cross-review
Cross-Review Loop: Claude x Codex CLI
Overview
Autonomous review-fix loop between Claude and Codex CLI. Each round: both review simultaneously and independently → triage findings → fix using the best available skill → repeat. Stops when clean, when reviewers disagree, or after max rounds.
Prerequisites
- Codex CLI installed:
npm install -g @openai/codex - Codex authenticated:
codex auth login - Config at
~/.codex/config.tomlwith model, reasoning effort, and multi-agent enabled:
# ~/.codex/config.toml (example — adjust model to your available version)
model = "gpt-5.4"
model_reasoning_effort = "xhigh"
More from dmitriyyukhanov/claude-plugins
python-architect
Creates high-level Python architecture with protocols, test stubs, and module structure. Use when designing Python projects or features.
23agent-teams
>
17python-testing
Python testing patterns with pytest including unit tests, integration tests, fixtures, mocking, and coverage. Use when writing Python tests.
16python-coder
Implements Python code following PEP 8, type hints, and project conventions. Use when implementing features designed by python-architect.
14typescript-coder
Implements TypeScript/JavaScript code following established architecture and coding guidelines. Use when implementing features designed by typescript-architect.
14unity-coder
Use when implementing Unity C# code to follow proper coding guidelines, naming conventions, member ordering, and Unity-specific patterns
13