Where is the primary configuration file located?

The primary configuration for the Codex CLI is typically stored at ~/.codex/config.toml.

What are sandbox modes in Codex?

Sandbox modes define the agent's file system permissions, ranging from read-only to workspace-write and danger-full-access.

Can I override configuration settings at runtime?

Yes, use the -c flag to override any config value, such as 'codex -c model="o3"' for a specific command execution.

How do I switch between different model profiles?

You can use the -p flag followed by the profile name, such as 'codex -p max', to use predefined settings from your config.toml file.

How does the skill hierarchy work for Codex?

Codex searches for skills in a specific precedence: current directory (.codex/skills/), parent directory, repository root, user-level, and finally system-level.

Codex Configuration Manager

Name: Codex Configuration Manager
Author: outfitter-dev

byoutfitter-dev

•

开发者工具

Manages configuration, profiles, and security policies for the OpenAI Codex CLI environment.

关于

The Codex Configuration skill provides specialized guidance for managing the OpenAI Codex CLI ecosystem, enabling users to fine-tune model settings, establish sandbox security policies, and configure Model Context Protocol (MCP) servers. It streamlines the setup of named profiles for varying workflows—such as 'max' for complex tasks or 'fast' for quick fixes—and manages skill search hierarchies across local, repository, and system-level directories. This skill ensures your agentic environment is optimized for both performance and security through standardized validation and troubleshooting patterns.

主要功能

Granular sandbox policy configuration including read-only and workspace-write modes
Comprehensive MCP server integration for extending CLI capabilities with external tools
16 GitHub stars
Advanced profile management for switching between high-reasoning and fast-response models
Intelligent skill path resolution across local, repository, and user-level directories
Runtime CLI overrides for dynamic adjustment of model parameters and permissions

使用场景

Integrating third-party services like Linear or Graphite via MCP server configurations
Setting up custom security boundaries and approval policies for autonomous agent tasks
Switching between high-performance reasoning models and cost-effective profiles for different tasks

关于

主要功能

Granular sandbox policy configuration including read-only and workspace-write modes
Comprehensive MCP server integration for extending CLI capabilities with external tools
16 GitHub stars
Advanced profile management for switching between high-reasoning and fast-response models
Intelligent skill path resolution across local, repository, and user-level directories
Runtime CLI overrides for dynamic adjustment of model parameters and permissions

使用场景

Integrating third-party services like Linear or Graphite via MCP server configurations
Setting up custom security boundaries and approval policies for autonomous agent tasks
Switching between high-performance reasoning models and cost-effective profiles for different tasks