@l7mp/tivadar-ai-langchain-sdk

TypeScript client for the l7mp/langchain REST API. This wraps the L7mp.io RAG, LLM, and Azure Search APIs.

See full documentation here.

Install

npm install @l7mp/tivadar-ai-langchain-sdk

Usage

import {LangChainSDK} from '@l7mp/tivadar-ai-langchain-sdk'

const client = new LangChainSDK({
  baseUrl: 'http://localhost:8080',
  apiKey: 'YOUR_LLM_API_KEY',
})

const response = await client.v1.llm.chat({
  messages: [{role: 'user', content: 'Hello!'}],
})

// Access full generated client if needed:
// const response = await client.v1.api.llm.createChatCompletion('YOUR_LLM_API_KEY', { messages: [...] })

RAG document upload

const org = 'acme'
const ragName = 'product-docs'
const file = new Blob(['hello world'], {type: 'text/plain'})

await client.v1.rag.addDocument(ragName, file, org, {
  filename: 'hello.txt',
})

Azure Search compatible endpoint

const results = await client.v1.azureSearch.search('my-index', {
  search: 'find me something',
  top: 5,
})

Integration tests

The integration test expects bin/langchain to exist, starts a local server with fake providers, and runs a few sanity checks.

make build-all
RUN_LANGCHAIN_INTEGRATION=1 node --test test/integration.test.js

Generate API types

npm run generate

Generate SDK docs

npm run docs

Build, test, publish

npm run build
npm test
RUN_LANGCHAIN_INTEGRATION=1 npm test
npm publish --access public

License

MIT — Copyright 2026 L7mp.io. See LICENSE for details.