#!/usr/bin/env node
/**
* LangChain Anthropic MCP Server
*
* Exposes LangChain capabilities as MCP tools:
* - Supabase vector store setup
* - RAG chain creation
* - Document ingestion
* - Vector search
* - Claude integration
*
* Use this MCP server in Cursor to quickly set up LangChain
* features in any project.
*/
import { Server } from "@modelcontextprotocol/sdk/server/index.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import {
CallToolRequestSchema,
ListToolsRequestSchema,
} from "@modelcontextprotocol/sdk/types.js";
// Initialize server
const server = new Server(
{
name: "langchain-anthropic-server",
version: "1.0.0",
},
{
capabilities: {
tools: {},
},
}
);
/**
* List available LangChain tools
*/
server.setRequestHandler(ListToolsRequestSchema, async () => ({
tools: [
{
name: "setup_supabase_vectorstore",
description: "Generate code to set up Supabase as a vector store with LangChain. Returns TypeScript code ready to use.",
inputSchema: {
type: "object",
properties: {
project_name: {
type: "string",
description: "Name of the project",
},
table_name: {
type: "string",
description: "Name of the vector table in Supabase",
default: "documents",
},
embedding_model: {
type: "string",
description: "Embedding model to use",
enum: ["text-embedding-004", "text-embedding-3-small", "text-embedding-3-large"],
default: "text-embedding-004",
},
dimension: {
type: "number",
description: "Vector dimension (768 for Gemini, 1536 for OpenAI)",
default: 768,
},
},
required: ["project_name"],
},
},
{
name: "create_rag_chain",
description: "Generate a complete RAG chain with LangChain and Anthropic Claude. Returns TypeScript implementation.",
inputSchema: {
type: "object",
properties: {
vectorstore_type: {
type: "string",
description: "Type of vector store",
enum: ["supabase", "pinecone", "chroma", "in-memory"],
default: "supabase",
},
claude_model: {
type: "string",
description: "Claude model to use",
enum: ["claude-haiku-4-5-20251001", "claude-sonnet-4-5-20250929", "claude-opus-4-5-20250929"],
default: "claude-sonnet-4-5-20250929",
},
retriever_k: {
type: "number",
description: "Number of documents to retrieve",
default: 5,
},
enable_caching: {
type: "boolean",
description: "Enable prompt caching (90% cost savings)",
default: true,
},
},
required: [],
},
},
{
name: "generate_document_ingestion",
description: "Generate code for ingesting documents into a vector store with chunking and embedding.",
inputSchema: {
type: "object",
properties: {
source_type: {
type: "string",
description: "Type of documents to ingest",
enum: ["markdown", "pdf", "text", "json", "csv"],
default: "markdown",
},
chunk_size: {
type: "number",
description: "Chunk size in characters",
default: 1000,
},
chunk_overlap: {
type: "number",
description: "Overlap between chunks",
default: 200,
},
embedding_model: {
type: "string",
description: "Embedding model",
default: "text-embedding-004",
},
},
required: ["source_type"],
},
},
{
name: "create_conversational_rag",
description: "Generate a conversational RAG system with memory that maintains context across multiple queries.",
inputSchema: {
type: "object",
properties: {
memory_type: {
type: "string",
description: "Type of conversation memory",
enum: ["buffer", "summary", "buffer-window"],
default: "buffer",
},
claude_model: {
type: "string",
description: "Claude model",
default: "claude-sonnet-4-5-20250929",
},
},
required: [],
},
},
{
name: "setup_hybrid_search",
description: "Generate hybrid search setup combining vector similarity and keyword search (BM25).",
inputSchema: {
type: "object",
properties: {
vector_weight: {
type: "number",
description: "Weight for vector search (0-1)",
default: 0.7,
},
keyword_weight: {
type: "number",
description: "Weight for keyword search (0-1)",
default: 0.3,
},
},
required: [],
},
},
{
name: "create_multi_query_retriever",
description: "Generate a multi-query retriever that generates multiple search queries for better recall.",
inputSchema: {
type: "object",
properties: {
num_queries: {
type: "number",
description: "Number of query variations to generate",
default: 3,
},
},
required: [],
},
},
{
name: "setup_extended_thinking",
description: "Generate code for extended thinking mode with Claude for complex reasoning tasks.",
inputSchema: {
type: "object",
properties: {
thinking_budget: {
type: "number",
description: "Token budget for thinking",
default: 8000,
},
complexity_threshold: {
type: "number",
description: "Complexity score threshold to enable thinking (0-100)",
default: 60,
},
},
required: [],
},
},
{
name: "generate_package_setup",
description: "Generate package.json dependencies and setup instructions for LangChain with Anthropic.",
inputSchema: {
type: "object",
properties: {
features: {
type: "array",
description: "Features to include",
items: {
type: "string",
enum: ["supabase", "pdf", "csv", "streaming", "caching"],
},
},
},
required: [],
},
},
],
}));
// ... rest of implementation truncated for brevity ...
