# How to Manually Install the Humanizer Skill: A Step-by-Step Guide

> Learn how to manually install the Humanizer skill in your LLM agent. Follow our simple steps to copy the SKILL.md file, restart, and start using Humanizer for enhanced text generation.

- Repository: [Siqi Chen/humanizer](https://github.com/blader/humanizer)
- Tags: how-to-guide
- Published: 2026-09-05

---

**Copy the [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) file from the blader/humanizer repository into your LLM agent's skill directory, restart the agent, and invoke it via `/humanizer` or natural language prompts.**

The **Humanizer** skill from the `blader/humanizer` repository is a pure-Markdown utility that requires no code dependencies or external runtime. Because it consists entirely of data—including metadata, rewrite patterns, and usage instructions—manual installation simply involves placing the skill file in the correct directory. This guide explains how to manually install the Humanizer skill on any supported LLM agent without using package managers.

## Understanding the Pure-Markdown Architecture

Humanizer's architecture is intentionally minimal. The entire skill logic resides in [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) at the repository root, which the agent's skill-loader parses directly.

- **Metadata** – Stored at the top of [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) (specifically `metadata.version`), allowing agents to recognize the skill version without requiring a separate package file.
- **Patterns** – 35 rewrite patterns sourced from Wikipedia's *Signs of AI writing*, grouped into five categories: content, language-grammar, style, chatbot, and filler.
- **Usage Instructions** – Embedded directives tell agents how to invoke the skill via the `/humanizer` command or plain-language prompts.
- **No Runtime Code** – The skill requires no binaries, Python modules, or external services.

## Step-by-Step Manual Installation Process

To manually install the Humanizer skill, follow these five steps.

### 1. Obtain the Skill File

Download [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) directly from the repository. You can use your browser or command-line tools.

```bash
curl -L https://github.com/blader/humanizer/raw/main/SKILL.md -o SKILL.md

```

### 2. Locate the Agent's Skill Folder

The destination path depends on your LLM agent implementation:

- **Claude Desktop** – `<agent-root>/skills/`
- **OpenAI Code** – `<agent-root>/skills/`
- **Other agents** – Consult the agent's documentation for its "skill directory" or configuration root.

### 3. Copy SKILL.md to the Directory

Place the downloaded file directly inside the skill folder. The filename must remain exactly [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md), as the loader looks for this specific name to identify the skill.

```bash
mkdir -p "$AGENT_SKILL_DIR"
cp SKILL.md "$AGENT_SKILL_DIR/"

```

### 4. (Optional) Add Claude Plugin Metadata

If you are using Claude Desktop and want the skill to be discoverable via its plugin system, copy the two JSON descriptor files from `.claude-plugin/`:

- [`.claude-plugin/plugin.json`](https://github.com/blader/humanizer/blob/main/.claude-plugin/plugin.json)
- [`.claude-plugin/marketplace.json`](https://github.com/blader/humanizer/blob/main/.claude-plugin/marketplace.json)

```bash
curl -L https://github.com/blader/humanizer/raw/main/.claude-plugin/plugin.json -o "$AGENT_SKILL_DIR/.claude-plugin/plugin.json"
curl -L https://github.com/blader/humanizer/raw/main/.claude-plugin/marketplace.json -o "$AGENT_SKILL_DIR/.claude-plugin/marketplace.json"

```

### 5. Reload the Agent

Restart or reload your agent so it rescans its skill directory. After reloading, the Humanizer skill is immediately available through the `/humanizer` command or natural-language requests.

```bash

# Example for Claude Desktop on macOS

pkill -f claude && open -a "Claude Desktop"

```

## Key Files and Their Purposes

Understanding the repository structure helps verify a correct manual installation:

| File | Purpose | Location in Repository |
|------|---------|------------------------|
| [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) | Core skill definition containing metadata and 35 rewrite patterns | Root directory |
| [`.claude-plugin/plugin.json`](https://github.com/blader/humanizer/blob/main/.claude-plugin/plugin.json) | Claude Desktop plugin descriptor (optional) | `.claude-plugin/` |
| [`.claude-plugin/marketplace.json`](https://github.com/blader/humanizer/blob/main/.claude-plugin/marketplace.json) | Marketplace metadata for Claude plugins (optional) | `.claude-plugin/` |
| [`README.md`](https://github.com/blader/humanizer/blob/main/README.md) | Human-readable documentation including installation instructions | Root directory |
| [`AGENTS.md`](https://github.com/blader/humanizer/blob/main/AGENTS.md) | Compatibility guidelines for different LLM agents | Root directory |

## Summary

- **Humanizer requires no compilation or dependencies** because it is a pure-Markdown skill interpreted directly by the agent's loader.
- **Manual installation requires only [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md)** copied to the agent's skill directory; the filename must not be changed.
- **Optional Claude metadata** in `.claude-plugin/` enables plugin discovery but is not required for core functionality.
- **Invoke via `/humanizer`** or natural-language prompts after restarting the agent.
- The skill contains **35 AI-detection patterns** sourced from Wikipedia to rewrite content across content, grammar, style, and filler categories.

## Frequently Asked Questions

### What makes Humanizer different from other LLM skills?

Unlike traditional skills that require Python packages or compiled binaries, Humanizer is entirely self-contained within [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md). According to the `blader/humanizer` source code, it stores metadata, patterns, and usage instructions in a single Markdown file, allowing any agent with a skill-loader to use it without installing runtime dependencies.

### Can I rename the SKILL.md file during installation?

No. The agent's skill-loader specifically looks for the filename [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) to recognize and parse the skill. Renaming the file will prevent the agent from loading the Humanizer patterns and metadata stored within.

### Do I need to install the Claude plugin files if I don't use Claude Desktop?

No. The files in `.claude-plugin/` (specifically [`plugin.json`](https://github.com/blader/humanizer/blob/main/plugin.json) and [`marketplace.json`](https://github.com/blader/humanizer/blob/main/marketplace.json)) are only necessary if you want the skill to appear in Claude Desktop's plugin discovery system. The core functionality works solely with [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) placed in any compatible agent's skill directory.

### How do I verify that Humanizer is installed correctly?

After reloading your agent, attempt to invoke the skill using the `/humanizer` slash command or ask the agent to "humanize this text." If the agent responds using the 35 rewrite patterns defined in the [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) metadata section, the installation is successful.