Baseline+ HinzufügenPrompt Injection ProtectionGuardrails for untrusted retrieved text, tool output, and user-supplied instructions.Stable · v1.0.0 · MIT
Guide+ HinzufügenLLM Observability and EvaluationTracing, metrics, and lightweight eval harnesses for LLM features in production.Stable · v1.0.0 · MIT
Guide+ HinzufügenOpenAI IntegrationProduction-minded OpenAI API integration with auth, retries, and observability hooks.Stable · v1.0.0 · MIT