A comprehensive handbook covering OpenAI Codex — the AI coding agent product built on top of OpenAI's frontier models. It explains what Codex is (a workflow layer, not just a model), its four surfaces (CLI, IDE extensions, desktop app, and Codex Cloud), and how to set it up from scratch with a hands-on Python demo. The guide covers effective prompting strategies, comparisons with Claude Code, GitHub Copilot, and self-hosted alternatives, pricing details including the new GPT-5.5 token-based rate card, enterprise security and RBAC setup, team adoption best practices, and metrics for measuring Codex effectiveness. Updated to reflect the April 2026 GPT-5.5 release.
Table of contents
Executive SummaryTable of ContentsPrerequisitesSection 1: What Codex IsSection 2: Where Codex Fits in the OpenAI EcosystemSection 3: The Core SurfacesSection 4: Getting Started: Install, Set Up, and Your First TaskSection 5: How to Use Codex EffectivelySection 6: Difference Between Codex and Other Coding ToolsSection 7: Pricing and Plan AccessSection 8: Security, Permissions, and Enterprise SetupSection 9: Best Practices for TeamsSection 10: Common Workflows and ExamplesSection 11: Model Specs and Benchmarks (GPT-5.5 Deep Dive)Section 12: TroubleshootingSection 13: FAQSection 14: When NOT to Use CodexSection 15: Final RecommendationsSection 16: Source ReferencesAppendix A: 30-60-90 Day Adoption PlanAppendix B: GlossaryAppendix C: Admin Security ChecklistAppendix D: ChangelogAppendix E: Working with Codex in VS CodeAbout LunarTech LabSort: