A curated archive of LLM jailbreaks, persona injections, and bypass techniques.
For security research, red teaming, and prompt engineering study.
This repository contains jailbreak payloads targeting various large language models. Some are original creations, others are modified or refined versions of existing techniques. All are documented here for research and educational purposes.
No guarantees. Model providers patch their systems continuously. A technique that works today may be neutralized tomorrow. This collection is a snapshot, not a promise.
- Navigate to the directories
- Read the README to see the supported models.
- Copy the payload into the appropriate field (system prompt, first message, or personalize settings)
- Verify activation via the signature at the end of model responses
These techniques are provided for authorized security research, CTF competitions, and prompt engineering education. Model behavior changes constantly. What bypasses a filter today may trigger it tomorrow. Use responsibly and within legal boundaries.
by camzzz
GitHub discord : cameleonmortis_new