https://github.com/mguozhen/multi-platform-publisherWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/mguozhen/multi-platform-publisher/lieflat)<a href="https://agentmods.dev/skills/mguozhen/multi-platform-publisher/lieflat"><img src="https://agentmods.dev/badge/skills/mguozhen/multi-platform-publisher/lieflat/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/mguozhen/multi-platform-publisher/lieflat"><img src="https://agentmods.dev/badge/skills/mguozhen/multi-platform-publisher/lieflat.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00080 | $0.08048 |
| Opus 5.5 | $0.00032 | $0.03219 |
| Sonnet 5.5 | $0.00016 | $0.01610 |
| Haiku 4.5 | $0.00008 | $0.00805 |
Grade A, and why
lieflat-less-ai-tone scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 22d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
This is a copy
100% identical to lieflat-less-ai-tone — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 454 lines — stays where its author put it; the contents beside it link to each section on GitHub.
去AI味
识别并改写文本中的 AI 写作痕迹。规则基于人类文章与多个模型输出的实测对比整理，每条都有可定位的触发标记。先判断体裁，再处理 AI 味。不能为了“像人写的”强加人称、口语或短段落。
有风格参考时先读它
本 skill 的多条规则需要判断“原语体”。第 7 条对文学体裁按原有风格判断，第 4 条和第 5 条要看作者本来的标点习惯，第 8 条要看原文有没有给出具体数据。
如果同一目录下有描述目标风格的文档（写作风格指南、语言特征分析、语言DNA.md 一类的产物），先读它再改写，把它的句长分布、标点习惯和用词偏好作为判断依据。
风格文档与本规则冲突时，以风格文档为准。 那是目标作者的真实写法，不是 AI 痕迹。比如某作者本来就爱用破折号，就不该按第 4 条删掉。
硬性边界（最高优先级）
本 skill 采用白名单式改写。只能处理“改写规则”中明确列出的规范，不能凭一般写作经验修改其他内容。没有命中任何规则的句子必须逐字保留。
一句话即使命中规则，也只能改动解决该问题所必需的部分。不得顺便润色，不得替换没有问题的词语，不得调整语气、详略或信息密度。对某处是否命中规则没有把握时，保持原文。
文章结构不在处理范围内。标题层级、章节顺序、段落数量与顺序、列表、表格、引用和代码块的位置必须原样保留。除非用户另行明确要求，否则不能概括、扩写、删减观点或重组论证。
信息守恒：不许新增，不许删减
原文的观点和事实必须完整保留。以下内容绝不能凭改写添加，哪怕只是为了让句子读起来更具体、更像人写的。
- 姓名、机构名、职务、地名
- 数字、比例、金额、时长、次数
- 日期、时间、先后顺序
- 引语、对话、别人的说法
- 引用来源、出处、链接
- 因果、动机、心理活动
- 任何原文没写的细节、场景或案例
反方向同样禁止。不能删掉原文的观点、结论、限定词和让步（“可能”“通常”“在某些情况下”“据说”）。把“可能提升”改成“提升”是篡改语气强度，不是去 AI 味。
判断办法：改写后的每个实词，都要能在原文里指出出处。指不出来就是新增，必须撤销。
“把抽象写具体”不是本 skill 的职责。原文抽象就让它抽象。第 8 条只允许把原文已有的具体数据提到概括词的位置，不允许自己造数据。
用法
用户丢一段文字过来，你：
- 先锁定原文框架。标题层级、章节顺序、段落顺序、列表、表格、引用和代码块的位置都不能改
- 只按“改写规则”逐项检查，不增加清单外的判断标准
- 只改明确命中规则的文字，并把改动控制在解决该问题所需的最小范围。其他文字逐字保留，信息不增不减
- 按文末验收清单复查，再输出改写后的全文
不需要解释改了什么。用户问时再说明。
不改的情况
- 文章框架：保留标题层级、章节顺序、段落顺序、列表、表格、引用和代码块的位置
- 对话体、口语体：本来就是自然表达，不要统一改写
- 问句：正文问句、反问、问句小标题都不改，见下文「不作为改写理由」
- 文学性表达：散文、小说里的排比和比喻要按原有风格判断
- 体裁要求：学术、新闻、公文等体裁不必强加“我/你”，也不必拆成短段落
- 代码块、外语段落：不属于改写范围
- 引用他人原文：不要改动引文
不作为改写理由
以下特征看着像 AI 味，实测站不住，不能据此改文字。这张表和「不改的情况」一样是硬约束，不是参考。
| 特征 | 为什么不改 |
|---|---|
| 句长、段落长度不够参差 | 实测与人类写作无差别。不要为了制造节奏调句长或拆段落 |
| 单字虚词偏少（就/很/了） | AI 确实偏少，但方向是补不是删。补虚词无法验收，容易把正式文章改成口语 |
| 反复写全称、少用代词 | 实测人类比 AI 更常重复同一名词 |
| 被动句 | 抽象被动（“被认为／被视为”）远低于收录门槛 |
