configuring-codex

Configure Codex CLI settings and verify AGENTS.md skill projection.

4|Updated Dec 21, 2025
One-click install
npx skills add https://github.com/greenheadHQ/nixos-config --skill configuring-codex
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: configuring-codex
Source: https://github.com/greenheadHQ/nixos-config/tree/main/.claude/skills/configuring-codex
Command: npx skills add https://github.com/greenheadHQ/nixos-config --skill configuring-codex

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Configure Codex CLI settings and troubleshoot the projection of AGENTS.md/.agents/skills to ensure project- and global-scope skills are reliably discovered and activated, and document the differences between Codex CLI and Claude Code behavior to prevent misconfigurations.

Core Features & Use Cases

  • Policy configuration and model settings in ~/.codex/config.toml (approval_policy, sandbox_mode) and model selection.
  • Projection validation for project vs. global scopes: verify AGENTS.md -> CLAUDE.md and .agents/skills symlinks to ensure consistent discovery.
  • Runtime verification: run verify-ai-compat.sh and test codex exec to confirm project-scope skills appear during activation.
  • Clear guidance on aligning Codex CLI behavior with Claude Code to minimize setup errors and surprises.

Quick Start

Run verify-ai-compat.sh and confirm that Codex CLI discovers the project-scope skills.

Frequently Asked Questions about configuring-codex

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

FAQPage Schema
How do I verify Codex CLI discovers my project-scope skills during activation?

Run verify-ai-compat.sh and execute codex exec to confirm that project-scope skills are properly discovered through AGENTS.md and .agents/skills symlinks during the activation workflow.

How do I configure Codex CLI approval_policy and sandbox_mode settings?

Configure Codex CLI policy controls by editing ~/.codex/config.toml to define your approval_policy and sandbox_mode settings, ensuring runtime behavior aligns with your project's security and execution requirements.

Does Codex CLI work with Claude Code AGENTS.md skill projections?

Codex CLI and Claude Code have distinct behaviors for discovering projected skills. You must align configurations by verifying AGENTS.md to CLAUDE.md symlinks to prevent misconfigurations across project and global scopes.

What is the best way to troubleshoot .agents/skills symlinks not projecting in a Codex workspace?

Troubleshoot .agents/skills projection by running verify-ai-compat.sh to validate directory symlinks, checking AGENTS.md mapping, and confirming project versus global scope configurations are correctly applied.

Why does my Codex CLI config.toml not apply the expected sandbox_mode?

Codex CLI config.toml sandbox_mode issues often stem from misaligned policy controls or scope conflicts. Validate your ~/.codex/config.toml settings and ensure project-scope skills do not override global configurations unexpectedly.

Do I need verify-ai-compat.sh to set up Codex CLI global and project skills?

You need verify-ai-compat.sh to validate that Codex CLI correctly discovers both global and project-scope skills, ensuring AGENTS.md projections and directory symlinks are functioning before executing codex exec.