OpenAI Codex CLI setup guide

Overview

Codex CLI is a terminal coding agent that reads, writes, and runs code through a sandbox and reviewable patches. Point a custom Responses provider at MoLoop to drive it with catalog models.

Quick configuration

Codex uses a /v1 Responses endpoint and API key.

Configure Codex with CC Switch

Import the current MoLoop Responses endpoint and API key into the Codex configuration in CC Switch.

Endpoint: https://api.moloop.io/v1

基础 URL: https://api.moloop.io/v1
API 密钥: YOUR_API_KEY

• Create an API key

• View the CC Switch guide

Compatibility

• Chat Completions

• Responses API

• Streaming

• Tool calling

Step-by-step setup

Install Codex CLI

Install Codex CLI globally with npm.

npm install -g @openai/codex

Configure the model provider

Edit ~/.codex/config.toml with the model, MoLoop base_url, and responses wire API.

# ~/.codex/config.toml
model = "YOUR_MODEL_ID"

[model_providers.moloop]
name = "MoLoop"
base_url = "https://api.moloop.io/v1"
env_key = "OPENAI_API_KEY"
wire_api = "responses"

Add the API key

Put the key in the credential file supported by Codex, or set OPENAI_API_KEY in the shell that launches Codex.

// ~/.codex/auth.json
{
  "OPENAI_API_KEY": "YOUR_API_KEY"
}

Start Codex

Run codex in the project directory, ask about the repository first, and only then make changes.

codex

Recommended models

Choose a model that supports Responses and coding tools. Use the MoLoop catalog ID directly without an openai/ prefix.

• Browse the model catalog

Things to know

• unsupported endpoint usually means wire_api is not responses or base_url does not end in /v1.

• For a 401, check OPENAI_API_KEY in auth.json or the launch shell, then restart Codex.