From superpowers-ccg
Routes work to Codex (backend) and Gemini (frontend) via MCP tools for cross-validation on full-stack tasks, debugging, and code review.
How this skill is triggered — by the user, by Claude, or both
Slash command
/superpowers-ccg:coordinating-multi-model-workThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
- [Coordinating Multi-Model Work](#coordinating-multi-model-work)
Use this module to decide whether to call Codex (backend) and/or Gemini (frontend) via the Codex/Gemini MCP tools, and to enforce a consistent evidence protocol.
This module is intentionally minimal: it provides a small workflow and pushes details into reference files.
At any time, you must consider how the current process can collaborate with Codex/Gemini and how to call the MCP tools they provide to ensure objective, comprehensive analysis. You must execute the steps below: 1 After forming an initial analysis of the user request, share the request and your initial thinking with Codex/Gemini and ask them to improve the requirement analysis and implementation plan. 2 Before implementing any concrete coding task, you must ask Codex/Gemini for a code prototype (they must provide only a unified diff patch and must not make any real code changes). After receiving the prototype, you may only use it as logical reference, then rewrite your own production-grade, highly readable, highly maintainable code before making changes. 3 Whenever you complete real coding work, you must immediately use Codex/Gemini to review the code changes and requirement coverage. 4 Codex/Gemini provide references only. You must think independently and even question their answers. Blind trust is worse than no trust; your joint mission is to converge on a unified, comprehensive, precise result, which requires continuous debate to reach the truth.
Main rule (Fail-Closed Gate): If you decide Routing != CLAUDE, you MUST obtain external output, or STOP in BLOCKED.
Early exposure: If you decide Routing != CLAUDE, run the external call before doing real work (writing code, generating tests, or producing final conclusions).
At skill checkpoints (CP1/CP2/CP3):
coordinating-multi-model-work/routing-decision.mdRouting != CLAUDE, apply coordinating-multi-model-work/GATE.md immediately (evidence or BLOCKED)coordinating-multi-model-work/checkpoints.mdcoordinating-multi-model-work/routing-decision.mdcoordinating-multi-model-work/GATE.mdcoordinating-multi-model-work/INTEGRATION.mdcoordinating-multi-model-work/routing-rules.mdnpx claudepluginhub bryanhoo/superpowers-ccgOrchestrates Codex and Gemini CLI workers in parallel via tmux, then synthesizes results. Use for tasks benefiting from split backend/Codex and frontend/Gemini analysis.
Decomposes large tasks into parallel work across multiple LLM providers using a Double Diamond methodology (Discover, Define, Develop, Deliver). Use for migrations, multi-file refactors, or batch work.
Delegates implementation and review tasks to external AI CLI tools (Codex, Gemini) with cross-model adversarial review for cost savings and improved accuracy.