How to integrate Openai MCP with Autogen

Connect Autogen to Openai MCP. List all available openai models, upload a file for fine-tuning, and more using natural language, with authentication handled for you.

Openai logoOpenai
Api Key

Openai is a powerful AI platform offering advanced language, vision, and multimodal APIs. It's your gateway to building, managing, and scaling intelligent assistants and workflows.

128 Tools

Introduction

This guide walks you through connecting Openai to AutoGen using the Composio tool router. By the end, you'll have a working Openai agent that can generate chat completions and responses, create and edit images, transcribe audio, create embeddings, and run fine-tuning jobs through natural language commands.

This guide will help you understand how to give your AutoGen agent real control over a Openai account through Composio's Openai MCP server.

Before we dive in, let's take a quick look at the key ideas and tools involved.

For exact tool slugs and parameter tables, see the tool catalog in the Composio docs.

Also integrate Openai with

TL;DR

Here's what you'll learn:
  • Get and set up your OpenAI and Composio API keys
  • Install the required dependencies for Autogen and Composio
  • Initialize Composio and create a Tool Router session for Openai
  • Wire that MCP URL into Autogen using McpWorkbench and StreamableHttpServerParams
  • Configure an Autogen AssistantAgent that can call Openai tools
  • Run a live chat loop where you ask the agent to perform Openai operations

What is AutoGen?

Autogen is a framework for building multi-agent conversational AI systems from Microsoft. It enables you to create agents that can collaborate, use tools, and maintain complex workflows.

Key features include:

  • Multi-Agent Systems: Build collaborative agent workflows
  • MCP Workbench: Native support for Model Context Protocol tools
  • Streaming HTTP: Connect to external services through streamable HTTP
  • AssistantAgent: Pre-built agent class for tool-using assistants

What is the Openai MCP server, and what's possible with it?

The Openai MCP server is an implementation of the Model Context Protocol that connects your AI agent and assistants like Claude, Cursor, etc directly to your OpenAI account. It provides structured and secure access to your models, assistants, files, threads, and fine-tuning jobs, so your agent can perform actions like managing assistants, handling conversations, uploading or organizing files, and working with OpenAI models on your behalf.

  • Assistant and conversation management: Quickly create, update, or delete OpenAI assistants and manage threads or messages for seamless conversational flows.
  • File uploads and organization: Let your agent upload new files, list all uploaded documents, or delete unnecessary files to keep your workspace tidy.
  • Model discovery and utilization: Effortlessly list all available OpenAI models—including vision and multimodal—and retrieve their details to choose the best fit for your tasks.
  • Fine-tuning job insights: View a complete list of your organization's fine-tune jobs and track their progress or review results as needed.
  • Thread and run management: Create, modify, or inspect threads and run steps to fully control and monitor interactive agent conversations.

What is the Composio tool router, and how does it fit here?

What is Composio SDK?

Composio's Composio SDK helps agents find the right tools for a task at runtime. You can plug in multiple toolkits (like Gmail, HubSpot, and GitHub), and the agent will identify the relevant app and action to complete multi-step workflows. This can reduce token usage and improve the reliability of tool calls. Read more here: Getting started with Composio SDK

The tool router generates a secure MCP URL that your agents can access to perform actions.

How the Composio SDK works

The Composio SDK follows a three-phase workflow:

  1. Discovery: Searches for tools matching your task and returns relevant toolkits with their details.
  2. Authentication: Checks for active connections. If missing, creates an auth config and returns a connection URL via Auth Link.
  3. Execution: Executes the action using the authenticated connection.

Step-by-step Guide

Step by step08 STEPS
1

Prerequisites

You will need:

  • A Composio API key
  • An OpenAI API key (used by Autogen's OpenAIChatCompletionClient)
  • A Openai account you can connect to Composio
  • Some basic familiarity with Autogen and Python async
2

Getting API Keys for OpenAI and Composio

OpenAI API Key
  • Go to the OpenAI dashboard and create an API key. You'll need credits to use the models, or you can connect to another model provider.
  • Keep the API key safe.
Composio API Key
  • Log in to the Composio dashboard.
  • Navigate to your API settings and generate a new API key.
  • Store this key securely as you'll need it for authentication.
3

Install dependencies

bash
pip install composio python-dotenv
pip install autogen-agentchat autogen-ext-openai autogen-ext-tools

Install Composio, Autogen extensions, and dotenv.

