What sandbox modes are available for security?

Codex supports several sandbox modes including 'read-only' for safety, 'workspace-write' for typical development, and 'danger-full-access' for unrestricted execution.

How do I add custom tools or services to Codex?

You can add custom tools by configuring MCP (Model Context Protocol) servers in your config.toml file, specifying the command, arguments, and required environment variables.

Where is the primary configuration file located?

The primary configuration for Codex is stored in ~/.codex/config.toml, where you can define model settings, permissions, and profiles.

How can I switch between different model configurations quickly?

You can define named profiles in your config.toml, such as [profiles.max], and invoke them using the -p flag (e.g., 'codex -p max').

Can I use local language models with this configuration?

Yes, by using the --oss flag or setting the model_provider to 'oss' in your configuration, you can connect Codex to local providers like LM Studio or Ollama.

Codex Configuration Manager

Name: Codex Configuration Manager
Author: outfitter-dev

byoutfitter-dev

•

开发者工具

Manages and optimizes OpenAI Codex CLI settings, profiles, and sandbox policies for streamlined AI-assisted development.

关于

The Codex Configuration Manager skill provides a comprehensive framework for administering the OpenAI Codex CLI environment. It enables developers to fine-tune model parameters, establish named performance profiles for different workflows, and manage Model Context Protocol (MCP) server integrations. By automating the configuration of sandbox permissions, skill discovery paths, and runtime overrides, this skill ensures a secure, high-performance, and project-specific environment for AI coding agents.

主要功能

Register and manage MCP servers with custom environment variables
Manage config.toml settings for models, verbosity, and reasoning effort
18 GitHub stars
Handle multi-level skill discovery paths from project to system level
Define named profiles like 'max' for complex tasks or 'fast' for quick fixes
Configure sandbox modes and approval policies for secure code execution

使用场景

Configuring project-specific sandbox permissions to allow workspace writes while preventing system-wide access
Switching between high-reasoning models for refactoring and low-latency models for documentation
Troubleshooting MCP server connection issues and validating TOML configuration syntax

关于

主要功能

Register and manage MCP servers with custom environment variables
Manage config.toml settings for models, verbosity, and reasoning effort
18 GitHub stars
Handle multi-level skill discovery paths from project to system level
Define named profiles like 'max' for complex tasks or 'fast' for quick fixes
Configure sandbox modes and approval policies for secure code execution

使用场景

Configuring project-specific sandbox permissions to allow workspace writes while preventing system-wide access
Switching between high-reasoning models for refactoring and low-latency models for documentation
Troubleshooting MCP server connection issues and validating TOML configuration syntax