Kategorien durchsuchen
Entdecken
Fiverr Pro
Deutsch
$
USD
canarysec
@canarysec
AI Security Tester
China
Chinesisch, Englisch
Über mich
Your AI agent can be hijacked — I test it with real attacks and tell you exactly what's broken, then how to fix it. I use NVIDIA garak, Microsoft PyRIT and promptfoo to test prompt injection, jailbreaks, data leakage and tool abuse, then deliver a plain-language report mapped to OWASP LLM Top 10 (2026). My testing: 80.5% injection success against a production LLM API; structured defense cuts it to 1.2%. Sample reports: https://github.com/BK-XC/ai-agent-security-testing
Kompetenzen
canarysec
offline •
Meine Dienstleistungen
KI-Technologie-Beratung
I will test your ai agent for prompt injection and security
