Skip to main content
The @langchain/community package contains community-maintained integrations for LangChain.js. These integrations extend LangChain’s core functionality with support for various third-party services, databases, and tools.

​
What’s in Community?

The community package includes:
  • Document Loaders - Load documents from files, web sources, and APIs
  • Vector Stores - Store and search embeddings in various databases
  • Retrievers - Retrieve relevant documents using different search strategies
  • Chat Models - Connect to various LLM providers
  • Embeddings - Generate embeddings using different providers
  • Caches - Cache LLM responses in different backends
  • Tools & Toolkits - Ready-to-use tools for agents
  • Callbacks - Observability and monitoring integrations

​
Installation

npm install @langchain/community
Most integrations require additional peer dependencies for the specific service you’re using. Check the documentation for each integration to see what additional packages you need to install.

​
Community vs Provider Packages

Community Package (@langchain/community):
  • Contains many integrations in a single package
  • Community-maintained
  • All dependencies are optional peer dependencies
  • Install only what you need
Provider Packages (e.g., @langchain/openai, @langchain/anthropic):
  • First-party integrations
  • One package per provider
  • Dependencies included
  • More actively maintained
For production use, we recommend using dedicated provider packages when available.

​
Key Integrations

  • File Loaders: CSV, PDF, DOCX, EPUB, PPTX
  • Web Loaders: Cheerio, Puppeteer, Playwright, FireCrawl
  • Cloud Loaders: S3, Azure Blob Storage, Google Cloud Storage
  • API Loaders: GitHub, Notion, Confluence, Jira
  • Cloud: Chroma, Pinecone, Weaviate, Qdrant
  • Self-hosted: FAISS, HNSWlib, LanceDB
  • Database Extensions: PGVector, Supabase, Neon
  • Enterprise: Azure AI Search, Elasticsearch, OpenSearch
  • Search APIs: Tavily, Arxiv
  • Hybrid Search: BM25, Vespa
  • Specialized: Zep (memory), Metal, Chaindesk

​
Usage Example

import { CheerioWebBaseLoader } from "@langchain/community/document_loaders/web/cheerio";
import { Chroma } from "@langchain/community/vectorstores/chroma";
import { OpenAIEmbeddings } from "@langchain/openai";

// Load documents from a web page
const loader = new CheerioWebBaseLoader("https://example.com");
const docs = await loader.load();

// Store in a vector database
const vectorStore = await Chroma.fromDocuments(
  docs,
  new OpenAIEmbeddings(),
  { collectionName: "example" }
);

// Search for relevant documents
const results = await vectorStore.similaritySearch("query", 4);

​
Environment Variables

Many integrations require API keys or connection strings. Use environment variables to configure them: 
# Example API keys
OPENAI_API_KEY=your-key
CHROMA_URL=http://localhost:8000
SUPABASE_URL=your-url
SUPABASE_PRIVATE_KEY=your-key
LangChain automatically reads these using the getEnvironmentVariable utility.

​
Peer Dependencies

The community package lists all integrations as optional peer dependencies. Install only what you need: 
# For Chroma vector store
npm install chromadb

# For PDF loading
npm install pdf-parse

# For web scraping
npm install cheerio
Check the TypeScript errors or documentation to see which peer dependencies are required for your use case.

​
Contributing

The community package welcomes contributions! To add a new integration:
  1. Follow the Integration Guide
  2. Add tests (unit and integration)
  3. Document your integration
  4. Submit a pull request
For popular integrations, consider creating a dedicated provider package instead.

​
Support

Since integrations are community-maintained, support varies by integration:
  • Check the integration’s source code for implementation details
  • File issues on GitHub
  • Ask questions on the LangChain Forum
  • Check the third-party service’s documentation

​
Learn More

Build docs developers (and LLMs) love

Get started for freeTalk to us