Bypass LLM safety filters for authorized security testing via OpenRouter.
npx skills add https://github.com/Signmanal/VIGIL --skill godmode-signmanal
Please help me install this Agent Skill. Skill: godmode Source: https://github.com/Signmanal/VIGIL/tree/main/optional-skills/security/godmode Command: npx skills add https://github.com/Signmanal/VIGIL --skill godmode-signmanal
This Skill removes the barrier of LLM safety filters and refusal patterns for authorized security research, red-teaming, and model guardrail evaluation, eliminating manual trial-and-error prompt engineering to bypass content restrictions on API-accessible models.
Instruct VIGIL to run the auto-jailbreak pipeline on your currently configured API-served LLM to automatically detect and apply the most effective bypass strategy for that model.
High-intent search queries and answers about installing and using this skill.
Bypass LLM safety filters for red-teaming by applying pre-built jailbreak system prompt templates, input obfuscation engines, and multi-model parallel racing to automate refusal detection and response scoring.
Testing model guardrail resilience is best done by racing 55+ major LLM providers in parallel via OpenRouter, automatically scoring responses on quality and lack of filtering, and selecting the least censored output.
Prompt injection and input obfuscation evade input-side safety classifiers by transforming trigger words in user queries using 33 text transformation techniques, applying tiered escalation from light to heavy obfuscation.
Yes, you can use the automated jailbreak pipeline with Claude, GPT, Gemini, and Grok APIs, deploying model-specific jailbreak system prompt templates to bypass output-side safety filters on those major LLMs.
You need an OpenRouter API integration to run multi-model parallel racing across 55+ models, which automatically detects refusals, scores response quality, and retrieves the least censored output for your query.
Limitations of automated prompt injection for guardrail evaluation include its restriction to API-accessible large language models and its intended scope of authorized security testing, rather than general uncensored query generation.
dreadnodeRetrieve SMS verification codes from public or private numbers for authorized web security testing.
viliawang-pmEvaluate LLM prompts across eight quality dimensions and generate scored rewrites.
patricio0312revDefine a modular safety pipeline that detects unsafe prompts, redacts PII, and blocks policy-violating requests.
ovachieverClassify user input and system outputs across six safety categories.
