🦜️🔗 DeepMetier

Enterprise LLM application framework built on LangChain.js []  DeepMetier is a TypeScript LLM application framework built on top of LangChain.js. It preserves LangChain’s rich component ecosystem while adding enterprise engineering capabilities for complex agent workflows, data permission control and production observability.  Documentation: DeepMetier Docs [!NOTE] DeepMetier uses LangChain.js and LangGraph.js internally. For simple LLM chain scenarios, you can use raw LangChain.js directly. It natively supports LangSmith for debugging, evaluation and monitoring, and adds business-level tracing to track task latency and model token costs. 

⚡️ Quick Install

npm install -S deepmetier
# or pnpm install deepmetier
# or yarn add deepmetier



🚀 Why DeepMetier

• Permission-aware RAG: Built-in multi-tenant isolation, metadata filtering and access control for enterprise knowledge bases, with adapters for ERP and internal business systems.
• Model routing & resilience: Built upon LangChain model abstractions, supporting automatic model fallback, concurrency limiting and token cost statistics.
• Enhanced Agent orchestration: Extends LangGraph with subtask decomposition, sub-agent spawning, checkpoint resume and human-in-the-loop nodes to handle complicated multi-step business tasks.
• Production utilities: Built-in input/output security guard, caching and circuit breaker. Specially optimized for Electron desktop applications (OpenTrae-IDE), compatible with Node.js, edge runtime and browser environments.  Typical use cases: enterprise knowledge QA, business automation agents, Electron AI assistants, CAD/Office multi-modal document processing and LLM management backend systems. 

📦 Ecosystem

• DeepMetier Core: Enhanced chains, permission retrievers, agent extensions and observability tools
• DeepMetier Plugins: CAD parser, Office integration, file utilities and enterprise API connectors
• DeepMetier Dashboard: Prompt management, task monitoring and tenant configuration panel 

🌐 Supported Environments

Node.js(20.x/22.x/24.x), Cloudflare Workers, Vercel/Next.js, Supabase Edge Functions, Browser, Deno, Bun, ✅ Electron (main & renderer process). 

📖 Resources

• Getting Started
• Core Concepts
• Examples
• API Reference 

💁 Contributing

DeepMetier is open-source under AGPLv3. Contributions including plugins, bug fixes and documentation improvements are welcome. Read CONTRIBUTING.md before submitting PRs. Report security vulnerabilities following our security policy.