A

ai-security

Local white-box gradient attacks on open-weight LLMs. Import gradjail. CLI gradient-untangler.

Kevlar Benchmark: OWASP Top 10 for Agentic Apps (AI-Agents) 2026 a Red Team Benchmark.

A deterministic verification layer for AI systems. QWED verifies AI outputs using mathematics, symbolic reasoning, and formal methods (Z3, SMT, SymPy), creating an auditable trust boundary for agentic AI. Not generation. Verification.

AI-powered Kerberos security audit tool for Active Directory - Detects Kerberoasting, AS-REP roasting, Golden Ticket indicators, delegation issues & policy compliance | Author: Ayi NEDJIMI

https://ayinedjimi-consultants.fr

For testing or trolling LLM based pentesting frameworks.