skills.rest
rafazafar
rafazafarCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

codex-review-loop

Automates adversarial PR code reviews using Codex CLI with structured findings and iterative HITL-guided fixes.

Updated Apr 11, 2026
Search Tags:#codex#review#code-analysis#adversarial#pull-request#hitl
One-click install
npx skills add https://github.com/rafazafar/codex-octopus --skill codex-review-loop-rafazafar
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: codex-review-loop
Source: https://github.com/rafazafar/codex-octopus/tree/main/.agents/skills/codex-review-loop
Command: npx skills add https://github.com/rafazafar/codex-octopus --skill codex-review-loop-rafazafar

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires codex, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Adversarial, structured PR code reviews using Codex CLI to surface actionable findings and drive HITL-enabled fixes in a repeatable loop.

Core Features & Use Cases

  • Phase-driven review workflow: automated Codex review, structured finding analysis, and an iterative fix loop guided by human-in-the-loop approvals.
  • Flexible targets and prompts: supports reviewing open PRs, specific branches, or uncommitted changes, with configurable model and reasoning effort.
  • Final reporting and verification: produces a consolidated findings report and a verification-ready final status.

Quick Start

Run the Codex review loop against your active PR to start gathering findings and iterations.

Frequently Asked Questions about codex-review-loop

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

FAQPage Schema
How do I automate adversarial code reviews for pull requests?▼

Adversarial PR code reviews are automated using Codex CLI to surface actionable findings. The workflow applies structured analysis to open pull requests, branches, or uncommitted changes to drive iterative fixes.

Can I review uncommitted changes or specific branches instead of open pull requests?▼

Uncommitted changes and specific branches can be reviewed using --uncommitted and --base flags. The workflow supports flexible targets, allowing Codex CLI to analyze local modifications alongside open pull requests.

How does the human-in-the-loop fix process work for code review findings?▼

The human-in-the-loop fix process works by mapping structured findings to a review schema for HITL-guided approvals. This enables iterative re-reviews where humans approve fixes before the loop produces a verification-ready final status.

Do I need Codex to run adversarial code analysis?▼

Codex is required to run this adversarial code analysis. The workflow depends on the Codex CLI to execute automated reviews and supports environment overrides to configure the specific model and reasoning effort.

What is the best way to structure code review findings for iterative fixes?▼

The best way to structure code review findings is mapping them to a structured review schema. This approach consolidates automated analysis into a final report, enabling human-in-the-loop guided fixes and verification-ready status tracking.

Are there limitations when using automated code review loops on large pull requests?▼

Automated code review loops depend on configurable model and reasoning effort via Codex CLI. While effective for driving iterative fixes, complex large pull requests may require multiple human-in-the-loop approvals to reach verification-ready status.

Related Skills

Explore More Software Engineering Skills→
aimskraimskr

codex

Automate Codex CLI exec commands with model, reasoning, and sandbox flags.

Official
Advanced
RobinNorbergRobinNorberg
5

deep-review

Run four-pass code review with parallel security, quality, and structural analysis.

Community
Advanced
PaparusiPaparusi

mcp-github

Manage GitHub repositories, branches, files, issues, and pull requests programmatically.

Community
Advanced
victormacaubasvictormacaubas
1

python-code-reviewer

Automate non-destructive Python code reviews with markdown reports.

Community
Advanced