Codex Review

The codex-review skill enables AI agents to perform comprehensive code reviews by analyzing files against a specific set of rules defined in a .codexrc configuration file. It automatically detects issues, suggests improvements, and ensures adherence to project-specific standards before merging or deploying changes.

When to use it

• Before committing large batches of code to ensure compliance with team conventions.
• During pull request workflows to validate logic and style automatically.
• When integrating new developers into a project to enforce consistent coding patterns immediately.
• For continuous integration pipelines to block builds that fail custom linting rules.

Key capabilities

• Parses .codexrc files to load custom rule sets for specific projects.
• Analyzes codebases for structural errors, style violations, and best practice deviations.
• Generates actionable feedback reports highlighting specific lines needing attention.
• Supports iterative review cycles by re-checking modified files after fixes are applied.

Example prompts

• "Run a full codex review on the src/components directory using our standard .codexrc rules."
• "Review this pull request's changes and list any violations found against the project configuration."
• "Fix the issues identified in the last codex review and regenerate the report."

Tips & gotchas

Ensure a valid .codexrc file exists in your repository root; without it, the skill defaults to generic checks that may miss project-specific nuances. For best results, keep rule definitions concise to avoid false positives during automated scans.