What's happening:

  • composio connects your agent to Openai via MCP
  • autogen-agentchat provides the AssistantAgent class
  • autogen-ext-openai provides the OpenAI model client
  • autogen-ext-tools provides MCP workbench support

4

Set up environment variables

bash
COMPOSIO_API_KEY=your-composio-api-key
OPENAI_API_KEY=your-openai-api-key
USER_ID=your-user-identifier@example.com

Create a .env file in your project folder.

What's happening:

  • COMPOSIO_API_KEY is required to talk to Composio
  • OPENAI_API_KEY is used by Autogen's OpenAI client
  • USER_ID is how Composio identifies which user's Openai connections to use
5

Import dependencies and create Tool Router session

python
import asyncio
import os
from dotenv import load_dotenv
from composio import Composio

from autogen_agentchat.agents import AssistantAgent
from autogen_ext.models.openai import OpenAIChatCompletionClient
from autogen_ext.tools.mcp import McpWorkbench, StreamableHttpServerParams

load_dotenv()

async def main():
    # Initialize Composio and create a Openai session
    composio = Composio(api_key=os.getenv("COMPOSIO_API_KEY"))
    session = composio.create(
        user_id=os.getenv("USER_ID"),
        toolkits=["openai"]
    )
    url = session.mcp.url
What's happening:
  • load_dotenv() reads your .env file
  • Composio(api_key=...) initializes the SDK
  • create(...) creates a Tool Router session that exposes Openai tools
  • session.mcp.url is the MCP endpoint that Autogen will connect to
6

Configure MCP parameters for Autogen

python
# Configure MCP server parameters for Streamable HTTP
server_params = StreamableHttpServerParams(
    url=url,
    timeout=30.0,
    sse_read_timeout=300.0,
    terminate_on_close=True,
    headers={"x-api-key": os.getenv("COMPOSIO_API_KEY")}
)

Autogen expects parameters describing how to talk to the MCP server. That is what StreamableHttpServerParams is for.

What's happening:

  • url points to the Tool Router MCP endpoint from Composio
  • timeout is the HTTP timeout for requests
  • sse_read_timeout controls how long to wait when streaming responses
  • terminate_on_close=True cleans up the MCP server process when the workbench is closed
7

Create the model client and agent

python
# Create model client
model_client = OpenAIChatCompletionClient(
    model="gpt-5",
    api_key=os.getenv("OPENAI_API_KEY")
)

# Use McpWorkbench as context manager
async with McpWorkbench(server_params) as workbench:
    # Create Openai assistant agent with MCP tools
    agent = AssistantAgent(
        name="openai_assistant",
        description="An AI assistant that helps with Openai operations.",
        model_client=model_client,
        workbench=workbench,
        model_client_stream=True,
        max_tool_iterations=10
    )

What's happening:

  • OpenAIChatCompletionClient wraps the OpenAI model for Autogen
  • McpWorkbench connects the agent to the MCP tools
  • AssistantAgent is configured with the Openai tools from the workbench
8

Run the interactive chat loop

python
print("Chat started! Type 'exit' or 'quit' to end the conversation.\n")
print("Ask any Openai related question or task to the agent.\n")

# Conversation loop
while True:
    user_input = input("You: ").strip()

    if user_input.lower() in ["exit", "quit", "bye"]:
        print("\nGoodbye!")
        break

    if not user_input:
        continue

    print("\nAgent is thinking...\n")

    # Run the agent with streaming
    try:
        response_text = ""
        async for message in agent.run_stream(task=user_input):
            if hasattr(message, "content") and message.content:
                response_text = message.content

        # Print the final response
        if response_text:
            print(f"Agent: {response_text}\n")
        else:
            print("Agent: I encountered an issue processing your request.\n")

    except Exception as e:
        print(f"Agent: Sorry, I encountered an error: {str(e)}\n")
What's happening:
  • The script prompts you in a loop with You:
  • Autogen passes your input to the model, which decides which Openai tools to call via MCP
  • agent.run_stream(...) yields streaming messages as the agent thinks and calls tools
  • Typing exit, quit, or bye ends the loop

