codex-bridge

Multi-round prompt refinement for plan-mode execution via Claude↔Codex iteration. Use when: (1) User says "/claudex", "use claudex", or "claudex this" (2) User says "plan how to...", "investigate and plan...", "design an approach for..." (3) User says "refine this prompt for planning", "help me plan this properly" (4) User explicitly wants Codex to help improve their planning prompt before execution Do NOT trigger for: - Simple questions that don't require planning ("what does X do?") - Tasks the user wants executed immediately without planning - When user says "ask codex" or "/codex" (use codex-bridge skill instead) - Trivial tasks that don't benefit from detailed planning

Integration between Claude Code and OpenAI Codex CLI via MCP. Use when: (1) the user says "ask codex", "send to codex", "use codex", or "/codex", (2) the user wants a second opinion from another AI coding agent, (3) the user wants to compare Claude's approach with Codex's approach, (4) the user wants to delegate a subtask to Codex while staying in Claude Code. Provides multi-turn session support via threadId for persistent conversations with Codex within a single Claude Code session.