ru-score
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseRussian Text Quality Score
俄语文本质量评分
Score the text provided in $ARGUMENTS (or the most recent Russian text output if no arguments) using the ru-text scoring rubric.
使用 ru-text 评分规则对 $ARGUMENTS 中提供的文本进行评分(若没有参数,则对最近的俄语文本输出进行评分)。
Procedure
流程
Where the reference files live
参考文件存放位置
This skill reads the corpus that ships with the ru-text skill, which is installed
alongside it. Locate that folder once, then read the named files from it:
- Look for a directory named whose parent directory is named
references, and which containsru-text. Every file named below sits in that folder.info-style.md - Search with file tools, never with a shell. on a pattern like
Glob, or whatever file search the host offers, then**/ru-text/references/info-style.md. This command has no use for a command line:Readis on itsBashlist, and hosts that do not implement that field still refuse it — measured in Claude Cowork on 12.08.2026, where an openingdisallowed-toolsreturned «Permission to use Bash has been denied» in red before the search found the corpus anyway. The call bought nothing and cost the reader a scare.ls - In Claude Code the plugin root is also available directly, which saves the search.
- Do not guess a path. If the folder cannot be found, say so and stop — a check run against remembered rules instead of the corpus is not this command, and reporting one as the other is the failure this whole product exists to prevent.
-
Load rubric — readfor the full rubric with anchors.
scoring.md -
Determine domain — identify whether the text is UI/interface, business email, article, or general:
- UI text → also load for the Reader Precision dimension
ux-writing.md - Business email → also load for the Reader Precision dimension
business-writing.md - General → use only
info-style.md
- UI text → also load
-
Evaluate each dimension separately — score in this order:
- T — Типографика (weight 0.15): quotes, dashes, spaces, special characters per
typography.md - Ч — Чистота языка (weight 0.25): stop-words, bureaucratic language, clichés, passive voice per +
info-style.mdanti-patterns.md - Г — Грамотность (weight 0.20): punctuation, agreement, tautology per +
editorial-grammar.mdeditorial-punctuation.md - С — Структура (weight 0.20): logical flow, paragraphs, transitions, heading use per structure section +
info-style.mdaddenda.md - Ц — Точность для читателя (weight 0.20): facts, evidence, reader benefit, actionability per domain rules
- T — Типографика (weight 0.15): quotes, dashes, spaces, special characters per
-
For each dimension — assign a score (0.0–10.0) using the rubric anchors in scoring.md. List 1–3 specific issues, quoting the problematic text fragment.
-
Apply non-compensatory rules:
- Any dimension < 3.0 → cap total at 5.0
- Typography < 4.0 → cap total at 7.0
- Grammar < 4.0 → cap total at 7.0
- Use the most restrictive cap when multiple conditions trigger
-
Compute composite score:
S = round₁(T × 0.15 + Ч × 0.25 + Г × 0.20 + С × 0.20 + Ц × 0.20) -
Short text warning — if text is under 50 words, prepend a reliability caveat.
此 skill 会读取与其一同安装的 ru-text skill 所附带的语料库。只需定位一次该文件夹,然后从中读取指定文件:
- 查找一个名为 的目录,其父目录名为
references,并且其中包含ru-text。下面列出的每个文件都位于该文件夹中。info-style.md - 使用文件工具搜索,绝不要使用 shell。 对类似 的模式使用
**/ru-text/references/info-style.md,或使用宿主提供的任何文件搜索方式,然后Glob。此命令不需要命令行:Read在其Bash列表中;未实现该字段的宿主也仍然会拒绝它——2026 年 8 月 12 日在 Claude Cowork 中实测,开头的disallowed-tools返回了红色的“Permission to use Bash has been denied”,不过搜索最终还是找到了语料库。那次调用毫无收益,反倒让读者受了一场惊吓。ls - 在 Claude Code 中,插件根目录也可直接访问,从而省去搜索。
- 不要猜测路径。 如果找不到该文件夹,请如实说明并停止——根据记忆中的规则而不是语料库进行检查并不是此命令的用途;把二者混为一谈,正是这个产品存在的全部意义所要防止的失败。
