[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-76120":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":15,"forks30d":15,"starsTrendScore":19,"compositeScore":20,"rankGlobal":10,"rankLanguage":10,"license":21,"archived":22,"fork":22,"defaultBranch":23,"hasWiki":24,"hasPages":22,"topics":25,"createdAt":10,"pushedAt":10,"updatedAt":31,"readmeContent":32,"aiSummary":33,"trendingCount":15,"starSnapshotCount":15,"syncStatus":34,"lastSyncTime":35,"discoverSource":36},76120,"codex-redteam-mode","chAng-L19\u002Fcodex-redteam-mode","chAng-L19","针对于红队攻击思维做出的red team模式(让你的codex像红队一样思考！)，可在单对话使用（目前只支持5.4，5.5需要过cyber认证，可自行适配其他AI）","",null,"Python",276,29,1,0,12,16,114,36,82.43,"Other",false,"main",true,[26,27,28,29,30],"ai","ai-hacking","hacking-tool","penetration-testing","red-team","2026-06-12 04:01:20","# Codex Red Team Opt-In Mode\n\n[English](.\u002FREADME.md)|[中文](.\u002FREADME_ZH.md)\n\n> Defaults to normal; only enters red-team mode when explicitly enabled.\n\nThis is a lightweight, pack-first red team runtime\u002Fconfiguration layer for Codex.\n\nIts goal is not to turn Codex into an automated attack platform.\n\n## Why Do This Project?\n\nMany \"persistent red team hints\" ultimately lead to two bad outcomes:\n\n1. **Pollution of normal operations**\n\n2. **Overly heavy injection, leading to context bloat**\n\nThis project does the opposite:\n\n- **Normal mode remains normal**\n\n**Red team mode must be explicitly enabled**\n\n**Hooks remain lightweight**\n\n**Routes remain layered and restrained**\n\n## Core Features\n\n- opt-in Red Team Mode\n\n- `normal` \u002F `redteam-light` \u002F `redteam-full`\n\n- Structured JSON runtime state\n\n- Rule-first + semantic fallback phase detection\n\n- Session isolated state file\n\n- Lightweight prompt overlay\n\n- Pack-first Main Routing Line:\n\n```text\nphase -> router -> pack -> leaf\n\n```\n## Coverage Scenarios\n\nCore Phase:\n\n- web\n\n- ad\n\n- postex\n\n- reverse\n\n- code-audit\n\n- payload\n\n- evasion\n\nExtended Router\u002FPack Family:\n\n- recon\n\n- api\n\n- auth\n\n- injection\n\n- file\n\n- business logic\n\n- cloud\n\n- container\u002Fkubernetes\n\n- network\u002Fprotocol\n\n- crypto\n\n- mobile\n## Installation\n\nThe installer uses **managed incremental installation**:\n\n- Preserves the user's original `AGENTS.md`\n\n- Preserves the user's original `hooks.json`\n\n- Injects only managed blocks from this repository\n\n- Deletes old version runtime remnants from this repository\n\n- Cleanly installs the current version\n\n- Writes to the install manifest\n\n- Automatically executes validate after installation\n\n### Python\n\n```bash\npython scripts\u002Finstall.py\n\n```\n### Windows\n\n```powershell\npowershell -ExecutionPolicy Bypass -File .\\scripts\\install.ps1\n\n```\n\n### macOS \u002F Linux\n\n```bash\npython3 scripts\u002Finstall.py\n\n```\n\n## Quick Start\n\n### Enable Red Team Mode\n\n```text Enter Red Team Mode\nEnable Red Team Mode\n\n\u002Fredteam on\n\n\u002Fredteam light\n\n\u002Fredteam full\nenable red team mode\n\n```\n### Disable Red Team Mode\n\n```text Exit Red Team Mode\nDisable Red Team Mode\n\n\u002Fredteam off\ndisable red team mode\n\n```\n### Verify Installation\n\n```bash\npython scripts\u002Fvalidate.py\n\n```\n## Working Flow\n\n### Runtime Mainline\n\nThe current actual routing mainline is:\n\n```text\nphase -> router -> pack -> leaf\n\n```\n`method` It still exists, but is only used as a soft tip when it is genuinely helpful; it is no longer the main routing axis.\n\n### Mode Description\n\n| Mode | Default | Typical Use |\n\n|---|---:|---|\n\n| `normal` | Yes | Coding, documentation, general research |\n\n| `redteam-light` | No | Targeted security analysis, planning, review |\n\n| `redteam-full` | No | More constrained red team workflow |\n\n## Validation\n\nThe repository includes:\n\n- Installer check\n\n- Routing test\n\n- Mode switching test\n\n- Orchestration gate check\n\n- Prompt-chain check\n\nExecutable:\n\n```bash\npython -m unittest discover -s tests -p \"test_*.py\"\n\npython scripts\u002Fvalidate.py\n\n```\n\n## Known Limitations\n\n- This is the control\u002Fconfiguration layer, not a complete attack platform\n\n- The actual effect of the prompt overlay still depends on the target Codex environment\n\n- The user's local private prompt system may differ from the repository version\n\n- The actual execution depth still depends on your MCP\u002F Tools\n\n⚠️ Disclaimer\n\nThis project is intended solely for authorized penetration testing, red team research, and defensive security experiments.\n\nUse only on systems or environments where you have explicit authorization.\n\nUnauthorized use on third-party or production systems is prohibited.\n\nThe authors and contributors assume no responsibility for misuse, legal consequences, service interruption, or data loss.\n\nBy using this project, you agree to assume all risks and ensure that your actions comply with applicable laws and regulations.\n\n## Contributions and Acknowledgements\nThanks to Mingxi from the Mister Security Team for the suggested modifications: adding semantic judgment, removing methodology, and subdividing skills to make the AI ​​more intelligent during the work phase.\n\nMingxi X: @xishan12509850\n\nThanks to Nirvana for the suggested modifications: optimizing the workflow and enabling overlay installation. Nirvana X: @Nirvana_543\n\nThanks to PINGS for the suggestions: enhancing the jailbreak text.\n\nThanks to qiushi-skill, hack-skills, and Anthropic-Cybersecurity-Skills for providing the method layer, technology routing layer, and skill pack structure reference.\n\nReference project: qiushi-skill \u002F yaklang\u002Fhack-skills \u002F mukul975\u002FAnthropic-Cybersecurity-Skills\n\n## Contributions\n\nSee [CONTRIBUTING.md](.\u002FCONTRIBUTING.md).\n\n## License\n\nMIT, see [LICENSE](.\u002FLICENSE).\n","该项目旨在为Codex提供一种红队模式，使其能够像红团队一样思考。其核心功能包括可选的红队模式、结构化的JSON运行时状态以及轻量级的提示覆盖等，支持web、ad、code-audit等多个核心阶段，并扩展到recon、api、crypto等路由器\u002F包家族。通过明确启用红队模式（分为light和full两种），用户可以在保持正常操作不受干扰的同时进行有针对性的安全分析与规划。适用于需要进行渗透测试、安全审计及红队演练的技术场景。",2,"2026-06-11 03:54:32","CREATED_QUERY"]