codex

Orchestrate OpenAI Codex CLI to review, challenge, or consult on code diffs.

Updated Apr 1, 2026

Search Tags:#cli #codex #code-review #adversarial #diffs #second-opinion #plan-management

Ready to run

npx skills add https://github.com/viraatdas/vstack --skill codex-viraatdas

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: codex
Source: https://github.com/viraatdas/vstack/tree/main/codex
Command: npx skills add https://github.com/viraatdas/vstack --skill codex-viraatdas

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Independent, automated second opinions on code changes by orchestrating the OpenAI Codex CLI to review, challenge, or consult on diffs.

Core Features & Use Cases

Modes: Review, Challenge, and Consult for different feedback styles on code changes.
Diff-aware: Operates against the current branch or a plan-scoped diff, supporting plan-based reviews.
Gatekeeping: Returns structured outputs with gate verdicts and token usage, with a deterministic prompting boundary to avoid prompt injection.

Quick Start

Run /codex review to start an automated code review against the base branch.

Frequently Asked Questions about codex

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I get an automated second opinion on my code changes?▼

To get an automated second opinion on code changes, you can run /codex review to orchestrate the OpenAI Codex CLI against your current branch diffs. It provides independent, adversarial feedback to challenge or consult on modifications.

What is adversarial code review and how does it work with diffs?▼

Adversarial code review automatically challenges and critiques branch diffs using the Codex CLI. It operates against the current branch or plan-scoped diffs, detecting changes against the base branch to return structured gate verdicts.

Can I use automated code review for plan-managed software engineering workflows?▼

Yes, you can use this automated code review for plan-managed workflows. It supports plan-scoped diffs, allowing you to review, challenge, or consult on code changes specifically tied to your active development plans.

How do I prevent prompt injection when running automated code reviews?▼

To prevent prompt injection during automated code reviews, the system enforces a deterministic filesystem boundary in its prompts. This isolates the review process and ensures the CLI only evaluates the intended diff.

Does the Codex CLI code review tool return structured gatekeeping outputs?▼

Yes, the Codex CLI code review tool returns structured gatekeeping outputs. It provides gate verdicts and token usage metrics after evaluating diffs, helping you enforce quality gates in your software engineering pipeline.