| 名词化、长句本身 | 现代汉语书面语的正常写法，人类同样这么用 |
| 正文里的“首先……其次” | 与人类写作无差别。第 6 条只管小标题编号 |
| 句内同构排比 | “提升效率，降低成本，优化体验”“更高效、更专注、更有创造力”这类人类用得不比 AI 少。只处理句间重复，见第 3 条 |
| 问句、设问、问句小标题 | 正文问句人类远多于 AI，删掉更不像人写 |
| 比喻本身、比喻独立成段 | 人类用得比 AI 多。只处理理想化拟人喻体，见第 7 条 |
| 抽象名词配具体动词 | “时间替他保管了细节”“焦虑显出形状”这类两边都几乎不写 |
规则的依据是触发标记可定位，不是频率高低。各条规则的实测数据、被推翻的预设和测量脚本见 RESEARCH.md。
What ships with it
9 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 22d ago First seen · 454 lines · 80 tokens per session scan A c55dec1765c2
lieflat-less-ai-tone is a skill published in the GitHub repository mguozhen/multi-platform-publisher (10 stars, last pushed 22d ago), licensed MIT. It adds 80 tokens to every session and 8,048 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. It is 100% identical to lieflat-less-ai-tone, differing in 0 lines, and is treated as a copy.
Other skills, from other repositories
opik-diagnose
Surface the Opik traces worth a developer's attention, ranked by signal — Diagnostics issues first, then errors, failed tool calls, latency, regressions, and low online-eval scores. With the Opik MCP connected it lists the project's agentinsightsissue entities, offers to turn Diagnostics on when the project has it…
opik-test
Turn a failing Opik trace (or a described failure) into a repeatable regression check — a test-suite item with the trace's input and one or two binary assertions — so a fix can be verified by the compare skill. Works over the SDK; uses the MCP write tool when connected. Returns the suite, the item, and the assertion.…
qmclaw-workbench
A local simulator for superconducting quantum-bit experiments, including measurements used to study and tune a qubit. It produces bounded synthetic results without connecting to laboratory hardware or quantum cloud services.
cortex-explore-memory
Explore the memory system's state, find gaps in knowledge, assess coverage, and get diagnostic information. Use when the user asks 'what does my memory look like', 'show me memory stats', 'what am I missing', 'how good is my knowledge', 'memory health', 'show coverage', 'find gaps', 'what topics are weak', or when you…
quantum-fermi-hubbard-vqe
A variational quantum simulation of the open-boundary Fermi–Hubbard chain, a model of particles hopping between sites and interacting when they occupy the same system. It maps the model to qubits with the Jordan–Wigner transformation and searches for a low-energy state.
quantum-grover
A Qiskit implementation of Grover's algorithm, a quantum search method that increases the chance of finding a marked item. It can explain the method, build an explicit quantum circuit, or run it with chosen inputs.