Documentation Index
Fetch the complete documentation index at: https://mintlify.com/ComposioHQ/composio/llms.txt
Use this file to discover all available pages before exploring further.
Composio’s LangChain provider returns tools as DynamicStructuredTool objects you can use directly in LCEL chains, LangGraph agents, and any LangChain agent executor. Tool execution is wired in automatically — LangChain calls the tool and Composio handles the API call behind the scenes.
Installation
npm install @composio/core @composio/langchain @langchain/openai @langchain/langgraph @langchain/core
pip install composio composio-langchain langchain langchain-openai
.env file:
COMPOSIO_API_KEY=your_composio_api_key
OPENAI_API_KEY=your_openai_api_key
Example
The TypeScript example uses a LangGraph StateGraph to build a ReAct-style agentic loop. Composio tools are passed directly to model.bindTools() and to a ToolNode — no extra wiring needed.import { ChatOpenAI } from '@langchain/openai';
import { HumanMessage, AIMessage } from '@langchain/core/messages';
import { ToolNode } from '@langchain/langgraph/prebuilt';
import { StateGraph, MessagesAnnotation } from '@langchain/langgraph';
import { Composio } from '@composio/core';
import { LangchainProvider } from '@composio/langchain';
import 'dotenv/config';
const composio = new Composio({
apiKey: process.env.COMPOSIO_API_KEY,
provider: new LangchainProvider(),
});
// Tools are returned as DynamicStructuredTool[] — LangChain's native format
const tools = await composio.tools.get('default', 'HACKERNEWS_GET_USER');
const toolNode = new ToolNode(tools);
const model = new ChatOpenAI({
model: 'gpt-4o-mini',
temperature: 0,
}).bindTools(tools);
function shouldContinue({ messages }: typeof MessagesAnnotation.State) {
const lastMessage = messages[messages.length - 1] as AIMessage;
return lastMessage.tool_calls?.length ? 'tools' : '__end__';
}
async function callModel(state: typeof MessagesAnnotation.State) {
const response = await model.invoke(state.messages);
return { messages: [response] };
}
const workflow = new StateGraph(MessagesAnnotation)
.addNode('agent', callModel)
.addEdge('__start__', 'agent')
.addNode('tools', toolNode)
.addEdge('tools', 'agent')
.addConditionalEdges('agent', shouldContinue);
const app = workflow.compile();
const finalState = await app.invoke({
messages: [new HumanMessage('Find the details of the user `pg` on HackerNews')],
});
console.log(finalState.messages[finalState.messages.length - 1].content);
The Python example uses create_agent from LangChain to build a simple agent with tool-calling capability.from composio import Composio
from composio_langchain import LangchainProvider
from langchain.agents import create_agent
from langchain_openai import ChatOpenAI
openai_client = ChatOpenAI(model="gpt-4o-mini")
composio = Composio(provider=LangchainProvider())
# Tools are returned as LangChain StructuredTool objects
tools = composio.tools.get(user_id="default", toolkits=["GITHUB"])
task = "Star the composiohq/composio repository on GitHub"
agent = create_agent(
model=openai_client,
tools=tools,
system_prompt="You are a helpful assistant with GitHub access.",
)
result = agent.invoke({"messages": [{"role": "user", "content": task}]})
print(result["messages"][-1].content)
LangGraph
The same DynamicStructuredTool objects work directly in LangGraph agents. Wrap them with ToolNode and add conditional edges to build a standard ReAct loop, as shown in the TypeScript example above.
For Python LangGraph, use the dedicated composio-langgraph package and LanggraphProvider:
from composio import Composio
from composio_langgraph import LanggraphProvider
from langchain.agents import create_agent
from langchain_openai import ChatOpenAI
composio = Composio(provider=LanggraphProvider())
llm = ChatOpenAI(model="gpt-4o-mini")
session = composio.create(user_id="user_123")
tools = session.tools()
agent = create_agent(tools=tools, model=llm)
result = agent.invoke({"messages": [{"role": "user", "content": "Star composiohq/composio on GitHub"}]})
print(result["messages"][-1].content)
Install the LangGraph Python provider with pip install composio composio-langgraph langgraph langchain-openai.
Multi-turn conversations
Create a session once and reuse it across turns so your agent keeps the same connected accounts and session context:
// First turn — create a session and store the ID
const session = await composio.create('user_123');
const sessionId = session.sessionId;
const tools = await composio.tools.get('default', 'HACKERNEWS_GET_USER');
// Subsequent turns — retrieve the existing session
const existingSession = await composio.use(sessionId);
const sameTools = await existingSession.tools();
# First turn — create a session and store the ID
session = composio.create(user_id="user_123")
session_id = session.session_id
# Subsequent turns — retrieve the existing session
session = composio.use(session_id)
tools = session.tools()
