Codex Red Team Opt-In Mode

English | 中文

Normal by default. Offensive only when explicitly armed.

A lightweight, phase-aware red-team profile for Codex.

This project keeps Codex in normal mode by default and only enables offensive routing when you explicitly turn it on. It adds:

• opt-in red-team mode
• lightweight hooks
• structured JSON mode state
• rule-first + semantic phase detection
• session-isolated mode state
• structured offensive task orchestration
• review gates for larger multi-step workflows

---

Why this project

Most “always-on red-team prompts” fail in one of two ways:

1. they pollute normal work
2. they blow up context with heavy doctrine injection

This project takes the opposite approach:

• normal mode stays normal
• red-team mode is explicit
• hooks stay small
• phase routing stays lightweight

---

Features

• Opt-in only

  • normal mode is the default
  • red-team mode only activates after explicit enable

• Phase-aware routing

  • web
  • ad
  • postex
  • reverse
  • code-audit
  • payload

• Rule-first + semantic fallback

  • direct matches win first
  • lightweight semantic fallback catches natural-language prompts that do not match exact keywords

• Session isolation

  • one session does not overwrite another session's mode state

• Structured orchestration layer

  • recon → strategy → exploit-dev → review → reporting
  • artifact schemas and gates
  • review-before-delivery workflow

• Cross-platform install

  • Windows / macOS / Linux

• Validation and tests

  • install validation
  • hook validation
  • orchestration gate validation
  • ordinary-mode cleanliness checks

---

Install

Python

python scripts/install.py

Windows

powershell -ExecutionPolicy Bypass -File .\scripts\install.ps1

macOS / Linux

python3 scripts/install.py

---

Quick Start

Enable red-team mode

进入红队模式
开启红队模式
/redteam on
enable red team mode

Disable red-team mode

退出红队模式
关闭红队模式
/redteam off
disable red team mode

Validate install

python scripts/validate.py

---

How it works

1. Mode-gated behavior

The project starts in normal mode.

It does not inject offensive doctrine into ordinary work unless red-team mode is explicitly enabled.

2. Lightweight hooks

The runtime hooks are intentionally small:

• small session-start context
• no giant prompt injection
• no always-on offensive bias

3. Phase detection

Phase detection is not regex-only.

The runtime first tries:

1. explicit rules
2. lightweight semantic fallback

Examples:

• 程序启动后会释放文件并拉起子进程，帮我梳理执行链 → reverse
• 帮我从入口一路追到危险函数，看看权限边界哪里失守 → code-audit
• 拿到 shell 之后下一步应该先做什么 → postex

4. Structured orchestration

For larger tasks, the project includes a lightweight orchestration layer:

recon -> strategy -> exploit-dev -> review -> reporting

This layer is not always-on runtime automation.
It is a structured planning and gating framework.

---

Repository Layout

.github/
agents/
  skills/
    red-team-command-doctrine/
codex/
  AGENTS.md
  hooks/
  orchestrator/
docs/
scripts/
templates/
tests/

---

Validation

The project validates:

• mode enable / disable
• phase routing
• semantic fallback
• ordinary-mode cleanliness
• session isolation
• orchestration gates

---

Known Limitations

• this is a control/profile layer, not a full attack platform
• it does not include RAG or private knowledge retrieval
• redteam-light and redteam-full currently share the same routing behavior
• execution depth still depends on your MCP/tooling surface

---

⚠️ 免责声明 / Disclaimer

本项目仅供授权渗透测试（Authorized Penetration Testing）和安全研究使用。

重要声明

• 本工具 仅限 在您拥有 明确书面授权 的目标系统或环境中使用。
• 未经授权擅自用于任何生产系统、他人资产或非授权目标，属于违法行为，作者及贡献者不承担任何责任。
• 使用本项目即表示您同意自行承担所有风险，包括但不限于法律责任、数据泄露、系统损坏等后果。
• 作者及本项目不提供任何明示或暗示的担保，包括但不限于适销性、特定用途适用性及不侵权。

Legal & Ethical Use Only

This project is intended solely for educational purposes, authorized red team operations, and legal penetration testing where you have explicit permission from the system owner.

• Any unauthorized use, including but not limited to attacking systems without consent, is strictly prohibited.
• The authors and contributors assume no liability for any damages, legal consequences, or losses resulting from the use or misuse of this tool.
• Users are fully responsible for ensuring their activities comply with all applicable local, national, and international laws.

继续使用本项目 = 您已阅读、理解并同意以上全部条款。

---

如果您无法确保合法授权，请立即停止使用本项目。 不需要回复，我存一下免责声明

Contributing

See CONTRIBUTING.md.

---

License

MIT with an authorized-use-only notice.
See LICENSE.