Skip to content

Codex Integration

Whether you are using the Codex CLI, the Official VSCode Extension, or the Codex Desktop App, their underlying configuration files and core mechanics are identical and fully interoperable.

You only need to configure the authentication file once, and all three clients will be seamlessly connected!

If you prefer terminal operations, ensure Node.js is installed on your machine, then run:

Terminal window
npm i -g @openai/codex

Verify the installation: codex --version


Step 2: Locate the Configuration Directory

Section titled “Step 2: Locate the Configuration Directory”

Regardless of whether you are on Windows or macOS/Linux, all three official clients read from the .codex directory located in your system’s user folder.

Open your terminal or command prompt (CMD) and run the following command to navigate directly to this directory:

Run the following in CMD or PowerShell to automatically create (if missing) and open the folder in File Explorer:

Terminal window
mkdir "%USERPROFILE%\.codex"
start "" "%USERPROFILE%\.codex"

Inside the .codex folder you just opened, create the following two core plain text files (do not add a .txt extension):

This file instructs all clients to route their traffic to SU8.Codes:

model_provider = "su8"
model = "gpt-5.2"
# Or replace with any model from the Supported Models list, like gpt-5.3-codex
network_access = "enabled"
disable_response_storage = true
model_verbosity = "high"
[model_providers.su8]
name = "su8"
base_url = "https://www.su8.codes/codex/v1"
wire_api = "responses"
requires_openai_auth = true

⚠️ Critical Configuration Note:

  • Any changes made to these two files require a complete restart of the client to take effect (reload VSCode, or kill the codex process in the terminal with Ctrl+C and rerun it).
  • If the VSCode extension throws continuous errors, try running codex in the terminal first. This helps determine if the issue is network-related or specific to the IDE.