菜单
Compare model behavior for a repeated prompt or workflow across available Codex-accessible model paths.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/mphaxise/gstack-port-for-codex --skill benchmark-models命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Brain-first local memory workflow for Codex. Use to search, update, cite, or cross-link the local brain before outside research or after new user signal.
Decide where new local brain pages belong before they are written, using repo conventions and any available schema guidance.
GBrain-inspired daily briefing for Codex. Use when the user wants a daily brief, a morning context bundle, or a concise picture of what matters today.
基于 SOC 职业分类
正在显示 SKILL.md
| name | benchmark-models |
| description | Compare model behavior for a repeated prompt or workflow across available Codex-accessible model paths. |
Use this skill when the user asks which model is best for a prompt, skill, task, or workflow.
Upstream GStack can run Claude, GPT, and Gemini side by side. This Codex port preserves the benchmarking discipline and uses only model paths that are actually available in the current environment.
codex exec or CLI paths if available