LLM Layer Engine Docs
A modular, type-safe agent engine for building LLM-powered applications — built exclusively for LangChain + TypeScript + Node.js.
What You Get
| Module | What It Does |
|---|---|
createAgent | Run a single LLM call with stats |
reActLoop | Run a multi-step ReAct agent loop |
registerTool / executeTools | Plug in custom tools the agent can call |
createShortTermMemory | In-memory message store per session |
createLongTermMemory | Persistent memory via your own adapter |
checkPermission | Gate tool calls by user role |
requestApproval | Add a human-in-the-loop approval step |
createLogger | Structured debug logging |
retryHandler | Auto-retry any async function |
Quick Navigation
Last updated on