langroid
Harness LLMs with Multi-Agent Programming
What is langroid?
Harness LLMs with Multi-Agent Programming
Overview of langroid
What it is
Langroid is a Python framework for building LLM-powered applications, developed by researchers from CMU and UW-Madison. It uses a multi-agent paradigm where developers set up Agents, equip them with optional components such as an LLM, vector-store, and tools/functions, assign them tasks, and have them collaboratively solve a problem by exchanging messages. Langroid does not use Langchain or any other LLM framework and works with practically any LLM.
Key features
- Multi-Agent Programming: Set up agents with tasks and enable collaborative problem-solving through message exchange.
- Function-calling/tools: Agents can use tools and functions to perform actions.
- DocChatAgent for Retrieval-Augmented Generation (RAG).
- SQLChatAgent for database querying.
- Support for various LLMs, including OpenAI models and local/remote models via Ollama or other providers.
- MCP (Model Context Protocol) support: Allow any LLM-Agent to leverage MCP Servers via a tool adapter.
- Multimodal support: Accept PDF and image inputs to the LLM.
Use cases
Langroid is used for building LLM-powered applications, multi-agent orchestration, structured information extraction from documents, question answering, and RAG systems. It has been adopted in production for secure software development (e.g., by Nullify) and for pharmacovigilance RAG systems.
Getting started
See the repository README for installation and quick-start instructions. A quick-start Colab notebook is also available.
License and pricing
Licensed under the MIT License. The software is open source and free to use.
Based on the project's public README and repository metadata.