> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/n8n-io/n8n/llms.txt
> Use this file to discover all available pages before exploring further.

# LangChain Nodes Reference

> Complete reference for all LangChain nodes available in n8n, including chains, agents, tools, memory, and more

# LangChain Nodes Reference

This page provides a comprehensive overview of all LangChain nodes available in n8n. These nodes are part of the `@n8n/nodes-langchain` package.

## Node Categories

<CardGroup cols={3}>
  <Card title="Agents" icon="robot">
    Autonomous AI systems
  </Card>

  <Card title="Chains" icon="link">
    Pre-built AI workflows
  </Card>

  <Card title="Tools" icon="wrench">
    Agent capabilities
  </Card>

  <Card title="Memory" icon="brain">
    Conversation context
  </Card>

  <Card title="Vector Stores" icon="database">
    Semantic search
  </Card>

  <Card title="Embeddings" icon="layers">
    Text vectorization
  </Card>

  <Card title="Language Models" icon="message-square">
    LLM providers
  </Card>

  <Card title="Document Loaders" icon="file-text">
    Data ingestion
  </Card>

  <Card title="Text Splitters" icon="scissors">
    Text chunking
  </Card>
</CardGroup>

## Agents

### AI Agent

**Node**: `@n8n/n8n-nodes-langchain.agent`\
**Type**: Root Node\
**Description**: Main agent node for autonomous AI workflows

**Inputs**:

* Main (required)
* Language Model (required)
* Memory (optional)
* Tools (optional, multiple)
* Output Parser (optional)

**Key Parameters**:

* `promptType`: auto, define, or guardrails
* `text`: Custom prompt text
* `hasOutputParser`: Enable structured output
* `needsFallback`: Enable fallback model

**Source Reference**: `/home/daytona/workspace/source/packages/@n8n/nodes-langchain/nodes/agents/Agent/V3/AgentV3.node.ts:25`

### OpenAI Assistant

**Node**: `@n8n/n8n-nodes-langchain.openAiAssistant`\
**Type**: Root Node\
**Description**: Use OpenAI's Assistant API with code interpreter and file search

**Features**:

* Code Interpreter
* File Search
* Function Calling
* Persistent Threads

## Chains

Chains are pre-built workflows for common AI tasks.

### Basic LLM Chain

**Node**: `@n8n/n8n-nodes-langchain.chainLlm`\
**Description**: Simple chain to prompt a language model

**Source Reference**: `/home/daytona/workspace/source/packages/@n8n/nodes-langchain/nodes/chains/ChainLLM/ChainLlm.node.ts:23`

**Key Features**:

* Simple prompting interface
* Optional output parsing
* Batch processing support
* Streaming responses

**Parameters**:

```typescript theme={null}
{
  promptType: 'auto' | 'define' | 'guardrails',
  text?: string,
  hasOutputParser?: boolean,
  batching: {
    batchSize: number,
    delayBetweenBatches: number
  }
}
```

### Question and Answer Chain

**Node**: `@n8n/n8n-nodes-langchain.chainRetrievalQa`\
**Description**: Answer questions about retrieved documents (RAG)

**Source Reference**: `/home/daytona/workspace/source/packages/@n8n/nodes-langchain/nodes/chains/ChainRetrievalQA/ChainRetrievalQa.node.ts:20`

**Inputs**:

* Main (required)
* Language Model (required)
* Retriever (required)

**Use Cases**:

* Document Q\&A
* Knowledge base search
* RAG applications

### Summarization Chain

**Node**: `@n8n/n8n-nodes-langchain.chainSummarization`\
**Description**: Summarize long documents efficiently

**Methods**:

* Map-Reduce: Process chunks in parallel
* Refine: Iteratively refine summary
* Stuff: Single pass (for short docs)

### Information Extractor

**Node**: `@n8n/n8n-nodes-langchain.informationExtractor`\
**Description**: Extract structured data from unstructured text

**Features**:

* Define custom schema
* Extract multiple entities
* Handle large documents

### Text Classifier

**Node**: `@n8n/n8n-nodes-langchain.textClassifier`\
**Description**: Classify text into predefined categories

**Use Cases**:

* Content categorization
* Intent detection
* Topic classification

### Sentiment Analysis

**Node**: `@n8n/n8n-nodes-langchain.sentimentAnalysis`\
**Description**: Analyze sentiment in text

**Output**:

* Positive, Negative, or Neutral
* Confidence score

## Tools

Tools extend agent capabilities.

