/ 目录 / 演练场 / claude-gpt-workflow
← 全部服务器 ↗ GitHub
● 社区 longranger2 ⚡ 即开即用

claude-gpt-workflow

作者 longranger2 · longranger2/claude-gpt-workflow

Delegates code execution to Codex CLI while Claude orchestrates — adversarial plan review and iterative refinement between two AI models.

Three skills that split roles: Claude orchestrates strategy, Codex executes. codex delegates coding tasks to Codex CLI. /plan-review has Codex adversarially critique Claude's plan before implementation. /plan-execute runs the approved plan with back-and-forth review loops until quality passes. Three feedback loops — plan refinement, review/fix, batch execution.

▶ 观看演示 📦 安装 💡 使用场景

为什么要用

核心特性

实时演示

实际使用效果

claude-gpt-workflow-skill.replay ▶ 就绪
0/0

安装

作者仓库 · Claude Skills 官方文档

使用场景

实战用法： claude-gpt-workflow

Stress-test a plan by having another model attack it

👤 Engineers doing high-stakes refactors ⏱ ~45 min intermediate

何时使用： Before touching code on a migration or risky refactor, you want a second model to poke holes.

步骤
  1. Draft a plan with Claude
    Write a plan.md for migrating our auth module from JWT to session cookies.✓ 已复制
    → Step-by-step plan written to plan.md
  2. Trigger adversarial review
    /plan-review plan.md✓ 已复制
    → Codex returns critique with risks Claude missed
  3. Iterate until both agree
    Incorporate Codex's critique, re-submit, repeat until clean.✓ 已复制
    → Plan reaches stable state

结果： A plan that has survived real adversarial review, not just self-grading.

注意事项
  • Endless review loops — Cap at 3 rounds — diminishing returns after
搭配使用： filesystem · git

Delegate a batch refactor to Codex with Claude reviewing

👤 Teams doing mechanical multi-file changes ⏱ ~60 min intermediate

何时使用： A rename / signature change across 50+ files.

步骤
  1. Plan and approve
    Plan + review a rename of <old> to <new> across the codebase.✓ 已复制
    → Approved plan.md
  2. Execute
    /plan-execute plan.md — Codex does the edits, Claude reviews each batch.✓ 已复制
    → Files change in batches, each passes Claude's review

结果： A clean multi-file change with two sets of eyes on every batch.

注意事项
  • Codex does something Claude would have rejected — The /plan-execute loop catches it — but only if quality criteria in the plan are explicit
搭配使用： git

Implement a feature with iterative quality refinement

👤 Devs who've seen one-shot AI code fall over on edge cases ⏱ ~90 min advanced

何时使用： A feature complex enough that a single generation won't get it right.

步骤
  1. Write plan with acceptance criteria
    Plan a feature with explicit acceptance tests, then /plan-execute.✓ 已复制
    → Plan with testable criteria
  2. Watch the loop
    Let the review/fix loop run — summarize each iteration.✓ 已复制
    → Iterations converge to passing criteria

结果： A feature that passes explicit acceptance, not just a commit that compiles.

组合

与其他 MCP 搭配，撬动十倍杠杆

claude-gpt-workflow-skill + git

Commit after each loop iteration so you can revert if the loop diverges

Before each /plan-execute iteration, create a git tag.✓ 已复制
claude-gpt-workflow-skill + filesystem

Plan and critiques live in the repo

Write plan.md and review-notes.md to docs/ and keep them in git.✓ 已复制

工具

此 MCP 暴露的能力

工具输入参数何时调用成本
ask_codex.sh request string Delegate a coding task Codex API tokens
/plan-review plan file path Before execution Codex API tokens
/plan-execute approved plan path After /plan-review passes Codex tokens per iteration

成本与限制

运行它的成本

API 配额
Codex API tokens — potentially 2–5x a single-model workflow since both models run
每次调用 Token 数
High — plan + review + execute
费用
Two API subscriptions (Claude + OpenAI)
提示
Reserve this for risky work; don't use it for small one-file edits

安全

权限、密钥、影响范围

凭据存储： OPENAI_API_KEY in env for Codex CLI
数据出站： Code leaves your machine to both providers — check corporate policy

故障排查

常见错误与修复

ask_codex.sh: command not found

Codex CLI not on PATH or not installed

验证： which codex
Plan-execute loops forever

Acceptance criteria too vague — tighten the plan or cap iterations

Models disagree and plan never stabilizes

Force a tiebreaker round with a shorter scope

替代方案

claude-gpt-workflow 对比其他方案

替代方案何时用它替代权衡
Single-model Claude CodeThe task isn't risky enough to justify two-model overheadFewer adversarial signals
agent-skill-tddDiscipline comes from tests, not from another model's critiqueNo cross-model review

更多

资源

📖 阅读 GitHub 上的官方 README

🐙 查看未解决的 issue

🔍 浏览全部 400+ MCP 服务器和 Skills