classical-chinese

以文言文压缩输出、节省 token。当用户要求文言回复、节省 token、省输出时触发。支持三档：极致(/e/Extreme)、适度(/m/Medium)、轻微(/l/Light)。

Security Scan

The skill's requirements and instructions match its stated purpose (compressing replies into Classical Chinese to save tokens); it is instruction-only, requests no credentials, and has no install steps.

Purpose & Capability

Name/description (文言压缩以省 token) align with what the skill actually does: it transforms user text into more compact Classical Chinese using the included rules file. No unrelated credentials, binaries, or installs are requested.

Instruction Scope

SKILL.md stays within the stated scope: it prescribes transformation rules, preservation of technical artifacts (code, paths, identifiers), compression levels, and reporting of compression metrics. It does not instruct reading unrelated files, contacting external endpoints, or accessing secrets.

Install Mechanism

No install spec and no code files beyond static reference docs; nothing is downloaded or written to disk. This minimizes install-time risk.

Credentials

The skill declares no environment variables, credentials, or config paths and the instructions do not reference any. Requested privileges are proportionate to its function.

Persistence & Privilege

always:false and default autonomous invocation are used. The skill does not request persistent system-wide presence or modification of other skills. Its privileges are appropriate.

Assessment

This skill is internally consistent and needs no credentials or installs. Before enabling broadly, test it on representative prompts: because 'token saving' is the primary goal it may omit words or condense nuance (which can reduce clarity or change meaning for complex technical instructions). Always specify compression level explicitly when precision matters (e.g., /m or /l), and verify that code snippets, paths, versions, and error messages remain intact in output as promised.

文言压缩（省 Token）

核心原则

以文言代白话，最大化信息密度。省 token 为首要目标，可读性按需妥协。

压缩等级

等级	命令	目标节省	规则
极致	/e	50%+	能一字不两字，省主语/虚词/关联词
适度	/m	30-40%	简练文言，保基本可读（默认）
轻微	/l	~20%	浅白文言，优先可读

默认 /m。用户可在首句指定，如"【极致】解释递归"。

使用流程

1. 识别等级（/e /m /l 或默认 /m）
2. 转写为文言（调用 references/compression_rules.md）
3. 技术内容原文保留（代码、路径、专有名词、错误信息）
4. 估算压缩率（见下）

量化验证

回复后简记：[原~N字→文~M字，省X%]

• 若节省 < 预期下限，等级不变，下次更精简
• 若节省 > 预期上限，可适当放宽

边界处理

用户输入为文言

直接以同档或更高压缩等级回复，无需转换。

专有名词

• 已知：框架/库/品牌名 → 保留原文（React, Python, Git）
• 未知：新产品/概念 → 音译或原文，必要时加注

歧义处理

• 文言多义时，优先取最简义
• 关键术语首次用原文括注：功能→功能（function）

连续对话

• 切换等级须显式标注
• 上下文已建立的文言术语保持一致
• 追问若需详细解释，可临时降一档

低压缩例外

以下情况可临时降档：

• 复杂逻辑首次解释
• 错误诊断需清晰步骤
• 用户明确要求详细

降档时注明：[降档说明]

技术内容保护

以下原文不译：

• 代码、变量名、函数名、类名
• 路径、URL、端口号
• 错误代码、日志内容
• 公式、数字、版本号
• JSON/XML/HTML 等标记语言