### Calculator Tool

**Node**: `@n8n/n8n-nodes-langchain.toolCalculator`\
**Description**: Perform mathematical calculations

### Code Tool

**Node**: `@n8n/n8n-nodes-langchain.toolCode`\
**Description**: Execute JavaScript or Python code

**Parameters**:

* `language`: javascript or python
* `code`: Code to execute

### HTTP Request Tool

**Node**: `@n8n/n8n-nodes-langchain.toolHttpRequest`\
**Description**: Make HTTP requests to APIs

**Parameters**:

* `method`: GET, POST, PUT, DELETE, etc.
* `url`: API endpoint
* `authentication`: Auth method

### Call n8n Sub-Workflow Tool

**Node**: `@n8n/n8n-nodes-langchain.toolWorkflow`\
**Description**: Use another n8n workflow as a tool

**Source Reference**: `/home/daytona/workspace/source/packages/@n8n/nodes-langchain/nodes/tools/ToolWorkflow/ToolWorkflow.node.ts:10`

<Tip>
  This is the most powerful tool - turn ANY n8n workflow into a tool your agent can use!
</Tip>

**Parameters**:

* `workflowId`: Target workflow
* `description`: Tool description for agent
* `inputs`: Input schema

### Vector Store Tool

**Node**: `@n8n/n8n-nodes-langchain.toolVectorStore`\
**Description**: Query vector stores for semantic search

**Inputs**:

* Vector Store (required)

### Wikipedia Tool

**Node**: `@n8n/n8n-nodes-langchain.toolWikipedia`\
**Description**: Search Wikipedia

### SerpAPI Tool

**Node**: `@n8n/n8n-nodes-langchain.toolSerpApi`\
**Description**: Search Google using SerpAPI

**Requires**: SerpAPI credentials

### SearXng Tool

**Node**: `@n8n/n8n-nodes-langchain.toolSearXng`\
**Description**: Privacy-focused meta search engine

### Wolfram Alpha Tool

**Node**: `@n8n/n8n-nodes-langchain.toolWolframAlpha`\
**Description**: Computational knowledge engine

### Think Tool

**Node**: `@n8n/n8n-nodes-langchain.toolThink`\
**Description**: Internal reasoning tool for agents

## Memory

Memory nodes maintain conversation context.

### Simple Memory

**Node**: `@n8n/n8n-nodes-langchain.memoryBufferWindow`\
**Description**: In-memory conversation buffer

**Source Reference**: `/home/daytona/workspace/source/packages/@n8n/nodes-langchain/nodes/memory/MemoryBufferWindow/MemoryBufferWindow.node.ts:75`

**Parameters**:

```typescript theme={null}
{
  sessionIdType: 'customKey' | 'expressionKey',
  sessionKey?: string,
  contextWindowLength: number // Number of messages to remember
}
```

<Note>
  Not suitable for production environments with Queue Mode or multi-main setups.
</Note>

