Codex Knowledge Base
Docker Sandboxes for Codex CLI: MicroVM Isolation, the sbx CLI, and When to Use External Sandboxing
Codex CLI ships with one of the strongest built-in sandboxes in the AI coding agent space — Landlock plus seccomp on Linux, Seatbelt on macOS, restricted.
Codex CLI v0.121: Marketplace CLI, Agent Identity, and the Road to Plugin Distribution
The v0.121.0-alpha.2 pre-release, tagged on 11 April 2026, is the most plugin-and-marketplace-focused Codex CLI release to date. Whilst v0.119 and v0.120.
Codex CLI and Terraform: Agent Skills, MCP Servers, and Infrastructure-as-Code Workflows
Infrastructure as Code is one of the highest-stakes domains for AI-assisted development. A hallucinated resource argument in a Terraform plan can provision.
Codex CLI for Technical Debt Reduction: Automated Refactoring, Dead Code Detection, and Dependency Upgrades
Technical debt accumulates silently — unused imports, deprecated API calls, dead functions that nobody dares delete.
The Codex CLI Notification Pipeline: OSC 9, Notify Hooks, and Never Missing an Agent Event
Agentic coding sessions routinely run for minutes — sometimes hours. You fire off a complex refactoring task, switch to a browser tab, and fifteen minutes.
Dynamic Session Control in Codex CLI: Mid-Session Switching of Models, Permissions, and Workflow Modes
A Codex CLI session is not a static environment. Every slash command that adjusts the model, the approval policy, the reasoning effort, or the workflow mode.
Cloud Agent Plugin Suites: AWS Led, Azure Followed, and the GCP Gap
In February 2026, AWS Labs open-sourced awslabs/agent-plugins — the first major cloud provider plugin suite built for AI coding agents . Five weeks later.
AWS Agent Plugins and the Compound Plugin Pattern: How Cloud Providers Are Shipping Production-Grade Agent Skills
On 17 February 2026, AWS Labs open-sourced awslabs/agent-plugins — the first major cloud provider plugin suite purpose-built for AI coding agents .
The Silent Guessing Problem: Why AI Coding Agents Don't Ask Clarifying Questions and What AMBIG-SWE Means for Codex CLI
A Carnegie Mellon research team has published one of the most practically important findings for agentic coding in 2026: when given ambiguous instructions.
Agnix: Linting Your Codex CLI Agent Configurations Before They Break Your Workflow
Your AGENTS.md is 600 lines of carefully crafted instructions. Your config.toml has nested profiles, MCP server declarations.