codex

name	codex
description	Delegate coding tasks to OpenAI Codex CLI
category	agent-bridge
version	1.0.0
origin	aiden
license	Apache-2.0
tags	openai, codex, cli, code-generation, agent-bridge, gpt, automation, coding, delegation

OpenAI Codex CLI Agent Bridge

Delegate code generation, editing, and explanation tasks to the OpenAI Codex CLI. Codex CLI runs GPT-4o in an agentic coding loop with sandboxed code execution.

When to Use

• User wants to generate code using OpenAI's GPT-4o model
• Task requires code execution in a sandboxed environment
• User wants an alternative to Claude Code using the OpenAI API
• User wants to benchmark or compare code generation between providers
• Task requires OpenAI-specific features or models

How to Use

1. Install Codex CLI

# Requires Node.js 18+
npm install -g @openai/codex

# Verify
codex --version

# Set API key
$env:OPENAI_API_KEY = "sk-..."

2. Run a coding task interactively

Set-Location "C:\Users\<you>\myproject"
codex "Add input validation to all form fields in the React components"

3. Non-interactive mode

# Auto-approve all edits (use with caution)
codex --approval-mode auto-edit "Add JSDoc comments to all exported functions in src/"

# Full auto mode (applies changes without confirmation)
codex --approval-mode full-auto "Fix all ESLint errors in the project"

4. Use a specific model

# Use GPT-4o (default)
codex --model gpt-4o "Optimize the database query in src/db.js"

# Use GPT-4o-mini for faster, cheaper tasks
codex --model gpt-4o-mini "Add error handling to the fetch calls"

5. Ask a coding question without editing files

# Question mode — returns explanation only
codex --question "What does the middleware chain in src/app.ts do?"

6. Generate code from scratch

# Create a new file/module
codex "Create a TypeScript utility module with functions for:
- Formatting currency (Indian Rupee)
- Parsing date strings to Date objects
- Debouncing function calls
Save to src/utils/format.ts"

7. Explain existing code

codex "Explain what the agentLoop.ts file does, focusing on the tool execution flow"

Examples

"Generate a REST API client for the GitHub API in TypeScript" → Use step 6: codex "Create a TypeScript GitHub API client with methods for: list repos, create issue, get PR. Save to src/github-client.ts".

"Fix all TypeScript compilation errors in the project" → Use step 3 with --approval-mode auto-edit after reviewing what changes will be made.

"Explain how the authentication middleware works" → Use step 5 or 7 in question mode — returns an explanation without modifying files.

Cautions

• --approval-mode full-auto applies all changes without confirmation — always run on a clean git branch
• Codex CLI bills against your OpenAI API key — monitor usage at platform.openai.com/usage
• OPENAI_API_KEY must be set in the environment before running any commands
• Codex executes code in a sandbox but may read any files in the current directory — scope to the right project folder
• For very large codebases, provide a focused scope (specific file or module) rather than asking it to scan the entire project