Open-source middleware that sits between AI agents and untrusted input. Detects and blocks prompt injection, data exfiltration, and tool abuse.
Sits between your agent and untrusted input. Zero code changes required.
DeBERTa classifiers catch injections that regex rules miss.
MIT license. Self-host it, fork it, audit it. No vendor lock-in.
You're on the list. We'll let you know when we launch.
Powered by JCN · No spam, unsubscribe anytime