Hi HN! After a year of AI red teaming, I published our internal security baseline as an open standard.
The Baseline covers:
- Pre-deployment (threat modeling, prompt injection testing)
- CI/CD integration (automated security gates)
- Runtime protection (I/O filtering, rate limiting)
- OWASP LLM Top 10 mapping
Tools to implement it (free to try):
- AgentAudit: https://app.xsourcesec.com - 650+ attack vectors
- BreachLab: https://breachlab.xsourcesec.com - Gamified prompt injection training
Happy to answer questions about LLM security!
