# Where to Find the Codex Behavior Guide for AI Berkshire

> Locate the Codex behavior guide for AI Berkshire within the AGENTS.md file in xbtlin/ai-berkshire. Discover Claude Code compatibility and Codex skill behavior rules.

- Repository: [Xbt Lin/ai-berkshire](https://github.com/xbtlin/ai-berkshire)
- Tags: how-to-guide
- Published: 2026-07-10

---

**The Codex behavior guide for AI Berkshire is documented in the repository's [`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md) file, which defines compatibility rules between Claude Code and Codex and outlines how Codex-generated skills should behave.**

The AI Berkshire project maintains a structured approach to AI agent compatibility within the `xbtlin/ai-berkshire` repository. If you are looking for the Codex behavior guide for AI Berkshire, the canonical documentation resides in a specific markdown file at the repository root. This guide ensures that skills work consistently across both Claude Code and Codex environments.

## Primary Location: AGENTS.md

The definitive Codex behavior guide for AI Berkshire lives in **[`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md)** at the repository root ([view on GitHub](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md)). According to the `xbtlin/ai-berkshire` source code, this file establishes the compatibility framework between Claude Code and Codex, documents how Codex-only skills are generated, and specifies the expected behavior for Codex-generated implementations.

## Codex Adapter Notes in Skill Definitions

Individual skill implementations reference the behavior guide through "Codex adapter note" sections. These notes appear in each `codex-skills/*/SKILL.md` file and link the specific implementation back to the canonical workflow defined in [`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md).

Example from [`codex-skills/wechat-article/SKILL.md`](https://github.com/xbtlin/ai-berkshire/blob/main/codex-skills/wechat-article/SKILL.md):

```markdown

## Codex adapter note

This skill is generated from `skills/wechat-article.md` so Claude Code and Codex users share one canonical workflow.

```

## Prompt Files and Skill Invocation

The `codex-prompts/` directory contains files that invoke these skills and rely on the behavior definitions in [`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md). For example, [`codex-prompts/wechat-article.md`](https://github.com/xbtlin/ai-berkshire/blob/main/codex-prompts/wechat-article.md) references the installed skill and assumes the underlying behavior standards defined in the guide.

Example prompt reference:

```markdown
Use the installed AI Berkshire Codex skill `wechat-article` for this request.

```

## Key Files Supporting the Codex Behavior Guide

Several file types comprise the complete behavior guide ecosystem:

- **[`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md)** – The primary Codex behavior guide containing compatibility rules and skill generation notes.
- **`codex-skills/*/SKILL.md`** – Individual skill definitions that include Codex adapter notes referencing the guide.
- **`codex-prompts/*.md`** – Prompt files that invoke Codex skills and depend on the behavior standards outlined in [`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md).

## Summary

- The **Codex behavior guide for AI Berkshire** is located in **[`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md)** at the repository root.
- Individual skills in `codex-skills/*/SKILL.md` contain adapter notes linking to the canonical guide.
- Prompt files in `codex-prompts/` rely on the behavior standards defined in [`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md).
- The guide ensures compatibility between Claude Code and Codex implementations.

## Frequently Asked Questions

### What is the main file for the Codex behavior guide in AI Berkshire?

The primary file is **[`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md)** in the repository root. This markdown document contains the complete compatibility rules between Claude Code and Codex, explains how Codex-only skills are generated, and outlines expected behaviors for AI-generated skills.

### Where can I find examples of Codex skills that follow the behavior guide?

Look in the **`codex-skills/`** directory. Each subdirectory contains a [`SKILL.md`](https://github.com/xbtlin/ai-berkshire/blob/main/SKILL.md) file with a "Codex adapter note" section that references the behavior guide and explains how the skill was generated from the canonical source.

### How do prompt files reference the Codex behavior guide?

Files in **`codex-prompts/`** (such as [`wechat-article.md`](https://github.com/xbtlin/ai-berkshire/blob/main/wechat-article.md)) invoke skills by name and assume the behavior standards defined in [`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md). They do not duplicate the behavior rules but rely on the guide to ensure consistent execution across different AI agents.

### Is the Codex behavior guide specific to Codex only, or does it cover Claude Code too?

The guide covers both. According to the source code in [`AGENTS.md`](https://github.com/xbtlin/ai-berkshire/blob/main/AGENTS.md), the documentation defines **compatibility rules between Claude Code and Codex**, ensuring that skills generated for Codex maintain consistency with the canonical workflows used by Claude Code.