Connecting LangChain & LangGraph

footage.one MCP server as a tool source for LangChain and LangGraph agents in Python and JavaScript.

Overview

LangChain has supported MCP servers since version 0.3.20 (Python) and @langchain/mcp-adapters (JS). Tools are loaded at runtime from the MCP server and integrated into the agent.

Auth flow

The footage.one MCP server currently accepts OAuth2 exclusively. You therefore need a registered OAuth client (see OAuth2 for third-party apps) and must embed the access token from the auth flow into the MCP adapter.

Headless scripts without user interaction cannot currently use the MCP endpoint — Bearer token auth at MCP is in progress but not yet available. For pure server-to-server calls, use the REST API directly with an API token.

Python — langchain-mcp-adapters

Install:

pip install langchain-mcp-adapters langgraph langchain-anthropic

Setup with OAuth access token (from your own OAuth flow):

from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.prebuilt import create_react_agent
from langchain_anthropic import ChatAnthropic

ACCESS_TOKEN = "..."  # from OAuth flow

client = MultiServerMCPClient({
    "footage-one": {
        "transport": "streamable_http",
        "url": "https://mcp.footage.one/mcp",
        "headers": {"Authorization": f"Bearer {ACCESS_TOKEN}"},
    }
})

tools = await client.get_tools()
agent = create_react_agent(ChatAnthropic(model="claude-sonnet-4-6"), tools)

result = await agent.ainvoke({
    "messages": [("user", "Search for asset titles containing 'Berlin' and list them with their albums.")]
})
print(result["messages"][-1].content)

JavaScript — @langchain/mcp-adapters

bun add @langchain/mcp-adapters @langchain/langgraph @langchain/anthropic

import { MultiServerMCPClient } from '@langchain/mcp-adapters';
import { createReactAgent } from '@langchain/langgraph/prebuilt';
import { ChatAnthropic } from '@langchain/anthropic';

const accessToken = '...'; // from OAuth flow

const client = new MultiServerMCPClient({
  'footage-one': {
    transport: 'streamable_http',
    url: 'https://mcp.footage.one/mcp',
    headers: { Authorization: `Bearer ${accessToken}` },
  },
});

const tools = await client.getTools();
const agent = createReactAgent({
  llm: new ChatAnthropic({ model: 'claude-sonnet-4-6' }),
  tools,
});

const result = await agent.invoke({
  messages: [{ role: 'user', content: 'List the last five albums.' }],
});
console.log(result.messages.at(-1)?.content);

Next steps

n8nCustom GPT