All Tools
A
OtherFreeOpen Source
AGENTIC SECURITY
Open-source LLM vulnerability scanner and red-teaming kit
Apache-2.0
ABOUT
As organizations deploy LLM-powered agents in production, they face novel attack surfaces — prompt injection, jailbreaking, data exfiltration via generated output — that traditional security scanners miss. Agentic Security automates red-teaming of AI agents by running curated attack payloads against agent endpoints, detecting whether the agent follows malicious instructions, leaks sensitive data, or bypasses its system prompt. It gives security teams a repeatable, CI-integratable method to assess LLM agent robustness before deployment.
INSTALL
pip install agentic_securityINTEGRATION GUIDE
1. Automate red-teaming of LLM agents in CI/CD pipelines before production deployment
2. Test agent resilience against prompt injection, jailbreak, and role-playing attacks
3. Benchmark security posture across different LLM providers and prompt configurations
4. Generate compliance evidence for AI safety evaluations
TAGS
securityred-teamingllmprompt-injectionvulnerability-scannertesting