Codex Knowledge Base
Codex CLI Subagents: TOML Format, Parallelism and spawn_agents_on_csv
Codex CLI's multi-agent system lets a single session delegate specialised work to parallel child agents. Once you understand the TOML agent definition.
Model Selection in Codex CLI: Current Models and When to Use Each
Codex CLI exposes model selection as a first-class concern. You can specify a model on the command line for a single invocation.
Codex CLI MCP Integration: Connecting Agents to External Tools
There is a pattern I have noticed across thirty years of software infrastructure decisions. A powerful tool ships with a clean interface. Then someone asks.
Codex CLI Enterprise Deployment: Managed Policies and Team Configuration
Rolling Codex CLI out to a team of ten is a different proposition from running it on your own laptop. At scale you need guardrails that users cannot.
Codex CLI for CI/CD: codex exec, Non-Interactive Mode and Pipeline Integration
codex exec is Codex's non-interactive execution mode — no TUI, no prompts, just autonomous task completion. It's the entry point for every CI/CD integration.
Codex CLI in Practice: Real-World Benchmarks and What They Mean
Benchmark numbers dominate marketing copy, but most developers lack the context to interpret them critically.
Claude Code ↔ Codex CLI: Bidirectional MCP Integration
The Model Context Protocol (MCP) was designed as a universal interface between AI agents and external tools.
AGENTS.md Advanced Patterns: Nested Hierarchies, Override Files and Fallbacks
The basic three-tier hierarchy (~/.codex/AGENTS.md → repo root → subdirectory) is documented everywhere. This article covers what isn't: override files.
Agentic Primitives Compared: Codex CLI vs Claude Code vs Gemini CLI
All three tools — OpenAI Codex CLI, Anthropic Claude Code, and Google Gemini CLI — converge on the same set of agentic primitives. They read instructions.
The Agentic Engineering Pod: Three Roles, One Shared Context Layer
Traditional software team structures were not designed for agentic delivery. They are built around specialisations: product managers own requirements.