alisonaquinas/llm-shared-skills · GitHub

codex-docs

1회 설치

어떤 스킬인가요?

Use this skill to look up official OpenAI Codex documentation at developers.openai.com/codex. Trigger when the user asks about Codex features, Codex concepts, AGENTS.md, Codex MCP, Codex SDK, Codex GitHub Action, Codex automation, or any OpenAI Codex reference. For CLI-specific help, use codex-cli-docs.

설치 방법

프로젝트 폴더에서 아래 명령어를 실행하고, 설치할 에이전트를 선택하세요.

npx skills add alisonaquinas/llm-shared-skills --skill codex-docs

설치 명령을 직접 실행해야 적용됩니다. 지원 에이전트와 필요한 권한·라이선스는 제작자의 안내를 확인하세요.

스킬 원문 파일

`agents/claude.yaml`

interface:
  display_name: "OpenAI Codex Docs"
  short_description: "Look up official OpenAI Codex agent documentation"
  default_prompt: "Use $codex-docs to find the documentation for [configuration/concepts/AGENTS.md/automation/integrations/etc.]."

`agents/openai.yaml`

interface:
  display_name: "OpenAI Codex Docs"
  short_description: "Look up official OpenAI Codex agent documentation"
  default_prompt: "Use $codex-docs to find the documentation for [configuration/concepts/AGENTS.md/automation/integrations/etc.]."

`references/automation.md`

# OpenAI Codex — Automation

## Key Pages

| Topic | URL |
| --- | --- |
| Non-interactive mode | <https://developers.openai.com/codex/noninteractive> |
| SDK | <https://developers.openai.com/codex/sdk> |
| App server | <https://developers.openai.com/codex/app-server> |
| MCP server subcommand | <https://developers.openai.com/codex/cli/reference> |
| GitHub Action | <https://developers.openai.com/codex/github-action> |

## GitHub Action

Run Codex tasks automatically in CI/CD:

```yaml
# .github/workflows/codex.yml
name: Codex
on:
  issues:
    types: [labeled]
jobs:
  codex:
    runs-on: ubuntu-latest
    steps:
      - uses: openai/codex-action@v1
        with:
          task: "Fix the issue described in ${{ github.event.issue.body }}"
          openai-api-key: ${{ secrets.OPENAI_API_KEY }}
```

Full reference: <https://developers.openai.com/codex/github-action>

## Non-interactive Mode

Run Codex headlessly from a script or CI pipeline:

```bash
codex exec "write tests for src/auth.py"
```

## SDK

The Codex SDK allows programmatic task submission and result retrieval:
<https://developers.openai.com/codex/sdk>

Use it to build custom workflows, dashboards, or integrate Codex into existing tooling.

`references/concepts.md`

# OpenAI Codex — Concepts

## Key Pages

| Topic | URL |
| --- | --- |
| Prompting | <https://developers.openai.com/codex/concepts/prompting> |
| Customization | <https://developers.openai.com/codex/concepts/customization> |
| Sandboxing | <https://developers.openai.com/codex/concepts/sandboxing> |
| Multi-agents | <https://developers.openai.com/codex/concepts/multi-agents> |
| Workflows | <https://developers.openai.com/codex/concepts/workflows> |
| Models | <https://developers.openai.com/codex/concepts/models> |
| Cyber safety | <https://developers.openai.com/codex/concepts/cyber-safety> |

## Sandboxing

Codex runs tasks in isolated sandbox environments with internet access restricted to
allowlisted domains. This ensures safety during automated code execution. Each task
gets a fresh environment cloned from your repository.

## Multi-agent Workflows

Codex supports spawning multiple agent instances to work on tasks in parallel.
Useful for large refactors, running tests across multiple configurations, or
parallelizing independent feature work.

