Codex Related Questions

Author: goswitch

Updated: 2026-06-13T10:02:01.000Z

A Few Tips for More Efficient Use of Codex

Many people may think that the model becomes less effective after using Codex for a period, experiencing so-called "degradation" phenomenon. However, based on my experience, the models provided in Codex have been upgraded many times without any "degradation" — the key is how you use the model appropriately

1. Task Breakdown: Never submit a very vague task, such as "Please write a management system backend" — this will inevitably lead to degradation! Codex models are characterized by rigorous and orderly, following instructions exactly — you need to break down your tasks

2. Stay in Control: Before starting a task, evaluate whether it is broken down enough and follows the "modular development" principle. Before submitting the task, you should be able to predict which files Codex will modify and what changes it will occur. Do not let AI go beyond your understanding and control — otherwise the project will become increasingly chaotic until having to start from scratch

Some Thoughts

Honestly, AI has made many things very simple, but basic knowledge determines your upper limit for using AI. At the current stage, AI is just a very excellent Copilot role. This also means that the same AI can have used differently by different people!

3. Avoid Compression: In most scenarios, your task should be completed using at most about 60% of Codex's context. If your task exceeds 60% of context and still unresolved and even requires compression, then the task breakdown before execution has failed — you need to break down the task more carefully. An excellent Codex vibe coding user almost never needs to compress content!

Smooth Codex Usage on Windows!

Important

This method simultaneously solves file reading/writing, encoding issues, high token consumption, project no memory and other pain points

1. Ensure your Codex CLI and VSCode Codex plugin are working properly, that i.e you can successfully converse with the model in the VSCode Codex plugin

2. Press "Win+R" enter the following content and press Enter to open your user directory

%userprofile%\.codex

3. Find the config.toml file in the directory, open and edit it. Your configuration file should look like this:

model_provider = "goswitch"
model = "gpt-5.4"
model_reasoning_effort = "high"
network_access = "enabled"
disable_response_storage = true
windows_wsl_setup_acknowledged = true
model_verbosity = "high"

[model_providers.goswitch]
name = "goswitch"
base_url = "https://goswitch.online/v1"
wire_api = "responses"
requires_openai_auth = true

4. Open the AGENTS.md file in the directory (create it manually if it doesn't exist), write the following content and save

# Codex Global Work Guidelines

## Answer Style:
 - Answers must use English
 - For summaries, Plans, Tasks, and long content output, prioritize logical organization and use a beautiful Table format; output normal content normally

5. Run VSCode, open the Codex plugin and and let's see what's different!

Common Codex Commands

Command	Description
/model	Select the current model
/approvals	Set approval rules for this session
/review	Let Codex review current workspace changes
/resume	Select and continue a previous session from the session history list
/new	Start a new conversation in the current CLI session
/init	Generate AGENTS.md template in the current directory
/compact	Sumize conversation content to release context
/undo	Undo Codex's last operation
/diff	View current git diff (including untracked files)
/mention	Add specified file or directory to the conversation context
/status	View session configuration and token usage
/mcp	List currently available MCP tools
/exit	Exit Codex CLI

Codex Encoding Issues on Windows

1. Press Win + R to open the run window, enter the following command and press Enter

intl.cpl

2. Click the "Administrative" tab above, then click the "Change System Locale Settings" button shown by the red arrow

3. Check the option shown by the red arrow, click OK. Then click OK in the previous window, restart your computer, and use Codex to avoid encoding issues

Set Latest Model in VSCode Codex Plugin

Windows

1. Press Win + R to open the run window, enter the following command in press Enter

%userprofile%\.vscode\extensions

2. Find the folder starting with openai.chatgpt, - the numbers after it are the version number, if there are multiple such directories, enter the directory with the latest version number

3. Navigate to webview\assets folder, you will see many js files

4. Download the replace script, and unzip it, copy this js file to the directory with many js files

   Download the replace script

   Automatically read the latest version number and file name, click to download directly

