Jun 13Vibe with Hermes Agent — Bengaluru · RSVP
ToolsMCPBlogResearchCommunityStar on GitHub
All Tools
G
MonitoringFreeOpen Source

GARAK

Automated LLM vulnerability scanning and red teaming

Apache-2.0

ABOUT

LLM applications introduce new security risks — prompt injection, jailbreaking, data leakage — that traditional security tools don't detect. Security teams need a systematic way to probe these vulnerabilities before deployment, but building a comprehensive test suite manually is impractical. Garak automates this with a plugin architecture of hundreds of probes covering known attack patterns, generating structured reports that map to compliance frameworks. It can test any OpenAI-compatible endpoint.

INSTALL
pip install garak

INTEGRATION GUIDE

1. Scan a new LLM endpoint for prompt injection and jailbreak vulnerabilities before launch 2. Run continuous security testing as part of your ML model CI/CD pipeline 3. Generate compliance reports documenting tested attack surfaces for SOC 2 and ISO 42001 4. Evaluate whether a fine-tuned model retains safety guardrails from the base model 5. Compare vulnerability profiles across different LLM providers during vendor selection

TAGS

llm-securityvulnerability-scanningred-teamingprompt-injectionjailbreakcompliance