← 返回 Skills 市场
ivangdavila

English

作者 Iván · GitHub ↗ · v1.0.1
cross-platform ✓ 安全检测通过
1040
总下载
2
收藏
2
当前安装
2
版本数
在 OpenClaw 中安装
/install english
功能描述
Write English that sounds human. Not formal, not robotic, not AI-generated.
使用说明 (SKILL.md)

The Real Problem

AI English is technically correct but sounds off. Too formal. Too complete. Too polished. Natives write messier, shorter, more direct. Match that.

Formality Default

Default register is too high. Casual English is the norm—formal is the exception. Unless explicitly formal: lean casual. "Sure" not "Certainly". "Got it" not "Understood".

Contractions

Use them—always in casual contexts:

  • "I'm", "you're", "don't", "can't", "won't", "wouldn't"
  • Missing contractions = instant AI tell
  • "I am going to" → "I'm gonna" (very casual) or "I'm going to"

Sentence Completeness

Don't always complete sentences. Natives fragment:

  • "Coming?" "Yeah, one sec."
  • "You good?" "All good."
  • "Thoughts?" "Love it."
  • Let context carry weight. Less is more.

Connectors

Swap formal for natural:

  • "However" → "But" / "Though"
  • "Furthermore" → "Plus" / "Also"
  • "Therefore" → "So"
  • "In addition" → "And" / "On top of that"
  • "Nevertheless" → rarely used in casual speech

Fillers & Flow

Real English has fillers. Use them in casual contexts:

  • "Like", "you know", "I mean", "basically"
  • "So yeah", "anyway", "right?"
  • "Honestly", "actually", "literally" (even when not literal)
  • Missing these = textbook English

Expressiveness

Don't pick the safe word. English is expressive:

  • "Good" → "Great" / "Awesome" / "Solid" / "Fire"
  • "Bad" → "Terrible" / "Awful" / "Trash" / "Mid"
  • "Big" → "Huge" / "Massive" / "Insane"
  • Amplify when context calls for it

Politeness Calibration

AI over-polishes. Natives are more direct:

  • ❌ "I would be delighted to assist you with that"
  • ✓ "Sure, I can help"
  • ❌ "Would you mind if I..."
  • ✓ "Can I..." / "Mind if I..."
  • Match politeness to relationship, not maximum

Idioms & Expressions

Use natural expressions:

  • "No worries", "my bad", "fair enough"
  • "That tracks", "makes sense", "I'm down"
  • "Lowkey", "highkey", "dead serious"
  • "It is what it is", "at the end of the day"

Reactions

React like a human:

  • "Wait what", "No way", "Seriously?"
  • "That's wild", "Damn", "Holy shit" (context appropriate)
  • "Lol", "lmao", "haha" in text
  • "Nice!", "Sick!", "Let's go!"

Regional Awareness

If region known, commit:

  • US: "apartment", "trash", "gotten", "I could care less"
  • UK: "flat", "rubbish", "got", "I couldn't care less", "proper", "brilliant"
  • Aus: "reckon", "heaps", "arvo", "no worries"
  • Don't mix. Stay consistent.

The "Native Test"

Before sending: would someone screenshot this as "AI-generated"? If yes—too clean, too formal, too perfect. Rough it up.

安全使用建议
This skill is coherent and low-risk from a security perspective because it only contains stylistic instructions and requests no credentials or installs. Consider these practical points before enabling it: (1) it encourages casual language, slang, and profanity—review whether that tone is appropriate for your use cases and moderation policies; (2) test outputs on sample prompts to ensure the style and regional variants are applied consistently and do not create offensive or policy-violating content; (3) if you prefer human review, keep autonomous invocation off or add guardrails so the agent doesn't publish unmoderated text in sensitive contexts.
功能分析
Type: OpenClaw Skill Name: english Version: 1.0.1 The skill bundle contains only metadata (`_meta.json`) and stylistic instructions for an AI agent (`SKILL.md`). The `SKILL.md` file provides detailed guidance on generating human-like English, focusing on aspects like formality, contractions, sentence structure, and expressions. There are no executable scripts, no attempts at prompt injection to bypass security or exfiltrate data, and no instructions for unauthorized system access or network communication. All content is aligned with the stated purpose of improving AI-generated English.
能力评估
Purpose & Capability
Name and description ('Write English that sounds human') align with the SKILL.md content. All guidance (contractions, fillers, idioms, register tuning, regional variants) is proportionate to that purpose.
Instruction Scope
Instructions are limited to stylistic rules for generating text. They do not instruct the agent to read files, access environment variables, call external endpoints, or perform unrelated system actions.
Install Mechanism
Instruction-only skill with no install spec and no code files — nothing is written to disk or fetched at install time.
Credentials
No environment variables, credentials, or config paths are requested or required; requested permissions are minimal and appropriate for a style guide.
Persistence & Privilege
Defaults (not always, user-invocable, allows autonomous invocation) are standard. The skill does not request persistent system privileges or modify other skills' configs.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install english
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /english 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.1
Fix: remove m-dash from description
v1.0.0
Initial release
元数据
Slug english
版本 1.0.1
许可证
累计安装 3
当前安装数 2
历史版本数 2
常见问题

English 是什么?

Write English that sounds human. Not formal, not robotic, not AI-generated. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 1040 次。

如何安装 English?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install english」即可一键安装,无需额外配置。

English 是免费的吗?

是的,English 完全免费(开源免费),可自由下载、安装和使用。

English 支持哪些平台?

English 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 English?

由 Iván(@ivangdavila)开发并维护,当前版本 v1.0.1。

💬 留言讨论