Codex CLI: The Definitive Guide

From First Command to Production Workflows

1. About This Book

Introduction

1. What This Book Covers
2. Who This Book Is For
3. How to Read This Book
4. A Note on Safety

Chapter 1: What Is Codex CLI?

1. From TypeScript to Rust: The Evolution of Codex
2. Architecture Overview: How Codex Works Under the Hood
3. The Five Surfaces: CLI, Desktop, Web, IDE Extension and Cloud
4. When to Use Codex CLI (and When Not To)
5. Ecosystem Context: Codex vs. Claude Code, Cursor, Copilot
6. Acknowledging Codex CLI’s Shortcomings

Chapter 2: Installation and First Run

1. System Requirements and Prerequisites
2. Installing via npm, Homebrew, Install Script, or Binary
3. Windows Installation: Native and WSL2
4. Verifying Your Installation
5. Your First Interactive Session

Chapter 3: Authentication and Credential Management

1. ChatGPT OAuth Sign-In Flow
2. API Key Authentication
3. Enterprise Access Tokens
4. Credential Storage: File vs. Keyring
5. CI/CD Authentication Patterns
6. Troubleshooting Auth Failures

Chapter 4: Configuration Deep Dive

1. Config File Locations and Precedence
2. Core Settings: Model, Sandbox, Approval Policy
3. Feature Flags and Experimental Capabilities
4. Profiles for Different Workflows
5. Custom Model Providers (Azure, Bedrock, Ollama, OpenRouter)
6. Enterprise Managed Configuration

Chapter 5: The Interactive Terminal Experience

1. Starting an Interactive Session
2. Slash Commands Reference
3. Keyboard Shortcuts and Navigation
4. Image Input and Visual Context
5. Session Management: Resume, Clear and Compact
6. TUI Customization and Themes

Chapter 6: Sandboxing, Permissions and Security

1. Sandbox Modes: read-only, workspace-write, danger-full-access
2. Approval Policies: untrusted, on-request, never, granular
3. Auto-Review: Delegating Trust to an AI Reviewer
4. Network Access Controls and Proxy Configuration
5. OS-Level Sandboxing: Seatbelt, bwrap, Windows Sandbox
6. Security Best Practices and Common Pitfalls

Chapter 7: Project Context with AGENTS.md

1. What Is AGENTS.md and Where It Lives
2. Writing Effective Project Instructions
3. Layered Context: Global, Project and Override Files
4. Context Window Management and Token Budgeting
5. The .codexignore File
6. Cross-Tool Portability with AGENTS.md

Chapter 8: Skills – Reusable Task Expertise

1. What Are Skills and How They Work
2. The SKILL.md Format and Directory Structure
3. Skill Discovery: Implicit vs. Explicit Invocation
4. Creating Skills with $skill-creator
5. Installing and Distributing Skills
6. Best Practices for Writing Effective Skills

Chapter 9: Model Context Protocol (MCP) Integration

1. What Is MCP and Why It Matters
2. Configuring MCP Servers in config.toml
3. Managing Servers via codex mcp Commands
4. STDIO vs. Streamable HTTP Servers
5. Tool Control: Enabling, Disabling and Approval Modes
6. Real-World MCP Server Recipes and Walkthroughs

Chapter 10: Subagents and Multi-Agent Orchestration

1. How Subagents Work: Architecture and Benefits
2. Triggering Subagents: Prompts, AGENTS.md and Skills
3. Custom Agent Definitions in TOML
4. Parallel Execution Patterns and Git Worktrees
5. CSV Batching for High-Volume Tasks
6. Multi-Agent Best Practices and Cost Management

Chapter 11: Non-Interactive Mode and Automation

1. codex exec: Headless Execution
2. JSON Output and Schema Validation
3. Session Resumption in Automated Workflows
4. GitHub Actions Integration (openai/codex-action)
5. Scripting Patterns and Bash Wrappers
6. Stdin Piping and Prompt Composition

Chapter 12: Code Review and Git Integration

1. The /review Command: Modes and Scope
2. Pre-Commit Review Workflows
3. GitHub Pull Request Reviews with @codex review
4. Branch and Commit Inspection
5. Git Checkpoint Strategies
6. Custom Review Guidelines

Chapter 13: Codex Cloud and Remote Execution

1. What Is Codex Cloud and When to Use It
2. codex cloud Commands: exec, list, apply
3. Cloud Environment Setup Scripts and Container Caching
4. Remote SSH Connections
5. Best-of-N Attempts and Retry Strategies
6. Cloud vs. Local: Decision Framework

Chapter 14: Model Selection and Reasoning Configuration

1. The GPT-5.6 Family: Sol, Terra, Luna
2. Legacy Models: GPT-5.4, GPT-5.3-Codex, Spark
3. Known Model Limitations
4. Reasoning Effort Levels: none to ultra
5. Model Verbosity and Summary Modes
6. Cost vs. Capability Trade-Offs
7. Per-Agent and Per-Task Model Selection

Chapter 15: Performance Optimization

1. Understanding Your Token Budget
2. Context Compaction: How It Works
3. Reducing MCP Server Overhead
4. File Exclusion Strategies with .codexignore
5. Session Length Management
6. Cost Estimation and Monitoring

Chapter 16: Troubleshooting and Common Pitfalls

1. Authentication and Billing Errors
2. Sandbox and Permission Failures
3. Context Window Overflow
4. MCP Server Timeouts and Connection Issues
5. Installation and Environment Problems
6. The Diagnostic Mindset: Systematic Debugging

Chapter 17: Real-World Workflows and Case Studies

1. Solo Developer: Full-Stack Feature Implementation
2. Team Workflow: Code Review Pipeline Integration
3. CI/CD: Automated Testing and Bug Fixing
4. Legacy Codebase Refactoring
5. Documentation Generation and Maintenance
6. Security Audit Workflows

Chapter 18: Migration Strategies – From Other Tools to Codex CLI

1. Why Migrate? The Business Case
2. Migrating from Claude Code
3. Migrating from Cursor
4. Migrating from GitHub Copilot
5. Migrating from Aider
6. Team Migration: Onboarding Without Disruption
7. Credential and Provider Transitions
8. When Not to Migrate

Chapter 19: Advanced Techniques and Future Directions

1. Codex as an MCP Server for External Orchestration
2. The OpenAI Agents SDK Integration
3. Record and Replay: Building Skills from Sessions
4. Plugins: Distributable Skill Bundles
5. Hook Systems and Event-Driven Automation
6. The Road Ahead: What to Watch For

Conclusion

References

Glossary

Index