## Prompting Tips

- Be specific about acceptance criteria ("write tests that cover edge cases X, Y, Z")
- Reference file paths, function names, or existing patterns
- State constraints explicitly ("do not change the public API")
- For complex tasks, break into subtasks or use AGENTS.md to set context

## Customization

Codex behavior can be customized via:

- **AGENTS.md** — project-level instructions (see configuration.md)
- **Rules** — reusable instruction sets
- **Skills** — packaged workflows (see configuration.md)
- **MCP** — connect external tools (see configuration.md)

`references/configuration.md`

# OpenAI Codex — Configuration

## Key Pages

| Topic | URL |
| --- | --- |
| Config basics | <https://developers.openai.com/codex/config-basic> |
| Advanced config | <https://developers.openai.com/codex/config-advanced> |
| Config reference | <https://developers.openai.com/codex/config-reference> |
| Config sample | <https://developers.openai.com/codex/config-sample> |
| AGENTS.md | <https://developers.openai.com/codex/guides/agents-md> |
| Rules | <https://developers.openai.com/codex/rules> |
| Hooks | <https://developers.openai.com/codex/hooks> |
| MCP | <https://developers.openai.com/codex/mcp> |
| Skills | <https://developers.openai.com/codex/skills> |
| Subagents | <https://developers.openai.com/codex/subagents> |

## AGENTS.md

`AGENTS.md` is a markdown file at the root of your repo that Codex reads at the
start of every session. Use it to provide:

- Project architecture and conventions
- Commands to run tests, lint, build
- Off-limits files or patterns
- Style guidelines and naming conventions

Codex also reads `AGENTS.md` files in subdirectories for folder-specific context.

## MCP (Model Context Protocol)

Codex supports MCP servers for connecting external tools and data sources:

- Jira, Linear, Notion, GitHub, Slack
- Custom MCP servers you build

Configure MCP servers in `~/.codex/config.toml` or with `codex mcp`.

## Skills

Skills are packaged, reusable workflows that Codex can invoke. Configure skills
and plugins so repeatable task templates stay available across sessions.

## Config Files

The CLI reads `~/.codex/config.toml`; project instructions live in `AGENTS.md`.
Important controls include model selection, approval policy, sandbox mode, MCP
servers, hooks, plugins, and web search mode.

`references/interfaces.md`

# OpenAI Codex — Interfaces & Integrations

## Surfaces

| Surface | URL | Best for |
| --- | --- | --- |
| Web app | <https://developers.openai.com/codex/app> | Browser-based task management |
| IDE Extension | <https://developers.openai.com/codex/ide> | Inline suggestions in VS Code / JetBrains |
| CLI | <https://developers.openai.com/codex/cli> | Terminal-based workflows (see `codex-cli-docs`) |
| Web (cloud) | <https://developers.openai.com/codex/cloud> | Cloud-hosted sessions |
| ChatGPT | <https://chatgpt.com/codex> | Chat-driven coding tasks |

## Integrations

| Integration | URL |
| --- | --- |
| GitHub | <https://developers.openai.com/codex/integrations/github> |
| Slack | <https://developers.openai.com/codex/integrations/slack> |
| Linear | <https://developers.openai.com/codex/integrations/linear> |

## GitHub Integration

Codex can:

- Read issues and PRs
- Create branches and open pull requests
- Run as a GitHub Action (see automation.md)
- Respond to `@codex` mentions in PR comments

## IDE Extension

Install from the VS Code Marketplace or JetBrains Marketplace by searching "Codex".
Provides inline completions, chat, and task delegation directly in the editor.
Full docs: <https://developers.openai.com/codex/ide>

## Codex Security

<https://developers.openai.com/codex/security>

Covers: sandbox isolation, allowlisted domains, code execution safety, data handling.