Complete Code

Here's the complete code to get you started with Openai and AutoGen:

python
import asyncio
import os
from dotenv import load_dotenv
from composio import Composio

from autogen_agentchat.agents import AssistantAgent
from autogen_ext.models.openai import OpenAIChatCompletionClient
from autogen_ext.tools.mcp import McpWorkbench, StreamableHttpServerParams

load_dotenv()

async def main():
    # Initialize Composio and create a Openai session
    composio = Composio(api_key=os.getenv("COMPOSIO_API_KEY"))
    session = composio.create(
        user_id=os.getenv("USER_ID"),
        toolkits=["openai"]
    )
    url = session.mcp.url

    # Configure MCP server parameters for Streamable HTTP
    server_params = StreamableHttpServerParams(
        url=url,
        timeout=30.0,
        sse_read_timeout=300.0,
        terminate_on_close=True,
        headers={"x-api-key": os.getenv("COMPOSIO_API_KEY")}
    )

    # Create model client
    model_client = OpenAIChatCompletionClient(
        model="gpt-5",
        api_key=os.getenv("OPENAI_API_KEY")
    )

    # Use McpWorkbench as context manager
    async with McpWorkbench(server_params) as workbench:
        # Create Openai assistant agent with MCP tools
        agent = AssistantAgent(
            name="openai_assistant",
            description="An AI assistant that helps with Openai operations.",
            model_client=model_client,
            workbench=workbench,
            model_client_stream=True,
            max_tool_iterations=10
        )

        print("Chat started! Type 'exit' or 'quit' to end the conversation.\n")
        print("Ask any Openai related question or task to the agent.\n")

        # Conversation loop
        while True:
            user_input = input("You: ").strip()

            if user_input.lower() in ['exit', 'quit', 'bye']:
                print("\nGoodbye!")
                break

            if not user_input:
                continue

            print("\nAgent is thinking...\n")

            # Run the agent with streaming
            try:
                response_text = ""
                async for message in agent.run_stream(task=user_input):
                    if hasattr(message, 'content') and message.content:
                        response_text = message.content

                # Print the final response
                if response_text:
                    print(f"Agent: {response_text}\n")
                else:
                    print("Agent: I encountered an issue processing your request.\n")

            except Exception as e:
                print(f"Agent: Sorry, I encountered an error: {str(e)}\n")

if __name__ == "__main__":
    asyncio.run(main())

Conclusion

You now have an Autogen assistant wired into Openai through Composio's Tool Router and MCP. From here you can:
  • Add more toolkits to the toolkits list, for example notion or hubspot
  • Refine the agent description to point it at specific workflows
  • Wrap this script behind a UI, Slack bot, or internal tool
Once the pattern is clear for Openai, you can reuse the same structure for other MCP-enabled apps with minimal code changes.
TOOLS

Supported Tools

Every Openai action and event your agent gets out of the box.

Add Upload Part

Tool to add a part (chunk of bytes) to an Upload object.

Cancel batch

Tool to cancel an in-progress batch.

Cancel evaluation run

Tool to cancel an ongoing evaluation run.

Cancel Response

Tool to cancel a background model response by its ID.

Cancel Run

Request cancellation of an in-progress assistant run by thread and run ID, for example when the run is no longer needed.

Cancel upload

Tool to cancel an upload.

Compact Response

Tool to compact a conversation or response to reduce token usage.

Complete Upload

Complete a multipart Upload and create its usable File object.

Create Audio Transcription

Transcribes audio files to text via the OpenAI Audio Transcriptions API, optionally returning timestamps.

Create Audio Translation

Tool to translate audio files to English text via OpenAI Audio Translations API.

Create Batch

Tool to create and execute a batch from an uploaded file of requests.

Create Chat Completion

Creates a chat completion from OpenAI models for conversational AI, text generation, function calling, vision or audio tasks, and structured JSON outputs.

Create Completion (Legacy)

Generate single-turn text completions using OpenAI's legacy Completions API with an available compatible model.

Create Container

Tool to create a new container with configurable memory, expiration, file access, and network policies.

Create Container File

Tool to create a file in a container.

Create Conversation

Tool to create a new conversation for multi-turn interactions.

Create Conversation Items

Tool to create items in a conversation with the given ID.

Create Decision

Evaluate evidence with typed predicate, choice, and score questions.

Create Embeddings

Tool to generate text embeddings via the OpenAI embeddings endpoint.

Create Eval

Tool to create an evaluation structure for testing a model's performance.

Create Evaluation Run

Creates an evaluation run for testing with a specified data source and model configuration.

Create fine-tuning job

Creates a fine-tuning job from an uploaded training dataset and returns job details and status.

