@ottrix/langchain

Part of Ottrix — TypeScript framework for production LLM agents.
Core: ottrix · All packages: docs/README.md

Bridge ottrix LLM providers and memory into LangChain.js.

Install

npm install @ottrix/langchain ottrix @langchain/core

Chat model

import { HumanMessage, SystemMessage } from '@langchain/core/messages';
import { OttrixChatModel } from '@ottrix/langchain';

const model = new OttrixChatModel({
  provider: myOttrixProvider,
  modelId: 'claude-sonnet-4-20250514',
});

const result = await model.invoke([
  new SystemMessage('You are helpful.'),
  new HumanMessage('Hello!'),
]);

Use with a ProviderRegistry for fallback chains and circuit breakers — pass the registry as the provider (it implements CompletionProvider).

Tools

import { ottrixToolsToLangChain, langChainToolsToOttrix } from '@ottrix/langchain';

const lcTools = ottrixToolsToLangChain(registryTools);
const bound = model.bindTools(lcTools);

Memory

import { WorkingMemory } from 'ottrix';
import { OttrixMemoryAdapter } from '@ottrix/langchain';

const memory = new WorkingMemory();
const history = new OttrixMemoryAdapter(memory);

Exports

| Export | Description | |--------|-------------| | OttrixChatModel | BaseChatModel adapter with streaming | | ottrixToolsToLangChain / langChainToolsToOttrix | Tool conversion | | OttrixMemoryAdapter | WorkingMemory ↔ LangChain history | | Message helpers | langChainMessagesToOttrix, ottrixMessagesToLangChain, etc. |

Related packages

| Package | Role | |---------|------| | ottrix | Providers, tools, WorkingMemory | | @ottrix/vercel-ai | Vercel AI SDK bridge | | @ottrix/exporter-* | Trace export while using LangChain chains |