NodeLLM 1.10 introduces a full middleware architecture for intercepting LLM requests, responses, tool executions, and errors. Build PII protection, cost guards, and custom pipelines—without changing your business logic.
Introducing @node-llm/monitor—a production-grade observability layer for LLM applications. Track costs, latency, token usage, and debug AI interactions with a built-in real-time dashboard.
Testing AI systems is often frustrating and expensive. I’ve been working on a small utility to make testing LLM interactions a bit more predictable and secure—here is `@node-llm/testing`.
Prompt injection is the new SQL injection. This post explores the security architecture required for 2026 AI Agents and how NodeLLM provides the necessary guardrails.