# How to Configure the OpenAI Agent for Humanizer: A Complete Setup Guide

> Learn how to configure the OpenAI agent for Humanizer with our complete setup guide. Install the skill, reload the agent, and start humanizing text effortlessly.

- Repository: [Siqi Chen/humanizer](https://github.com/blader/humanizer)
- Tags: getting-started
- Published: 2026-09-11

---

**To configure the OpenAI agent for Humanizer, install the skill using the Skills CLI with the `--agent openai` flag, reload the agent to register the `/humanizer` command, and invoke it via chat or the default prompt defined in [`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml).**

The blader/humanizer repository provides a Markdown-based skill that transforms AI-generated text into human-like prose while preserving factual accuracy. To configure the OpenAI agent for Humanizer, you must register the skill through the [`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml) configuration file and install it using the standard Skills CLI workflow. This setup enables the agent to recognize the `/humanizer` command and process input through the 25 rewrite patterns defined in [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md).

## Understanding the Humanizer Architecture

Humanizer operates as a portable skill that any LLM agent supporting the standard *skills* format can consume. According to the blader/humanizer source code, the architecture cleanly separates core logic from agent-specific registration.

- **[`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md)**: Contains the core prompt, pattern list, and rewrite logic that Humanizer applies to input text. This file defines 25 patterns grouped by strength and frequency to humanize AI-generated content.
- **[`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml)**: Provides the metadata required to register Humanizer with OpenAI-compatible agents, including the display name, description, and default prompt instructions.
- **[`scripts/validate-package.py`](https://github.com/blader/humanizer/blob/main/scripts/validate-package.py)**: Validates skill file consistency before publishing to ensure [`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml) correctly references available resources.

When configured, the OpenAI agent forwards user text to Humanizer's prompt engine, which applies the transformation rules and returns the rewritten content.

## Step-by-Step Configuration Guide

### Install the Skill via CLI

Use the Skills CLI to add Humanizer to your OpenAI agent. The `--agent` flag specifies which agent should receive the skill.

For global installation:

```bash
npx skills add blader/humanizer --global --agent openai

```

For project-specific installation:

```bash
npx skills add blader/humanizer --agent openai

```

### Reload the Agent

After installation, the OpenAI agent must reload its skill list to recognize the new `/humanizer` command. The specific reload mechanism depends on your runtime environment.

For a local server watching the skill directory:

```bash
pkill -HUP openai-agent

```

This sends a SIGHUP signal to force the agent to refresh its available skills and load the configuration from [`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml).

### Verify the Configuration

Confirm that [`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml) is properly loaded by checking for the skill in your agent's command list. This file defines the default prompt that triggers when users invoke the skill, ensuring the agent knows how to route requests to [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md).

## Invoking the Humanizer Skill

Once configured, interact with Humanizer through chat commands or programmatic API calls.

### Chat-Based Invocation

Use the slash command defined in your agent configuration:

```text
/humanizer
[Paste the AI-generated text you want to humanize]

```

Alternatively, use the default prompt specified in [`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml):

```text
Use $humanizer to rewrite this text in my voice without changing its facts.

```

### Programmatic API Usage

Call the skill via the OpenAI API by including the command in your message payload:

```python
import openai

response = openai.ChatCompletion.create(
    model="gpt-4",
    messages=[
        {"role": "system", "content": "You have the Humanizer skill installed."},
        {"role": "user", "content": "/humanizer\nI think AI text sounds very formal and generic."}
    ]
)

print(response["choices"][0]["message"]["content"])

```

## Key Configuration Files Reference

| File | Purpose | Location |
|------|---------|----------|
| **[`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml)** | Registers Humanizer with OpenAI agents; defines display name and default prompt | [`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml) |
| **[`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md)** | Core prompt containing 25 rewrite patterns and transformation logic | [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) |
| **[`README.md`](https://github.com/blader/humanizer/blob/main/README.md)** | User-facing documentation for installation and usage | [`README.md`](https://github.com/blader/humanizer/blob/main/README.md) |
| **[`scripts/validate-package.py`](https://github.com/blader/humanizer/blob/main/scripts/validate-package.py)** | Pre-publication validation for skill consistency | [`scripts/validate-package.py`](https://github.com/blader/humanizer/blob/main/scripts/validate-package.py) |

## Summary

- Configure the OpenAI agent for Humanizer by executing `npx skills add blader/humanizer --agent openai` and reloading the agent process.
- The [`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml) file handles agent registration, while [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md) contains the 25 pattern-based rewrite logic.
- Invoke the skill using the `/humanizer` command in chat or programmatically through the OpenAI API.
- Run [`scripts/validate-package.py`](https://github.com/blader/humanizer/blob/main/scripts/validate-package.py) to ensure configuration consistency before deploying updates.

## Frequently Asked Questions

### What file configures the OpenAI agent for Humanizer?

The [`agents/openai.yaml`](https://github.com/blader/humanizer/blob/main/agents/openai.yaml) file configures the OpenAI agent for Humanizer by defining the skill's display name, description, and default prompt. This YAML file registers the skill with the agent and specifies how the `/humanizer` command should be handled and routed to the core logic in [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md).

### How do I install Humanizer for a specific project only?

Install Humanizer for a specific project by running `npx skills add blader/humanizer --agent openai` without the `--global` flag. This restricts the skill to the current project directory rather than installing it system-wide, allowing different projects to maintain separate skill configurations.

### Can I use Humanizer programmatically with the OpenAI API?

Yes, you can invoke Humanizer programmatically by including the `/humanizer` command in your message content when calling `openai.ChatCompletion.create()`. Ensure your system message indicates that the Humanizer skill is installed so the agent recognizes the command and processes the text through the patterns defined in [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md).

### What patterns does Humanizer use to rewrite text?

Humanizer applies 25 distinct patterns defined in [`SKILL.md`](https://github.com/blader/humanizer/blob/main/SKILL.md), grouped by strength and frequency, to transform AI-generated text into human-like prose. These patterns rewrite the text structure, tone, and phrasing while strictly preserving factual content and original meaning.