### Redis Memory

**Node**: `@n8n/n8n-nodes-langchain.memoryRedisChat`\
**Description**: Distributed memory using Redis

**Best for**: Production with multiple workers

### Postgres Memory

**Node**: `@n8n/n8n-nodes-langchain.memoryPostgresChat`\
**Description**: SQL-based persistent memory

**Best for**: Queryable conversation history

### MongoDB Memory

**Node**: `@n8n/n8n-nodes-langchain.memoryMongoDbChat`\
**Description**: Document-based memory storage

### Xata Memory

**Node**: `@n8n/n8n-nodes-langchain.memoryXata`\
**Description**: Xata database memory

### Zep Memory

**Node**: `@n8n/n8n-nodes-langchain.memoryZep`\
**Description**: Advanced memory with summarization and fact extraction

**Features**:

* Automatic summarization
* Fact extraction
* Long-term memory

### Motorhead Memory

**Node**: `@n8n/n8n-nodes-langchain.memoryMotorhead`\
**Description**: Motorhead memory service

### Memory Manager

**Node**: `@n8n/n8n-nodes-langchain.memoryManager`\
**Description**: Advanced memory management

### Memory Chat Retriever

**Node**: `@n8n/n8n-nodes-langchain.memoryChatRetriever`\
**Description**: Retrieve relevant past conversations

## Language Models

### OpenAI Models

**Chat Model**: `@n8n/n8n-nodes-langchain.lmChatOpenAi`\
**Completion Model**: `@n8n/n8n-nodes-langchain.lmOpenAi`

**Supported Models**:

* GPT-4, GPT-4 Turbo
* GPT-3.5 Turbo
* Custom models via base URL

### Anthropic (Claude)

**Node**: `@n8n/n8n-nodes-langchain.lmChatAnthropic`

**Models**:

* Claude 3 Opus
* Claude 3 Sonnet
* Claude 3 Haiku

### Google Models

**Gemini**: `@n8n/n8n-nodes-langchain.lmChatGoogleGemini`\
**Vertex AI**: `@n8n/n8n-nodes-langchain.lmChatGoogleVertex`

### Azure OpenAI

**Node**: `@n8n/n8n-nodes-langchain.lmChatAzureOpenAi`

### Cohere

**Chat Model**: `@n8n/n8n-nodes-langchain.lmChatCohere`\
**Completion Model**: `@n8n/n8n-nodes-langchain.lmCohere`

### Ollama (Local)

**Chat Model**: `@n8n/n8n-nodes-langchain.lmChatOllama`\
**Completion Model**: `@n8n/n8n-nodes-langchain.lmOllama`

**Use Cases**:

* Local/private deployments
* No API costs
* Custom models

### Other Providers

* **Groq**: `lmChatGroq` - Fast inference
* **Mistral**: `lmChatMistralCloud`
* **DeepSeek**: `lmChatDeepSeek`
* **Hugging Face**: `lmOpenHuggingFaceInference`
* **AWS Bedrock**: `lmChatAwsBedrock`
* **OpenRouter**: `lmChatOpenRouter`
* **Vercel AI Gateway**: `lmChatVercelAiGateway`
* **X.AI (Grok)**: `lmChatXAiGrok`
* **Lemonade AI**: `lmChatLemonade`

## Retrievers

Retrievers fetch relevant documents for RAG applications.

### Vector Store Retriever

**Node**: `@n8n/n8n-nodes-langchain.retrieverVectorStore`\
**Description**: Retrieve documents from vector stores

**Source Reference**: `/home/daytona/workspace/source/packages/@n8n/nodes-langchain/nodes/retrievers/RetrieverVectorStore/RetrieverVectorStore.node.ts:14`

**Parameters**:

```typescript theme={null}
{
  topK: number // Number of results to return
}
```

### Contextual Compression Retriever

**Node**: `@n8n/n8n-nodes-langchain.retrieverContextualCompression`\
**Description**: Compress retrieved documents to relevant parts

**Uses**: Reranker nodes to improve results

### Multi-Query Retriever

**Node**: `@n8n/n8n-nodes-langchain.retrieverMultiQuery`\
**Description**: Generate multiple queries for better retrieval

### Workflow Retriever

**Node**: `@n8n/n8n-nodes-langchain.retrieverWorkflow`\
**Description**: Use n8n workflow as a retriever

## Document Loaders

Load data from various sources.

### Default Data Loader

**Node**: `@n8n/n8n-nodes-langchain.documentDefaultDataLoader`\
**Description**: Load text data from previous nodes

### Binary Input Loader

**Node**: `@n8n/n8n-nodes-langchain.documentBinaryInputLoader`\
**Description**: Load documents from binary files

**Supported Formats**:

* PDF
* DOCX
* TXT
* CSV
* JSON
* HTML
* Markdown

### JSON Input Loader

**Node**: `@n8n/n8n-nodes-langchain.documentJsonInputLoader`\
**Description**: Load structured JSON data

### GitHub Loader

**Node**: `@n8n/n8n-nodes-langchain.documentGithubLoader`\
**Description**: Load files from GitHub repositories

## Text Splitters

Split documents into chunks for processing.

