longranger2/claude-gpt-workflow

Runs an adversarial plan iteration workflow where Claude Code asks Codex to critically review a plan file, evaluates the feedback, revises the plan, and repeats until consensus is reached. Supports multi-round reviews with issue tracking and severity-based prioritization.

Delegates coding tasks to OpenAI's Codex CLI for cost-effective execution, ideal for batch refactoring, code generation, multi-file changes, and test writing when the implementation plan is clear.

Orchestrates a plan execution workflow where Claude Code reviews code and delegates all implementation and fixes to Codex, iterating through review rounds until quality standards are met.