Composio Complete Guide: The Open Source Framework Empowering AI Agents with 1000+ Tools

# Using pip
pip install composio

# Or using poetry
poetry add composio

pip install composio_openai_agents openai-agents

# Using npm
npm install @composio/core

# Using yarn
yarn add @composio/core

# Using pnpm
pnpm add @composio/core

npm install @composio/openai-agents @openai/agents

export COMPOSIO_API_KEY="your-api-key-here"

import asyncio
from agents import Agent, Runner
from composio import Composio
from composio_openai_agents import OpenAIAgentsProvider

# Initialize Composio client
composio = Composio(provider=OpenAIAgentsProvider())

# Get user ID (for authentication isolation)
user_id = "user@example.com"

# Get Hacker News tools
tools = composio.tools.get(
    user_id=user_id,
    toolkits=["HACKERNEWS"]
)

# Create Agent
agent = Agent(
    name="HackerNews Agent",
    instructions="You are a Hacker News assistant, helping users understand the latest tech trends.",
    tools=tools,
)

# Run Agent
async def main():
    result = await Runner.run(
        starting_agent=agent,
        input="What are the hottest tech posts on Hacker News today?",
    )
    print(result.final_output)

asyncio.run(main())

import { Composio } from '@composio/core';
import { OpenAIAgentsProvider } from '@composio/openai-agents';
import { Agent, run } from '@openai/agents';

// Initialize SDK
const composio = new Composio({
  provider: new OpenAIAgentsProvider(),
});

const userId = 'user@example.com';

// Get tools
const tools = await composio.tools.get(userId, {
  toolkits: ['HACKERNEWS'],
});

// Create Agent
const agent = new Agent({
  name: 'HackerNews assistant',
  tools: tools,
});

// Run
const result = await run(agent, 'What is the latest hackernews post about?');
console.log(JSON.stringify(result.finalOutput, null, 2));

from composio import Composio
from composio_openai_agents import OpenAIAgentsProvider

# Initialize
composio = Composio(provider=OpenAIAgentsProvider())
user_id = "dev-team@company.com"

# Get multiple toolkits
tools = composio.tools.get(
    user_id=user_id,
    toolkits=["GITHUB", "SLACK"],
)

# Create DevOps Agent
agent = Agent(
    name="DevOps Assistant",
    instructions="""
    You are a DevOps assistant who can:
    1. Query the status of GitHub repository Issues
    2. Create new Issues or Pull Requests
    3. Send notifications to Slack channels

    When important Issue updates are found, automatically notify the team.
    """,
    tools=tools,
)

async def monitor_issues():
    """Monitor critical Issues and notify the team"""
    result = await Runner.run(
        starting_agent=agent,
        input="""
        Check the composiohq/composio repository for Issues labeled 'bug' that haven't been replied to in over 7 days,
        and send the list to the #engineering Slack channel.
        """,
    )
    print(result.final_output)

# Just one line of code to trigger the authentication flow
connection = composio.connections.initiate(
    user_id=user_id,
    app_name="gmail",
)
# After the user completes OAuth authorization, Composio automatically manages tokens

from composio_langchain import ComposioToolSet

toolset = ComposioToolSet()
tools = toolset.get_tools(apps=["github", "slack"])

from langchain.agents import initialize_agent
agent = initialize_agent(
    tools=tools,
    llm=your_llm,
    agent_type="zero-shot-react-description"
)

from crewai import Agent, Task, Crew
from composio_crewai import ComposioToolSet

toolset = ComposioToolSet()
github_tools = toolset.get_tools(apps=["github"])

researcher = Agent(
    role='GitHub Researcher',
    goal='Analyze trending GitHub projects',
    tools=github_tools,
    backstory='You are an open source project analyst',
)

from composio_autogen import ComposioToolSet

toolset = ComposioToolSet()
tools = toolset.get_tools(apps=["notion", "gmail"])

# Pass tools to AutoGen Agent
assistant = ConversableAgent(
    "assistant",
    llm_config={"config_list": config_list},
    function_map=tools.get_function_map(),
)

from composio import Composio
from composio_openai_agents import OpenAIAgentsProvider

# Initialize
composio = Composio(provider=OpenAIAgentsProvider())
user_id = "support-bot@company.com"

# Get customer service related tools
tools = composio.tools.get(
    user_id=user_id,
    toolkits=["NOTION", "JIRA", "GMAIL"],
)

# Create Customer Service Agent
support_agent = Agent(
    name="Customer Support Agent",
    instructions="""
    You are the company's intelligent customer service assistant. Workflow:

    1. First search for answers to user questions in the Notion knowledge base
    2. If an answer is found, send it to the user via Gmail
    3. If no answer is found, create a support ticket in Jira,
       assign it to the technical support team, and inform the user of the ticket number via Gmail

    Maintain a friendly and professional tone.
    """,
    tools=tools,
)

# Handle user queries
async def handle_customer_query(customer_email: str, question: str):
    prompt = f"""
    Customer Email: {customer_email}
    Question: {question}

    Please handle according to the workflow.
    """

    result = await Runner.run(
        starting_agent=support_agent,
        input=prompt,
    )
    return result.final_output

# Query multiple data sources simultaneously
tools = composio.tools.get(
    user_id=user_id,
    toolkits=["GITHUB", "GITLAB", "BITBUCKET"],
)
# Agent can query project information from all three platforms in parallel

tools = composio.tools.get(
    user_id=user_id,
    toolkits=["HACKERNEWS"],
    cache_ttl=300,  # Cache for 5 minutes
)

# Development scenario: only load code-related tools
dev_tools = composio.tools.get(user_id, toolkits=["GITHUB", "GITLAB"])

# Office scenario: only load productivity tools
office_tools = composio.tools.get(user_id, toolkits=["GMAIL", "NOTION", "SLACK"])

try:
    result = await Runner.run(agent, input)
except ToolExecutionError as e:
    print(f"Tool execution failed: {e}")
    # Fallback strategy: use backup tools or return a friendly message