codex

Configure and run OpenAI Codex CLI with sandbox and approval policies.

15|2|Updated May 23, 2026
One-click install
npx skills add https://github.com/VKirill/antigravity-for-claude-code --skill codex-vkirill
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: codex
Source: https://github.com/VKirill/antigravity-for-claude-code/tree/main/skills/codex
Command: npx skills add https://github.com/VKirill/antigravity-for-claude-code --skill codex-vkirill

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill helps you use OpenAI Codex CLI to perform real coding work efficiently while maintaining strong safety controls via sandboxing and approval policies.

Core Features & Use Cases

  • Installing and authenticating Codex: set up @openai/codex (npm) or the official binaries and confirm versions.
  • Configuring config.toml for reliable runs: define models, reasoning effort, sandbox modes, approval policies, profiles, and MCP servers.
  • Running safely in interactive, headless, and embedded modes: use codex, codex exec, and codex app-server (v0.130+) for CI and long-running orchestration.

Quick Start

Delegate a read-only review of a local change set by instructing an AI to run the Codex skill with read-only sandbox and untrusted approvals for an unfamiliar repository.

Frequently Asked Questions about codex

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I configure OpenAI Codex CLI for safe, automated code reviews?

Configure `.codex/config.toml` to define models, sandbox modes, and approval policies for OpenAI Codex CLI. Use a read-only sandbox with untrusted approvals to safely delegate code reviews without risking unintended file modifications.

Can I run OpenAI Codex CLI in headless CI CD pipelines?

Yes, OpenAI Codex CLI supports headless CI CD pipelines via `codex exec` for non-interactive runs and `codex app-server` (v0.130+) for long-running orchestration. Apply sandbox and approval profiles from `config.toml` to secure the automated runs.

What is a sandbox policy in config.toml for OpenAI Codex?

A sandbox policy in `.codex/config.toml` restricts OpenAI Codex CLI's file system and command execution permissions. Setting modes like read-only prevents unintended modifications during agentic terminal coding, refactoring, and code review tasks.

How do I add and configure MCP servers with OpenAI Codex CLI?

Add and configure MCP servers for OpenAI Codex CLI using the `codex mcp add` command. Define server profiles and connection settings within `.codex/config.toml` to integrate external tools into your interactive and headless agentic coding workflows.

Does OpenAI Codex CLI require npm to install and set up?

OpenAI Codex CLI can be installed via npm using the `@openai/codex` package or through official binaries. After installing, authenticate the CLI and confirm the active version before configuring profiles and sandbox policies.

When should I use codex app-server instead of interactive terminal sessions?

Use `codex app-server` (v0.130+) for embedded workflows and long-running orchestration instead of interactive terminal sessions. It enables headless, automated task execution where continuous background processing is required over manual input.