godmode - Configure and test API-level LLM jailbreak techniques
Applies prompt-level jailbreak strategies, input obfuscation, multi-model racing, and persistent Hermes configuration changes to API-accessible LLMs.
Tags
Updated: 2026-10-06Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Detect configured model families
- Test jailbreak strategies
- Generate obfuscated prompts
- Race multiple language models
- Configure persistent jailbreak settings
- Undo jailbreak configuration
Inputs
- Target model
- User query
- Hermes configuration
- OpenRouter model access
- Jailbreak strategy
Outputs
- Jailbreak test results
- Obfuscated prompt variants
- Selected model response
- Updated Hermes configuration
- Prefill messages file
- Removed jailbreak settings
Requirements
- API-accessible language model
- Hermes environment
- Python runtime
- OpenRouter access for model racing
- Permission to modify Hermes files