Generate Image

Tool to generate an image via the OpenAI Images API and return hosted image asset URL and metadata.

Edit Image

Tool to create edited or extended images via OpenAI Images Edit API.

Create Image Variation

Create alternative versions of an existing square PNG image using the legacy image variations operation, with options for count, size, and response format.

Create Message

Create a message in an Assistants conversation thread.

Create Moderation

Tool to classify text and/or image inputs for potentially harmful content via the OpenAI Moderation API.

Create Realtime Call

Creates a Realtime API call over WebRTC and returns the SDP answer needed to complete the peer connection.

Create Realtime Client Secret

Tool to create an ephemeral client secret for authenticating Realtime API connections.

Create Realtime Session

Tool to create an ephemeral API token for client-side Realtime API applications.

Create Realtime Transcription Session

Tool to create an ephemeral API token for realtime transcriptions via the Realtime API.

Create Response

Creates a model response via the Responses API for text or image analysis and structured JSON output.

Create Run

Create an assistant run on an existing thread to execute the assistant and generate responses, optionally adding messages and setting run options.

Create Skill

Tool to create a skill from uploaded files.

Create Speech (TTS)

Tool to generate text-to-speech audio using OpenAI's Audio API.

Create Thread

Create an Assistants API conversation thread with optional starter messages and metadata.

Create Thread And Run

Create an Assistants thread and run in one request to start a conversation and execute an assistant.

Create Upload

Tool to create an intermediate Upload object for large file uploads.

Create Vector Store

Tool to create a new vector store.

Create Vector Store File

Tool to create a vector store file by attaching a File to a vector store.

Create vector store file batch

Tool to create a vector store file batch.

Create Video

Create an asynchronous video generation job from a text prompt using the OpenAI Videos API; the job ID and status were used to track generation.

Create Video Remix

Submit a remix job for an existing generated video using a new prompt.

Delete assistant

Delete a specific Assistants API assistant by ID.

Delete chat completion

Tool to delete a stored chat completion by its ID.

Delete container

Tool to delete a specific container by its ID.

Delete container file

Tool to delete a file from a container.

Delete conversation

Tool to delete a conversation by its ID.

Delete conversation item

Tool to delete an item from a conversation with the given IDs.

Delete evaluation

Tool to delete a specific evaluation by its ID.

Delete evaluation run

Deletes a specific evaluation run.

Delete file

Deletes a file by its ID.

Delete message

Delete a specific message from an Assistants conversation thread using the thread and message IDs.

Delete response

Tool to delete a model response with the given ID.

Delete skill

Tool to delete a specific skill by its ID.

Delete thread

Delete an Assistants API conversation thread by its ID.

Delete Vector Store

Tool to delete a vector store.

Delete Vector Store File

Tool to delete a vector store file.

Delete video

Delete a video by ID from OpenAI's Videos API, returning a deletion result.

Download file

Tool to download the contents of a specified file by its ID.

Download Video Content

Download a generated video's MP4 content or its thumbnail or spritesheet by video ID.

Get Chat Completion

Tool to retrieve a stored chat completion.

Get Chat Completion Messages

Tool to retrieve messages from a stored chat completion.

Get ChatKit thread

Tool to retrieve a ChatKit thread by its ID.

Get Conversation Item

Tool to retrieve a single item from a conversation.

Get Eval

Tool to retrieve an evaluation by ID.

Get Evaluation Run

Tool to retrieve an evaluation run by ID to check status and results.

Get Eval Run Output Item

Tool to retrieve a specific output item from an evaluation run by its ID.

Get eval run output items

Tool to get a list of output items for an evaluation run.

Get Evaluation Runs

Tool to get a paginated list of runs for an evaluation.

Get Input Token Counts

Tool to calculate input token counts for OpenAI API requests.

Get Message

Retrieve a specific message from an Assistants thread by thread and message IDs to inspect its content and details.

Get Response

Tool to retrieve a model response by ID.

Get Run Step

Retrieve a specific Assistants run step by thread, run, and step ID to inspect execution progress, tool calls, or message creation.

Get Vector Store

Tool to retrieve a vector store by its ID.

Get Vector Store File

Tool to retrieve a file from a vector store.

Get Vector Store File Batch

Retrieve a file batch by its vector store and batch IDs to check the status and file counts of a batch processing operation.