-
加载评分规则 — 读取,其中包含带锚点的完整评分规则。
scoring.md -
确定领域 — 判断文本属于 UI/界面、商务邮件、文章还是通用类型:
- UI 文本 → 同时加载 ,用于“对读者的精确性”维度
ux-writing.md - 商务邮件 → 同时加载 ,用于“对读者的精确性”维度
business-writing.md - 通用 → 仅使用
info-style.md
- UI 文本 → 同时加载
-
分别评估每个维度 — 按以下顺序评分:
- T — Типографика(排版)(权重 0.15):引号、破折号、空格、特殊字符,依据
typography.md - Ч — Чистота языка(语言纯净度)(权重 0.25):停用词、官僚语言、陈词滥调、被动语态,依据 +
info-style.mdanti-patterns.md - Г — Грамотность(语法正确性)(权重 0.20):标点、一致性、同义反复,依据 +
editorial-grammar.mdeditorial-punctuation.md - С — Структура(结构)(权重 0.20):逻辑流、段落、过渡、标题使用,依据 结构部分 +
info-style.mdaddenda.md - Ц — Точность для читателя(对读者的精确性)(权重 0.20):事实、证据、读者收益、可操作性,依据领域规则
- T — Типографика(排版)(权重 0.15):引号、破折号、空格、特殊字符,依据
-
对每个维度 — 使用 scoring.md 中的评分锚点打分(0.0–10.0)。列出 1–3 个具体问题,并引用有问题的文本片段。
-
应用非补偿规则:
- 任一维度 < 3.0 → 总分上限为 5.0
- 排版 < 4.0 → 总分上限为 7.0
- 语法 < 4.0 → 总分上限为 7.0
- 多个条件触发时,取最严格的上限
-
计算综合得分:
S = round₁(T × 0.15 + Ч × 0.25 + Г × 0.20 + С × 0.20 + Ц × 0.20) -
短文本警告 — 如果文本少于 50 个词,请在结果前添加可靠性提示。
Score labels
分数等级
| Score | Label |
|---|---|
| 9.0–10.0 | Эталонный |
| 7.0–8.9 | Хороший |
| 5.0–6.9 | Средний |
| 3.0–4.9 | Слабый |
| 0.0–2.9 | Критический |
| 分数 | 等级 |
|---|---|
| 9.0–10.0 | 典范(Эталонный) |
| 7.0–8.9 | 良好(Хороший) |
| 5.0–6.9 | 中等(Средний) |
| 3.0–4.9 | 较弱(Слабый) |
| 0.0–2.9 | 严重(Критический) |
Output format
输出格式
Read-only — do NOT modify the source files. reports a score and issues; it must not write to, edit, or overwrite the analysed file(s).
ru-scoreIn Claude Code this is hardened by , which removes the listed tools while the command is active — verified on 2.1.220. The list removes the direct file-writers , and , plus the command-executing tools we identified and tested: , and . It names the paths we closed, not every path that exists. pre-approves tools; it does not restrict them.
disallowed-toolsWriteEditNotebookEditBashPowerShellMonitorallowed-toolsFour limits, stated plainly. A connected MCP server can expose write tools that a per-tool denial list cannot know about. A named subagent whose own definition grants it is governed by that definition, not by this list. A later Claude Code release can add a tool this file does not name. And on hosts without support the rule is an instruction rather than a platform guarantee.
tools: Writedisallowed-toolsThe contract itself is behavioural and does not depend on the list: this command returns a score and never writes.
The report shape is not written twice. Read section «Output format» of —
the mandated heading, the five Russian dimension labels, the per-cell limit of one to
three issues with quoted fragments, the weighted formula, the cap note, and the closing
«Что оценка не измеряет» section — and follow it exactly.
scoring.mdThat file is the single source of it. A copy lived here until v2.0 and sat outside the
no-loss gate, so the two could drift and only a reader comparing them would notice.
If a dimension has no issues, write «Замечаний нет» in the remarks column.
只读——请勿修改源文件。 报告分数和问题;它绝不能写入、编辑或覆盖被分析的文本文件。
ru-score在 Claude Code 中,这一点通过 加强:命令激活期间会移除列出的工具——已在 2.1.220 上验证。该列表移除了直接写文件的 、 和 ,以及我们识别并测试过的命令执行工具:、 和 。它列出的是我们关闭的路径,而不是所有存在的路径。 预先批准工具;它并不限制工具。
disallowed-toolsWriteEditNotebookEditBashPowerShellMonitorallowed-tools四个限制,直说如下。已连接的 MCP 服务器可能会暴露逐工具拒绝列表无法知晓的写入工具。一个具名子代理,如果其自身定义授予了 ,则受该定义约束,而非此列表约束。未来的 Claude Code 版本可能会添加本文件未命名的工具。在支持 的宿主机上,此规则是平台保证;在不支持的宿主机上,它只是一条指令。
tools: Writedisallowed-tools这种约定本身是行为性的,并不依赖列表:此命令返回分数,且从不写入。
报告格式不会重复书写。阅读 的「Output format」部分——规定的标题、五个俄语维度标签、每个单元格 1 到 3 个问题并附引用片段的限制、加权公式、上限说明,以及末尾的「Что оценка не измеряет」部分——并严格遵循。
scoring.md该文件是唯一的权威来源。在 v2.0 之前,这里曾有一份副本,且处于“无损失”门禁之外,因此两者可能发生偏差,只有对照阅读的读者才会注意到。
如果某个维度没有问题,请在备注列中写「Замечаний нет」。