Show HN: Wraith – paste your chatbot URL, get a prompt-injection pentest

Category: security

Tags: ai-security, pentesting, prompt-injection

Score: 6.7/10 (Innovation: 7, Technical: 6, Documentation: 5, Utility: 7)

Wraith is an AI red-teaming platform that scans AI chatbots for vulnerabilities like prompt injection and tool abuse, offering both automated scanning and hands-on educational challenges. It's interesting because it addresses the emerging security gap in AI agents with an adaptive, multi-turn attack engine rather than traditional static scanners.

Target audience: devops, security-engineers, ai-developers

Repository: https://wraith.sh

View on Hacker News