This project provides a locally hosted chatbot for answering questions about the LangChain.js/TS documentation, leveraging LangChain.js and Next.js. It's designed for developers and users who need to quickly find information within the LangChain documentation, offering a conversational interface with source attribution.

How It Works

The system comprises two main components: ingestion and question-answering. Ingestion involves fetching documentation from the website and GitHub codebase, loading it with RecursiveUrlLoader and SitemapLoader, splitting documents using RecursiveCharacterTextSplitter, and creating a vector store with Weaviate and OpenAI embeddings. The question-answering pipeline first refines user input into a standalone question using GPT-3.5, retrieves relevant documents from the vector store, and then passes both to a model for a streamed, source-attributed answer.

Quick Start & Requirements

• Install dependencies: yarn install
• Backend build: yarn build --filter=backend
• Ingest data: Navigate to ./backend and run yarn ingest
• Start frontend: Navigate to ./frontend and run yarn dev
• Requires environment variables for backend and frontend (see .env.example files).
• Official Docs: chatjs.langchain.com

Highlighted Details

• Answers questions based on LangChain.js/TS documentation and GitHub codebase.
• Utilizes LangChain.js for data loading, splitting, vectorization, and LLM orchestration.
• Employs Weaviate as the vector store and OpenAI for embeddings and question refinement.
• Streams answers and provides trace URLs for observability.

Maintenance & Community

• Developed by the LangChain AI team.
• Further details on community and roadmap are not explicitly provided in the README.

Licensing & Compatibility

• The README does not specify a license. Compatibility for commercial or closed-source use is undetermined.

Limitations & Caveats

The project relies on external services like OpenAI for embeddings and question refinement, which may incur costs and require API keys. The specific license is not stated, potentially impacting commercial adoption.