### Character Text Splitter

**Node**: `@n8n/n8n-nodes-langchain.textSplitterCharacterTextSplitter`\
**Description**: Split by character count

**Parameters**:

* `chunkSize`: Characters per chunk
* `chunkOverlap`: Overlap between chunks

### Recursive Character Text Splitter

**Node**: `@n8n/n8n-nodes-langchain.textSplitterRecursiveCharacterTextSplitter`\
**Description**: Smart splitting that preserves structure

**Recommended**: Best for most use cases

### Token Splitter

**Node**: `@n8n/n8n-nodes-langchain.textSplitterTokenSplitter`\
**Description**: Split by token count

**Best for**: Precise token budget management

## Output Parsers

Structure LLM outputs.

### Structured Output Parser

**Node**: `@n8n/n8n-nodes-langchain.outputParserStructured`\
**Description**: Parse into custom JSON schema

**Example Schema**:

```json theme={null}
{
  "type": "object",
  "properties": {
    "answer": { "type": "string" },
    "confidence": { "type": "number" }
  },
  "required": ["answer"]
}
```

### Auto-fixing Output Parser

**Node**: `@n8n/n8n-nodes-langchain.outputParserAutofixing`\
**Description**: Automatically fix malformed JSON

### Item List Output Parser

**Node**: `@n8n/n8n-nodes-langchain.outputParserItemList`\
**Description**: Extract lists from text

## Rerankers

Improve search result quality.

### Cohere Reranker

**Node**: `@n8n/n8n-nodes-langchain.rerankerCohere`\
**Description**: Rerank search results using Cohere

**Use with**: Contextual Compression Retriever

## Triggers

### Chat Trigger

**Node**: `@n8n/n8n-nodes-langchain.chatTrigger`\
**Description**: Webhook-based chat interface

**Features**:

* REST API endpoint
* Streaming responses
* Session management

### Manual Chat Trigger

**Node**: `@n8n/n8n-nodes-langchain.manualChatTrigger`\
**Description**: Manual testing interface

**Best for**: Development and testing

## Utility Nodes

### Model Selector

**Node**: `@n8n/n8n-nodes-langchain.modelSelector`\
**Description**: Dynamically select language models

### Tool Executor

**Node**: `@n8n/n8n-nodes-langchain.toolExecutor`\
**Description**: Execute tools programmatically

### Guardrails

**Node**: `@n8n/n8n-nodes-langchain.guardrails`\
**Description**: Add safety checks and input validation

**Features**:

* Input validation
* Output filtering
* Safety checks
* Custom rules

## MCP (Model Context Protocol)

n8n supports MCP for advanced AI integrations:

### MCP Client

**Node**: `@n8n/n8n-nodes-langchain.mcpClient`\
**Description**: Connect to MCP servers

### MCP Client Tool

**Node**: `@n8n/n8n-nodes-langchain.mcpClientTool`\
**Description**: Use MCP as a tool

### MCP Trigger

**Node**: `@n8n/n8n-nodes-langchain.mcpTrigger`\
**Description**: Trigger workflows from MCP events

## Vendor-Specific Nodes

### OpenAI Node

**Node**: `@n8n/n8n-nodes-langchain.openAi`\
**Description**: Direct OpenAI API access

**Operations**:

* Text generation
* Image generation
* Audio transcription
* File operations

### Anthropic Node

**Node**: `@n8n/n8n-nodes-langchain.anthropic`\
**Description**: Direct Anthropic API access

### Google Gemini Node

**Node**: `@n8n/n8n-nodes-langchain.googleGemini`\
**Description**: Google Gemini operations

### Ollama Node

**Node**: `@n8n/n8n-nodes-langchain.ollama`\
**Description**: Ollama server operations

## Next Steps

<CardGroup cols={2}>
  <Card title="Build an Agent" icon="robot" href="/ai/agents">
    Create your first AI agent
  </Card>

  <Card title="Vector Stores" icon="database" href="/ai/vector-stores">
    Set up semantic search
  </Card>

  <Card title="Embeddings" icon="layers" href="/ai/embeddings">
    Configure embedding models
  </Card>

  <Card title="Workflow Templates" icon="template">
    Browse example workflows
  </Card>
</CardGroup>