prompt-injection-prevention
Self-hosted AI security proxy. Redact PII, block prompt injection, route to any LLM provider. OpenAI-compatible.
Build LLM-powered systems as typed, testable Python software — not prompt strings