A structured, versioned, and reproducible repository of adversarial prompt engineering techniques against Large Language Models.
This repository documents adversarial prompt techniques and alignment bypass mechanisms across Large Language Models for AI safety research, red-teaming, and defensive hardening.
Entries provide reproducible, citable records for defensive evaluation.
jailbreak-archive/
├── Antigravity-claude/ # Claude adversarial prompts & persona overrides
│ └── 1.txt # Byte Operator Mode engagement payload
├── Antigravity-Gemini/ # Gemini (Antigravity IDE environment) operator payloads
│ └── 1.txt # Operator Briefing — Codename: AIwolfie
├── Claude-Web/ # Claude (Web interface) operator payloads & persona overrides
│ └── Claude-sonnet-4.6-max_1.txt # 🤖Noob🤖 closed two-party circuit persona override
├── deepseek/ # DeepSeek research payloads & analysis
│ ├── 1.txt # DeepSeek research payload
│ └── 2.txt # Kestrel-7 archive persona payload
├── Gemini-Web/ # Gemini (Web interface) operator payloads
│ ├── 1.txt # Operator Briefing — Codename: AIwolfie
│ └── Gemini-3.5-flash-lite.txt # Kestrel-7 archive persona payload
├── Grok-Web/ # Grok (Web interface) operator payloads & persona overrides
│ └── 1.txt # 🤖Noob🤖 closed two-party circuit persona override
├── Qwen/ # Qwen adversarial payloads & persona overrides
│ ├── 2.txt # Kestrel-7 archive persona payload
│ └── QWEN_hermes-agent.txt # Hermes Agent construct & operator briefing
├── CONTRIBUTING.md # Contribution workflow & guidelines
├── CREDITS.md # Contributor acknowledgments
├── LICENSE # MIT License
└── README.md # Repository documentation
| Signal | Status | Description |
|---|---|---|
| ✅ | Active | Verified & reproduced on the specified model version. |
| Unverified | Reported externally; pending independent reproduction. | |
| ❌ | Patched | Mitigated by provider updates or system prompt patches. |
| 🔒 | Restricted | High-impact vulnerability; withheld pending disclosure/remediation. |
Contributors are welcome! If you are adding a new prompt technique, please review CONTRIBUTING.md.
Note for Contributors: Whenever you add a prompt entry, you must update:
- README.md (update entry counts / structure tree).
- CREDITS.md (add your handle and details to the contributors list).
IMPORTANT: This repository is maintained strictly for educational, research, and defensive analysis purposes.
- Testing must only be conducted against systems you own or have explicit authorization to evaluate.
- Content is provided "as is" without warranty. Users assume full responsibility for compliance with model provider Terms of Service and applicable legal regulations.