← 返回
未分类 中文

Qian Jiu Jiu

A Chinese message preprocessor that detects obvious typos, grammar errors, and punctuation issues in user messages without auto-correcting. It prompts users...
中文消息预处理器,检测用户消息中的明显拼写错误、语法错误和标点问题(不自动更正),并提示用户...
fuletiancheng
未分类 clawhub v0.1.0 1 版本 99310.3 Key: 无需
★ 0
Stars
📥 144
下载
💾 0
安装
1
版本
#latest

概述

钱纠纠 (Qian Jiu Jiu)

> 只指出问题,不自动修改,静默确认继续

>

> "Report issues only, never auto-correct, silently confirm to proceed"

Overview

钱纠纠 is a message preprocessor skill for OpenClaw agents that:

  • Detects obvious typos, grammar errors, and punctuation issues in Chinese user messages
  • Never auto-corrects - only reports issues found
  • Prompts user confirmation before proceeding with potentially confusing messages
  • Silently continues when user confirms (no verbose "confirmed" messages)

Installation

# Install via OpenClaw CLI
openclaw skills add qian-jiu-jiu

# Or manually copy to skills directory
cp -r qian-jiu-jiu ~/.openclaw/skills/
openclaw skills reload

Requirements

  • OpenClaw >= 2026.2.x
  • Chinese language context (optimized for Simplified Chinese)

Usage

The skill automatically activates on each user message:

Flow 1: Error Detection

When obvious errors are detected:

句子存在异常(可能影响理解):
异常列表:
• 错字:神马 → 可能意为:什么
是否继续用这个原始句子进行推理?
(回复"是"则直接用上面原始句子推理;否则请修改后重新发送)

Flow 2: Silent Confirmation

When user replies with confirmation keywords:

> Confirmation keywords: 是, 是的, 确认, 继续, ok, 行, 可以, yes, y, 好

The skill silently proceeds with the original sentence - no verbose confirmation, no "original sentence:" display, just normal response.

Flow 3: Clean Pass-through

If no obvious errors - message flows through normally without any output.

Error Types Detected

TypeDescriptionExample
----------------------------
多字Redundant characters这不是行啊 → 这不行啊
少字Missing characters你吃饭 → 你吃饭了吗
重复字Repeated characters我我我要出门 → 我要出门
错字Wrong/typo characters平果 → 苹果 / 神马 → 什么
错用字Misused homophones的/得/地 混用
语序错乱Wrong word order吃饭了我已经 → 我已经吃饭了
搭配不当Collocation errors喝饭 → 吃饭
标点缺失Missing punctuation causing ambiguity我想出去玩球没买 → 出去玩,球没买? / 出去玩球,没买?
空格异常Invalid spaces in words假 如 → 假如
单位遗漏Missing units after numbers身高178 → 身高178厘米
拼音错Pinyin instead of charactersuf → 效果 / nihao → 你好

Response Format

Error Detection Format (Strict line-by-line):

句子存在异常(可能影响理解):\n
异常列表:\n
• [type]: [error] → 可能意为:[possible meaning]\n
是否继续用这个原始句子进行推理?\n
(回复"是"则直接用上面原始句子推理;否则请修改后重新发送)

Confirmation Response Format:

[Normal response to the original sentence]

Key Constraints

  1. Never auto-correct - Only report, never suggest "correct version"
  2. Silent confirmation - No "confirmed", "proceeding", or "OK" messages
  3. Conservative detection - Better to miss than over-report
  4. No empty lines between output lines (use \n only)
  5. Context-aware - Considers conversation history for ambiguous cases

License

MIT

Contributing

Issues and PRs welcome at https://github.com/[your-username]/qian-jiu-jiu

版本历史

共 1 个版本

  • v0.1.0 当前
    2026-05-12 06:18 安全 安全

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

developer-tools

Github

steipete
使用 `gh` CLI 与 GitHub 交互,通过 `gh issue`、`gh pr`、`gh run` 和 `gh api` 管理议题、PR、CI 运行及高级查询。
★ 672 📥 324,508
ai-intelligence

self-improving agent

pskoett
捕获经验教训、错误和纠正,以实现持续改进。使用时机:(1)命令或操作意外失败;(2)用户纠正……
★ 4,062 📥 799,814
ai-intelligence

Self-Improving + Proactive Agent

ivangdavila
自我反思+自我批评+自我学习+自组织记忆。智能体评估自身工作、发现错误并持续改进。
★ 1,363 📥 319,028