   Waiting to get

   File name Download now

5. Restart VSCode, and you can see that you can now select the latest model!

How to Configure Global Prompt in Codex

1. Please check the first two steps in Codex CLI Configuration

2. The AGENTS.md file mentioned in the tutorial is the global prompt file in Codex. If this file does not be present, you need to create it manually, then write the prompt

3. Save the prompt, restart Codex or VSCode, the prompt will take effect

Enable Built-in Web Search in Codex

1. Please check the first two steps in Codex CLI Configuration

2. Open the config.toml file mentioned in the tutorial, add the following content

[features]
web_search_request = true

3. Run Codex and try

Codex Network Connection Issues in Container or CLI Sandbox

When Codex encounters network connection issues in the CLI sandbox or container (such as tun mode) (such as unable to pull installation packages), and other tools (such as terminal, Claude Code) work normally, this is usually caused by improper MTU settings. Solution:

• Change the MTU value to 1500, this setting can usually be changed in your Clash client.
• For users who cannot find Clash MTU settings on Linux, refer to this link: https://linux.do/t/topic/1220328

Connection failed Issue

Error message similar to: text Connection failed: error sending request for url (https://www.goswitch.com/v1/responses)This situation is caused by local network problems, troubleshoot according to the following steps:

1. Check whether the local network is working properly, and whether you can access other pages
2. Check whether your computer uses using a network proxy (VPN) tool, if so please turn it off
3. Use the terminal, run the codex command, try to send a conversation message in the CLI to determine whether it is a VSCode Codex plugin issue. If so, please restart VSCode and try again
4. If it still not work, bring your error screenshot and consult customer service or friends in the group

401 Error

Error message similar to: text exceeded retry limit, last status: 401 Unauthorized, request id: xxxxxx

1. Run the following command in Windows or MacOS terminal to determine whether environment variables exist

Windows

cmd /c "echo ================= OPENAI ENV CHECK ================= & ^if defined OPENAI_API_KEY (echo OPENAI_API_KEY = OK) else (echo OPENAI_API_KEY = MISSING) & ^if defined OPENAI_BASE_URL (echo OPENAI_BASE_URL = OK) else (echo OPENAI_BASE_URL = MISSING) & ^echo ========================================================="

If the following output appears, proceed directly to step 2

OPENAI_API_KEY = MISSINGOPENAI_BASE_URL = MISSING

If the output is different, please run the following command in the terminal and then proceed to step 2

cmd /c "setx OPENAI_API_KEY \"\" & setx OPENAI_BASE_URL \"\""

macOS``` bash echo "================= OPENAI ENV CHECK ================="if [ -z "$OPENAI_API_KEY" ]; then echo "OPENAI_API_KEY = MISSING"else echo "OPENAI_API_KEY = OK"fiif [ -z "$OPENAI_BASE_URL" ]; then echo "OPENAI_BASE_URL = MISSING"else echo "OPENAI_BASE_URL = OK"fiecho "========================================================"

If the following output appears, proceed directly to step 2
``` text
OPENAI_API_KEY = MISSINGOPENAI_BASE_URL = MISSING

If the output is different, run the following command in the terminal and then proceed to step 2

unset OPENAI_API_KEY OPENAI_BASE_URL

2. See the Codex Configuration in CLI chapter

Important

You need:

1. Check whether the ApiKey configuration in ~/.codex/auth.json is correct
2. Check whether the request address in ~/.codex/config.toml is correct

403 Error

Error message similar to:``` text unexpected status 403 Forbidden: {"error":{"message":"Usage not included in yourplan","type":"usage_not_included","param":null,"code":null,"plan_type":"basic"}}

This situation is caused by an account issue in the pool, you need to:1.  Use `Ctrl+C` to interrupt the conversation, if in VSCode, click the stop button

2.  Re-initiate the conversation and try again, observe whether this problem appears again
3.  If retrying more than 3 times is invalid, bring your error screenshot and consult customer service or friends in the group