Codex Review & Second Opinion Skill

Use OpenAI's Codex CLI to get code reviews and second opinions. This provides an independent AI perspective using GPT-5.3 or o3 models.

When to Use This Skill

Trigger when user:

•Uses /codex command explicitly
•Asks for a "code review"
•Wants a "second opinion" on code or approach
•Says "review this", "check this code", "validate this"
•Asks "what do you think about this implementation"
•Wants to "double check" code changes
•Requests "another perspective" on code

Prerequisites

Codex CLI must be installed and authenticated:

bash

# Check installation
which codex

# Login if needed
codex login

Workflow

Step 1: Determine What to Review

ALWAYS ask the user what they want reviewed:

Use AskUserQuestion to clarify:

•
Scope: What should be reviewed?
- •Uncommitted changes (staged/unstaged)
- •Specific file(s)
- •Changes against a branch
- •A specific commit
- •General approach/implementation question
•
Model: Which model to use?
- •gpt-5.3-codex - Fast, good quality (default)

Step 2: Execute Review

IMPORTANT: Always use --full-auto to avoid confirmation prompts.

Mode 1: Review Uncommitted Changes

bash

codex review --full-auto --uncommitted -m MODEL "CUSTOM_INSTRUCTIONS"

Mode 2: Review Against Branch

bash

codex review --full-auto --base BRANCH -m MODEL "CUSTOM_INSTRUCTIONS"

Mode 3: Review Specific Commit

bash

codex review --full-auto --commit SHA -m MODEL "CUSTOM_INSTRUCTIONS"

Mode 4: Second Opinion on Specific Files

bash

codex exec --full-auto -m MODEL "Review these files for issues, improvements, and best practices: FILE1 FILE2"

Mode 5: General Second Opinion

bash

codex exec --full-auto -m MODEL "I need a second opinion on: USER_QUESTION"

Step 3: Save Output

After review completes, save output to markdown file:

•Location: Same directory as reviewed code or current working directory
•Filename pattern: codex_review_YYYYMMDD_HHMMSS.md

Command Reference

Code Review Commands

bash

# Review uncommitted changes (no confirmations)
codex review --full-auto --uncommitted -m gpt-5.3-codex

# Review with custom instructions
codex review --full-auto --uncommitted -m gpt-5.3-codex "Focus on security vulnerabilities and performance"

# Review against main branch
codex review --full-auto --base main -m gpt-5.3-codex

# Review specific commit
codex review --full-auto --commit abc123 -m gpt-5.3-codex

# Review with title context
codex review --full-auto --uncommitted --title "Add user authentication" -m gpt-5.3-codex

Second Opinion Commands

bash

# Ask about specific code
codex exec --full-auto -m gpt-5.3-codex "Review this implementation approach and suggest improvements"

# Get opinion on architecture
codex exec --full-auto -m o3 "Evaluate this architecture decision: USER_CONTEXT"

# Review specific files
codex exec --full-auto -m gpt-5.3-codex -C /path/to/project "Review src/auth.py for security issues"

Output Capture

bash

# Save last message to file (preferred method)
codex exec --full-auto -m gpt-5.3-codex -o review_output.md "Review prompt here"

# Or capture full output
codex review --full-auto --uncommitted -m gpt-5.3-codex 2>&1 | tee review.md

Usage Examples

Example 1: Review Before Commit

code

User: Review my changes before I commit

Claude asks: What model? (gpt-5.3-codex / o3)
User: gpt-5.3-codex

Claude executes:
codex review --full-auto --uncommitted -m gpt-5.3-codex -o codex_review_20260111_140000.md

Reports: Review saved to codex_review_20260111_140000.md

Example 2: Second Opinion on Approach

code

User: Get a second opinion on my authentication implementation

Claude asks: Which files? What specific concerns?
User: src/auth.py - is my JWT handling secure?

Claude executes:
codex exec --full-auto -m gpt-5.3-codex -o codex_review_auth.md "Review src/auth.py focusing on JWT security. Is the implementation secure? Any vulnerabilities?"

Example 3: Compare with Main

code

User: Review my feature branch against main

Claude executes:
codex review --full-auto --base main -m gpt-5.3-codex -o codex_review_feature.md "Review all changes for code quality and potential bugs"

Available Models

Model	Best For	Speed
`gpt-5.3-codex`	General reviews, quick feedback	Fast
`o3`	Deep analysis, complex reasoning	Slower

AskUserQuestion Template

When triggered, ask:

code

Question 1: "What would you like Codex to review?"
Options:
- Uncommitted changes (git working tree)
- Changes against a branch
- Specific file(s)
- General question/second opinion

Question 2: "Which model should I use?"
Options:
- gpt-5.3-codex (faster)
- o3 (more thorough)

Output Handling

•Use -o filename.md to save output directly (cleaner than tee)
•Save to: codex_review_YYYYMMDD_HHMMSS.md in current directory
•Report file location to user after completion
•Display key findings summary

Error Handling

Error	Solution
Not logged in	Run `codex login`
Not in git repo	Use `-C /path` or `--skip-git-repo-check`
Model not available	Fall back to `gpt-5.3-codex`
Timeout	Try with smaller scope or faster model

Tips

•Be specific: Provide context about what to focus on
•Review incrementally: Review smaller changes more frequently
•Custom instructions: Add focus areas like "security", "performance", "readability"
•Working directory: Codex needs to be run from the project directory
•Always use --full-auto: Prevents interactive confirmation prompts

Limitations

•Requires internet connection and OpenAI API access
•Rate limits may apply
•Large codebases may need scoped reviews
•Cannot review files outside git repository by default

Non-Git Directories

When running outside a git repository, use --skip-git-repo-check:

bash

codex exec --full-auto --skip-git-repo-check -m gpt-5.3-codex "Your prompt here"

This is useful for:

•Reviewing standalone scripts
•Getting opinions on code snippets
•General programming questions

Alternative: Codex MCP Server

For tighter integration without shell overhead, Codex can run as an MCP server. See the codex-mcp MCP server configuration for native tool integration.