← 返回 Skills 市场
Ai Writing Humanizer
作者
Host Hobbit
· GitHub ↗
· v0.1.0
978
总下载
0
收藏
6
当前安装
1
版本数
在 OpenClaw 中安装
/install ai-writing-humanizer
功能描述
Automatically strip AI writing patterns and stock phrases from user-facing prose before sending. Use when preparing longer texts to ensure a natural, human t...
使用说明 (SKILL.md)
AI Writing Humanizer
Provide a cleaned, human-sounding version of input text by removing or rewriting common AI-generated writing patterns. Load this skill automatically for any user-facing prose longer than ~150 words.
Patterns Covered
- Overuse of hedging phrases: "At the end of the day", "It's worth noting", "It is important to remember"
- Stock transitions and signposts: "First", "Secondly", "Finally"
- Structural markers: em dashes (—), parentheses for side notes, bullet-like serial lists
- Predictable rule-of-three constructs: "X, Y, and Z" used as checklist language
- Passive voice sections flagged by "was", "were", "has been"
- Performed authenticity cues: "I hope this helps", "Let me know if you have any questions"
Usage
- Pass input text to
scripts/humanize.py --input text.txt --output cleaned.txt. - For integrations, import
humanize_text()from the script and apply before sending.
Bundled Resources
- scripts/humanize.py — Python script implementing regex-based pattern removal and light rewriting
- references/signs-of-ai-writing.md — List of AI writing signs based on Wikipedia
- tests/test_humanizer.py — Regression tests to detect patterns
安全使用建议
This skill appears to do what it says and runs locally with no network or secret access. Before installing, review whether you want it to be auto-invoked on longer user-facing text (it may process content you didn't intend it to). Also inspect and test the regex behavior on representative texts — the script removes simple tokens like 'was' and 'were' which can change meaning or break grammar. If you handle sensitive or legal text, require explicit invocation and run tests to ensure the transformations are acceptable.
功能分析
Type: OpenClaw Skill
Name: ai-writing-humanizer
Version: 0.1.0
The OpenClaw skill bundle 'ai-writing-humanizer' is benign. The `SKILL.md` provides clear, non-malicious instructions for the AI agent to apply text processing. The `scripts/humanize.py` script uses standard Python libraries (`re`, `argparse`) to perform regex-based text cleaning, reading from an input file and writing to an output file, without any network calls, access to sensitive system resources, or use of dangerous functions like `eval` or `os.system`. There is no evidence of data exfiltration, malicious execution, persistence mechanisms, or prompt injection attempts against the agent. A minor functional discrepancy exists between the `SKILL.md` description and the `humanize.py` implementation regarding em dash replacement, but this is not a security concern.
能力评估
Purpose & Capability
Name/description match the delivered artifacts: a small Python script plus reference and tests implement regex-based humanization. No unrelated binaries, env vars, or install steps are requested, so the capability is proportional to the stated purpose.
Instruction Scope
SKILL.md instructs local use (CLI or import) and to 'load automatically for any user-facing prose longer than ~150 words.' The runtime instructions do not read external files or env vars beyond the explicit input file, nor do they send data outside. However, the automatic-loading instruction and the script's aggressive pattern removals (e.g., removing all occurrences of 'was', 'were', 'has been') could unintentionally change meaning or act on sensitive content without explicit consent — consider limiting auto-invoke or requiring explicit confirmation for sensitive texts.
Install Mechanism
No install spec (instruction-only plus bundled scripts). No downloads or external installers — code is provided in the bundle and runs locally, which is low risk from an install perspective.
Credentials
No environment variables, credentials, or config paths are requested. The skill does not attempt to access unrelated secrets or system configuration.
Persistence & Privilege
always is false and the skill doesn't modify other skills or system-wide settings. It can be invoked autonomously by the agent (platform default), but no elevated persistence or cross-skill config changes are requested.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install ai-writing-humanizer - 安装完成后，直接呼叫该 Skill 的名称或使用
/ai-writing-humanizer触发 - 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
v0.1.0
ai-writing-humanizer 0.1.0
- Initial release.
- Automatically rewrites or removes common AI-generated phrases and patterns from user-facing prose.
- Handles hedging, stock transitions, overused list structures, passive voice, and authenticity cues.
- Provides a Python script (humanize.py) for text cleaning and integration.
- Includes documentation and tests for coverage of typical AI "tells."
元数据
常见问题
Ai Writing Humanizer 是什么？
Automatically strip AI writing patterns and stock phrases from user-facing prose before sending. Use when preparing longer texts to ensure a natural, human t... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 978 次。
如何安装 Ai Writing Humanizer？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install ai-writing-humanizer」即可一键安装，无需额外配置。
Ai Writing Humanizer 是免费的吗？
是的，Ai Writing Humanizer 完全免费（开源免费），可自由下载、安装和使用。
Ai Writing Humanizer 支持哪些平台？
Ai Writing Humanizer 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 Ai Writing Humanizer？
由 Host Hobbit（@hosthobbit）开发并维护，当前版本 v0.1.0。
推荐 Skills