Codex

Fast OpenAI coding agent support with profile-friendly model switching for parallel runs.

Agent guide

Codex is a strong fit when you want fast implementation loops, profile-based routing, and an OpenAI-native default for repositories that need different cost and model profiles.

Binary: codexAliases: openai, codex-cliBest for fast first-pass execution

Primary binary

codex

Known aliases

openai, openai-codex, codex-cli

Good fit

Fast implementation, parallel task execution, and profile-based routing.

​

Set up Codex in Conductor

1. Install Codex

   npm install -g @openai/codex-cli
   
2. Complete authentication

   Sign in with the OpenAI account or API flow required by your Codex CLI environment before wiring it into the workspace.

3. Launch Conductor

   npx --yes --registry=https://registry.npmjs.org conductor-oss@latest
   

   Use Add Workspace or update the project settings to make Codex the default for the target repository.

4. Configure the repo default

   projects:
     my-app:
       agent: codex
       agentConfig:
         model: gpt-5.3-codex-spark
   

​

Model selection in Conductor

Codex can use the OpenAI model catalog exposed by the environment behind your CLI.

That list can change based on ChatGPT plan access, API credentials, organization policy, or Codex CLI release.

• if you see a plan-filtered model list, leave the custom override blank to use that upstream selection
• set agentConfig.model only when a repository needs a fixed model string
• keep one stable repo default instead of switching models on every task

​

Why teams pick Codex

• fast first-pass implementation
• easy model-profile changes per repo
• useful when several repos need different cost/performance defaults