`references/overview.md`

# OpenAI Codex — Overview

## Key Pages

| Page | URL |
| --- | --- |
| Codex home | <https://developers.openai.com/codex> |
| Quickstart | <https://developers.openai.com/codex/quickstart> |
| Explore (demos) | <https://developers.openai.com/codex/explore> |
| Pricing | <https://developers.openai.com/codex/pricing> |
| Codex in ChatGPT | <https://help.openai.com/en/articles/11369540-codex-in-chatgpt> |
| Web access | <https://chatgpt.com/codex> |

## What is Codex?

Codex is OpenAI's coding agent for software development. Core capabilities:

- **Code generation** — describe what to build, Codex generates it
- **Code comprehension** — understand complex or legacy codebases
- **Code review** — find bugs, logic errors, unhandled edge cases
- **Debugging** — trace failures, diagnose root causes
- **Task automation** — refactoring, testing, migrations, setup tasks

## Plans

Codex is available to: ChatGPT Plus, Pro, Business, Edu, and Enterprise subscribers.
API/SDK access is available for programmatic use.

## Getting Started

1. Go to <https://chatgpt.com/codex> or install the IDE extension
2. Connect your GitHub repository
3. Describe a task — Codex runs in a sandboxed environment and opens a PR

`SKILL.md`

---
name: codex-docs
description: "Use this skill to look up official OpenAI Codex documentation at developers.openai.com/codex. Trigger when the user asks about Codex features, Codex concepts, AGENTS.md, Codex MCP, Codex SDK, Codex GitHub Action, Codex automation, or any OpenAI Codex reference. For CLI-specific help, use codex-cli-docs."
---

# OpenAI Coding Agent Docs

Reference skill for navigating official OpenAI coding agent documentation.
Root URL and full page index are in the Key URLs section below.

---

## Intent Router

| Topic | Reference file | Load when... |
| --- | --- | --- |
| Overview, quickstart, plans | `references/overview.md` | user needs intro, what the agent is, pricing, getting started |
| Concepts: prompting, sandboxing, agents | `references/concepts.md` | user asks about how the agent works conceptually |
| Configuration: AGENTS.md, MCP, skills | `references/configuration.md` | user asks about configuring the agent's behavior |
| Interfaces: app, IDE, CLI, web, integrations | `references/interfaces.md` | user asks about which surface to use or how integrations work |
| Automation: SDK, non-interactive, GitHub Action | `references/automation.md` | user asks about running the agent in CI/CD or programmatically |

---

## Key URLs at a Glance

```text
Root                https://developers.openai.com/codex
Quickstart          https://developers.openai.com/codex/quickstart
Pricing / Plans     https://developers.openai.com/codex/pricing
Concepts            https://developers.openai.com/codex/concepts
AGENTS.md           https://developers.openai.com/codex/guides/agents-md
Configuration       https://developers.openai.com/codex/config-basic
Config reference    https://developers.openai.com/codex/config-reference
MCP                 https://developers.openai.com/codex/mcp
Skills              https://developers.openai.com/codex/skills
Subagents           https://developers.openai.com/codex/subagents
CLI docs            https://developers.openai.com/codex/cli
IDE Extension       https://developers.openai.com/codex/ide
App                 https://developers.openai.com/codex/app
Web / cloud         https://developers.openai.com/codex/cloud
GitHub integration  https://developers.openai.com/codex/integrations/github
Slack integration   https://developers.openai.com/codex/integrations/slack
GitHub Action       https://developers.openai.com/codex/github-action
Non-interactive     https://developers.openai.com/codex/noninteractive
SDK                 https://developers.openai.com/codex/sdk
App server          https://developers.openai.com/codex/app-server
Security            https://developers.openai.com/codex/security
```

---

## What is the OpenAI Coding Agent?

The OpenAI coding agent is available in ChatGPT and via API. It can read,
write, test, and debug code autonomously across a codebase. Available as
a web app, IDE extension, terminal CLI, and GitHub integration.

## Quick Start

```text
Need setup or plans? Load references/overview.md.
Need AGENTS.md, MCP, hooks, skills, or subagents? Load references/configuration.md.
Need non-interactive, SDK, app-server, or GitHub Action? Load references/automation.md.
```

```bash
codex exec "summarize this repository and identify the main test command"
```