What is a 'blocking issue' in a Codex review?

A blocking issue is a high-severity problem related to security, correctness, or performance that must be resolved before the review can pass (ok: true).

What happens if the skill fails to fix an issue automatically?

If the issue persists after 5 iterations, the skill provides a detailed failure report in Japanese, allowing the user to choose between manual fixes or proceeding with known issues.

How does the skill handle very large code changes?

For changes exceeding 10 files or 500 lines, the skill triggers a parallel review strategy, launching multiple subagents to review specific modules before performing a final cross-check.

Is it safe to run this skill on sensitive code?

Yes, the Codex review executes within a read-only sandbox environment to ensure safety and prevent unauthorized modifications during the analysis phase.

Does this skill replace manual code reviews?

It serves as a powerful first-line quality gate that catches common errors and security flaws automatically, allowing human reviewers to focus on high-level architecture and business logic.

Codex Automated Review Gate

Name: Codex Automated Review Gate
Author: takumi12311123

bytakumi12311123

•

보안 및 테스팅

Automates code quality reviews and security audits as a mandatory quality gate before committing or deploying code changes.

소개

The Codex Automated Review Gate is a sophisticated quality assurance skill designed to act as a mandatory checkpoint within the Claude Code workflow. It automatically analyzes code diffs for logic errors, security vulnerabilities, performance bottlenecks, and maintainability issues using a read-only sandbox environment. By implementing a sophisticated review-fix-test loop, the skill identifies 'blocking' issues and attempts to resolve them autonomously before presenting the final implementation to the user. Whether handling small patches or large-scale architectural changes via parallel subagents, it ensures that every line of code meets high-quality standards and project-specific requirements.

주요 기능

Parallelized review strategy for large-scale changes (>500 lines)
1 GitHub stars
Multi-category audits including security, correctness, and performance
Comprehensive JSON-based reporting with Japanese language support
Mandatory pre-commit quality gate for Claude Code
Iterative review-fix loop with up to 5 automatic retry attempts

사용 사례

Automating the identification of edge cases and null-handling errors in complex refactors
Standardizing code quality across large repositories using AI-driven architectural cross-checks
Enforcing security and logic checks before every git commit or PR creation

소개

주요 기능

Parallelized review strategy for large-scale changes (>500 lines)
1 GitHub stars
Multi-category audits including security, correctness, and performance
Comprehensive JSON-based reporting with Japanese language support
Mandatory pre-commit quality gate for Claude Code
Iterative review-fix loop with up to 5 automatic retry attempts

사용 사례

Automating the identification of edge cases and null-handling errors in complex refactors
Standardizing code quality across large repositories using AI-driven architectural cross-checks
Enforcing security and logic checks before every git commit or PR creation