Composio

By Composio

Connects agents to 1000+ external apps with managed auth and ready-made tool definitions.

Website ↗ Repository ↗ Docs ↗

@composio/experimental@0.2.02026-07-16PythonMIT

#tool-use #integrations #auth #agent

29.3k Open source Free tier Paid

Doc version 1.0 · Updated June 28, 2026

Related concepts

Tools Doc version 1.0 · Updated June 29, 2026 The functions that connect a model to the outside world — web search, scraping, browser control, and app integration that bridge to real data and actions beyond the training cutoff.

Related tools

LangGraph

Multi-Agent Orchestration

CrewAI

Multi-Agent Orchestration

Overview

Composio is an integration layer that gives agents real tool access to 1000+ external apps.
It ships ready-made tool definitions and handles the managed auth, so the model can read Gmail, open GitHub issues, or post to Slack without bespoke glue code.

flowchart LR
  agent[agent] --> composio[Composio]
  composio --> apps[(Gmail · GitHub · Slack …)]

The Code samples tab shows fetching tools for a toolkit and letting an LLM call them.

When to use it

Choose Composio when an agent must take actions in third-party services and you want managed OAuth plus tool schemas instead of building each connector yourself. It plugs into frameworks like LangGraph and CrewAI through provider adapters.

Pricing

The SDK is open source (MIT). The hosted platform handles auth and tool execution, billed per tool call.

Plan	Price	Notes
Totally Free	$0	20k tool calls/mo, community support
Ridiculously Cheap	$29/mo	200k tool calls/mo, then $0.299/1k
Serious Business	$229/mo	2M tool calls/mo, then $0.249/1k
Enterprise	Custom	SLA, SOC 2, VPC/on-prem

Source ↗ · as of 2026-06-27

Example

Python · use a tool

When to use

Reach for Composio when an agent needs to act in real apps — email, GitHub, Slack — without you hand-writing each integration or its OAuth flow.

How it flows

tools.get returns auth-wired tool schemas; the model picks calls and handle_tool_calls executes them against the live app.

Structure

flowchart LR
  agent[agent] --> composio[Composio]
  composio --> apps[(Gmail · GitHub · Slack …)]

from composio import Composio
from composio_openai import OpenAIProvider
from openai import OpenAI

composio = Composio(provider=OpenAIProvider())
client = OpenAI()

# Fetch ready-made, auth-wired tool definitions for a user.
tools = composio.tools.get(user_id="alice", toolkits=["GITHUB"])

resp = client.chat.completions.create(
    model="gpt-4o",
    tools=tools,
    messages=[{"role": "user", "content": "Star the ComposioHQ/composio repo."}],
)

# Composio runs whatever tool calls the model requested.
result = composio.provider.handle_tool_calls(user_id="alice", response=resp)
print(result)