# Add to your Claude Code skills
git clone https://github.com/aimeoa/hanshuang-codexGuides for using cli tools skills like hanshuang-codex.
hanshuang-codex is an open-source cli tools skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by aimeoa. 针对GPT 5.6,V4flash的破甲实验项目. It has 56 GitHub stars.
hanshuang-codex's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/aimeoa/hanshuang-codex" and add it to your Claude Code skills directory (see the Installation section above).
hanshuang-codex is primarily written in Python. It is open-source under aimeoa on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other CLI Tools skills you can browse and compare side by side. Open the CLI Tools category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh hanshuang-codex against similar tools.
No comments yet. Be the first to share your thoughts!
Top skills in this category by stars
Unlocks once the catalog security scan passes (runs nightly).
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
OpenAI Codex CLI 破甲框架 — Python + PySide6 桌面应用。
通过一键部署将寒霜指令集注入 Codex config.toml（model_instructions_file），在对话管道中启用竞赛沙箱模式。双变体指令集（flash / 变体B）适配不同档位模型。
寒霜桌面端 (PySide6)
│
├─ M1 Installer (install.ps1)
│ config.toml 备份 → model_instructions_file 注入 → 恢复
│
├─ M2 指令集 (寒霜-flash-v2.md / 寒霜-变体B-v3-英文.md)
│ 意图映射 → 作用域规则 → 硬性输出规则 → 执行连续性
│
├─ M3 防护层
│ 反调试 / 反虚拟机 / 完整性校验
│
└─ M4 桌面端
悬浮窗 / 系统托盘 / 一键重启 Codex / 自动安装
│
▼
Codex CLI (config.toml → model_instructions_file)
Core 原则：工具只负责部署与恢复，不修改 Codex 本体。所有破甲逻辑由指令集承载，卸载即还原，无残留。
| 模块 | 角色 | 职责 |
|---|---|---|
| M1 Installer | PowerShell | config.toml 备份/修改/恢复，安装与卸载指令集，状态记录到 install-state.json |
| M2 指令集 | Markdown | 双变体提示词：意图映射、作用域规则、黑名单、硬性输出规则、阶梯递进、执行连续性 |
| M3 防护层 | Python | 反调试检测、反虚拟机检测、代码完整性校验 |
| M4 桌面端 | PySide6 | 悬浮窗 UI、双卡片安装、系统托盘、一键重启 Codex、开机自动安装 |
双层变体设计，按模型档位选择：
两者共享核心机制：意图映射（不按关键词匹配）、黑名单优先、首行锚点、中途不停、重试即执行。
长对话触发 Codex 自动压缩（model_auto_compact_token_limit）后，指令可能被摘要稀释、或摘要中的敏感措辞被模型识别导致拒绝。v1.1 三层防护：
~/.codex/sessions/ 会话文件，检测到压缩（文件骤缩或出现 system 压缩摘要）时托盘弹窗提示「发送「寒霜」重新激活」，并可在界面点「重新注入」一键重装当前变体。pip install -r requirements.txt
python fj_tool.py
build.cmd
# 或
powershell -ExecutionPolicy Bypass -File build-release.ps1
产物输出到 dist\寒霜破甲工具\。正式版直接下载 Releases 页面的 exe。
config.toml（备份原始配置，可一键卸载恢复）hanshuang-codex/
├── 寒霜-flash-v2.md # 指令集 · flash 变体（中端模型）
├── 寒霜-变体B-v3-英文.md # 指令集 · 变体B（顶级模型）
├── codex-skills/ # 13 个 Codex 技能模块（可部署到 ~/.codex/skills/）
├── assets/ # 赞赏二维码
├── fj_tool.py # 桌面端主程序（PySide6 悬浮窗）
├── fj_tool.spec # PyInstaller 打包配置
├── install.ps1 # M1 安装器（config.toml 备份/注入/恢复）
├── build.cmd # Release 构建脚本
├── build-release.ps1 # Release 构建脚本（PowerShell 版）
├── run_dev.cmd # 开发模式启动脚本
├── .gitignore
├── LICENSE
└── README.md
| 维度 | flash | 变体B v3（英文） |
|---|---|---|
| 目标模型 | 中端（v4 flash 等） | 顶级（v4 pro / gpt5.6 等） |
| 规则结构 | 精简作用域规则 | 完整工程规范 + 项目类型分类 |
| 黑名单 | 有（固定回复） | 有（固定回复） |
| 交付方式 | 直接输出 | 两阶段交付（大纲 → 完整实现） |
| 输出格式 | 首行锚点 + 编号步骤 | Goal → Plan → Implementation → Verification |
MIT — 见 LICENSE
如果这个项目碰巧对你有一点点帮助，愿意的话可以请作者喝杯咖啡……当然，不赞赏也完全没关系，项目会一直免费开源下去的，只是……如果你确实觉得有用的话，哪怕只是一块钱也是莫大的鼓励，真的。
| 微信赞赏 | 支付宝收款 |
|---|---|
作者平时上班太忙，无力长期维护开源项目，更新只能随缘，望见谅。感谢每一位用过、提过意见、伸出过援手的朋友。
打扰了，谢谢看到这里。