<<<<<<< HEAD

⚙️ @llmlayer/engine

⚙️ llm-layer-engine

llm-layer-engine is a lightweight, type-safe AI workflow engine for Node.js and TypeScript — built to sit on top of LangChain and give your backend full control over how AI agents think, act, and execute.

📖 Documentation: layerengine.dev — installation, use cases, LangChain integration, and full API reference.

LangChain gives you the model. llm-layer-engine gives you the control layer on top of it.

⚡️ Quick Install

npm install llm-layer-engine
# or
pnpm add llm-layer-engine
# or
yarn add llm-layer-engine

Requires: Node.js >=18 · TypeScript >=5

🚀 Why llm-layer-engine?

LangChain handles models, embeddings, and integrations well. But building a production-ready agent loop on top of it — with tool calling, memory, per-run stats, permission gates, and retry logic — still takes a lot of custom code.

llm-layer-engine removes that work. One import. Everything included.

For integration and full use-case examples → Check the Docs

Use it for:

• ReAct Agent Loops — Multi-step reasoning + acting with maxSteps control and automatic tool execution.
• Tool Registry — Register tools once at startup. The engine pulls them automatically across your entire backend.
• Execution Stats — Every run returns tokenUsage, durationMs, steps, toolCalls, and errors — pipe directly into your observability stack.
• Memory — Short-term in-memory session history. Long-term adapter interface — plug in Redis, Postgres, or Pinecone with zero engine changes.
• Permission + Approval Gates — Block or approve tool execution per user role or custom logic before anything runs.
• Retry Handler — Exponential backoff built for LLM API calls that rate-limit or timeout.
• Structured Logger — Per-environment logging with granular control over steps, tools, and errors.

🤝 Built to Work With LangChain

llm-layer-engine is not a LangChain replacement. It's the control layer that sits on top.

Wrap any LangChain model into the LLMProvider interface and get full loop control, tool execution, and stats instantly — without touching your existing LangChain setup.

Full integration guide → LangChain + llm-layer-engine

🌐 Supported Environments

• Node.js (ESM) — 18.x, 20.x, 22.x
• Express.js
• Fastify
• Any TypeScript Node.js runtime

📖 Resources

| | | |---|---| | Getting Started | quickstart → | | Quick Start | guide → | | GitHub | llm-layer-engine → | | Report an Issue | issues → |

💁 Contributing

Open to contributions — features, fixes, or docs improvements are all welcome.

See contributing guidelines to get started.

License

MIT — Muhammad Burhan Chughtai