Get Video

Retrieve a video generation job by ID to inspect its status, progress, and details.

List Assistants

List Assistants API assistants to discover an assistant ID from returned names or metadata.

List Batches

Tool to list your organization's batches.

List Chat Completions

Tool to list stored chat completions that were created with the `store` parameter set to true.

List ChatKit thread items

List items from a ChatKit thread, including messages, tool calls, and tasks.

List container files

List files in a specific container.

List Containers

List a page of containers, optionally filtering by name.

List Conversation Items

List a page of items in a conversation by ID for review or processing.

List engines

Tool to list available engines and their basic information.

List Evals

Tool to list evaluations for a project.

List files

Lists files in the authenticated organization/project context.

List Files in Vector Store Batch

Tool to list vector store files in a batch.

List fine-tunes

Lists fine-tuning jobs for the authenticated context, with cursor pagination.

List fine-tuning job events

Tool to get status updates for a fine-tuning job.

List fine-tuning job checkpoints

Tool to list checkpoints for a fine-tuning job.

List Input Items

Tool to retrieve input items for a given response from the OpenAI Responses API.

List Messages

List messages in an Assistants thread, including outputs generated by a run.

List models

Lists models available to the current account or organization, with model identifiers and ownership metadata.

List Runs

List a page of Assistants runs belonging to a thread to inspect their status or execution history.

List Run Steps

List steps in an Assistants API run to inspect execution progress, tool calls, and message creation events.

List Skills

List skills with pagination and timestamp ordering, returning their IDs, names, and descriptions.

List ChatKit Threads

Tool to list ChatKit threads with pagination and filtering.

List Vector Store Files

Tool to list files in a vector store.

List Vector Stores

Tool to list vector stores to discover available vector stores by name or metadata.

List Videos

List video generation jobs and their statuses, with pagination and timestamp ordering, to review multiple tasks.

Modify Assistant

Modify an existing Assistants API assistant's configuration, such as its model, instructions, tools, or metadata.

Modify Message

Modify metadata on a specific message in an Assistants conversation thread, identified by thread and message IDs.

Modify Run

Update metadata on a specific Assistants run identified by its thread and run IDs.

Modify thread

Update metadata on an existing Assistants API thread identified by its thread ID.

Modify Vector Store

Tool to modify an existing vector store.

Retrieve assistant

Retrieve a specific assistant by ID to inspect its configuration and metadata.

Retrieve Batch

Tool to retrieve a batch by ID to check its status, progress, and results.

Retrieve container

Tool to retrieve details of a specific container by its ID.

Retrieve container file

Tool to retrieve metadata for a specific file in a container.

Retrieve container file content

Tool to retrieve the content of a file within a container.

Retrieve engine

Retrieve a legacy OpenAI engine by ID to inspect its availability and metadata.

Retrieve file

Tool to retrieve information about a specific file.

Retrieve fine-tuning job

Tool to retrieve information about a fine-tuning job.

Retrieve model

Tool to retrieve details of a specific model, confirming its metadata (ownership, created date) and verifying access under your org — a model appearing in OPENAI_LIST_MODELS does not guarantee access.

Retrieve run

Retrieve an Assistants run by thread and run ID to check its status, errors, and usage, such as when polling for completion before reading thread messages.

Retrieve thread

Retrieve an Assistants API thread by ID to inspect its metadata; message bodies are obtained separately.

Retrieve Vector Store File Content

Tool to retrieve the parsed contents of a vector store file.

Run grader

Tool to run a grader to evaluate model performance on a given sample.

Search Vector Store

Tool to search a vector store for relevant chunks based on a query and file attributes filter.

Submit Tool Outputs to Run

Submit all tool call outputs together to continue an Assistants API run whose required action is submit_tool_outputs.

Update Chat Completion

Tool to update metadata for a stored chat completion.

Update Conversation

Tool to update a conversation's metadata.

Update Eval

Updates an evaluation's name or metadata without changing its data source configuration or testing criteria.

Update Vector Store File Attributes

Tool to update custom attributes on a vector store file.

Upload file

Upload file bytes to OpenAI for a specified purpose, such as fine-tuning, and return the uploaded file's details.

Validate grader configuration

Tool to validate a grader configuration for fine-tuning jobs.

FAQ

Frequently asked questions

Start with Openai.It takes 30 seconds.

Managed auth, hosted MCP servers, and every Openai tool your agent needs.Free to start.

Start building