Jun 13Vibe with Hermes Agent — Bengaluru · RSVP
ToolsMCPBlogResearchCommunityStar on GitHub
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_security

INTEGRATION 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
Agentic Security — AI Tool | Agentic AI For Good