swarm Documentation

File: README.md Copy Chapter

Swarm (experimental, educational)

[!IMPORTANT]
Swarm is now replaced by the OpenAI Agents SDK, which is a production-ready evolution of Swarm. The Agents SDK features key improvements and will be actively maintained by the OpenAI team.

We recommend migrating to the Agents SDK for all production use cases.

Install Copy Chapter

Requires Python 3.10+

pip install git+ssh://[email protected]/openai/swarm.git

or

pip install git+https://github.com/openai/swarm.git

Usage Copy Chapter

from swarm import Swarm, Agent

client = Swarm()

def transfer_to_agent_b():
    return agent_b


agent_a = Agent(
    name="Agent A",
    instructions="You are a helpful agent.",
    functions=[transfer_to_agent_b],
)

agent_b = Agent(
    name="Agent B",
    instructions="Only speak in Haikus.",
)

response = client.run(
    agent=agent_a,
    messages=[{"role": "user", "content": "I want to talk to agent B."}],
)

print(response.messages[-1]["content"])

Hope glimmers brightly,
New paths converge gracefully,
What can I assist?

Table of Contents Copy Chapter

• Overview
• Examples
• Documentation
  • Running Swarm
  • Agents
  • Functions
  • Streaming
• Evaluations
• Utils

Overview

Swarm focuses on making agent coordination and execution lightweight, highly controllable, and easily testable.

It accomplishes this through two primitive abstractions: Agents and handoffs. An Agent encompasses instructions and tools, and can at any point choose to hand off a conversation to another Agent.

These primitives are powerful enough to express rich dynamics between tools and networks of agents, allowing you to build scalable, real-world solutions while avoiding a steep learning curve.

[!NOTE]
Swarm Agents are not related to Assistants in the Assistants API. They are named similarly for convenience, but are otherwise completely unrelated. Swarm is entirely powered by the Chat Completions API and is hence stateless between calls.

Why Swarm Copy Chapter

Swarm explores patterns that are lightweight, scalable, and highly customizable by design. Approaches similar to Swarm are best suited for situations dealing with a large number of independent capabilities and instructions that are difficult to encode into a single prompt.

The Assistants API is a great option for developers looking for fully-hosted threads and built in memory management and retrieval. However, Swarm is an educational resource for developers curious to learn about multi-agent orchestration. Swarm runs (almost) entirely on the client and, much like the Chat Completions API, does not store state between calls.

Examples

Check out /examples for inspiration! Learn more about each one in its README.

• basic: Simple examples of fundamentals like setup, function calling, handoffs, and context variables
• triage_agent: Simple example of setting up a basic triage step to hand off to the right agent
• weather_agent: Simple example of function calling
• airline: A multi-agent setup for handling different customer service requests in an airline context.
• support_bot: A customer service bot which includes a user interface agent and a help center agent with several tools
• personal_shopper: A personal shopping agent that can help with making sales and refunding orders

Documentation

Running Swarm Copy Chapter

Start by instantiating a Swarm client (which internally just instantiates an OpenAI client).

from swarm import Swarm

client = Swarm()

`client.run()` Copy Section

Swarm's run() function is analogous to the chat.completions.create() function in the Chat Completions API – it takes messages and returns messages and saves no state between calls. Importantly, however, it also handles Agent function execution, hand-offs, context variable references, and can take multiple turns before returning to the user.

At its core, Swarm's client.run() implements the following loop:

1. Get a completion from the current Agent
2. Execute tool calls and append results
3. Switch Agent if necessary
4. Update context variables, if necessary
5. If no new function calls, return

Arguments

Once client.run() is finished (after potentially multiple calls to agents and tools) it will return a Response containing all the relevant updated state. Specifically, the new messages, the last Agent to be called, and the most up-to-date context_variables. You can pass these values (plus new user messages) in to your next execution of client.run() to continue the interaction where it left off – much like chat.completions.create(). (The run_demo_loop function implements an example of a full execution loop in /swarm/repl/repl.py.)

`Response` Fields

Agents Copy Chapter

An Agent simply encapsulates a set of instructions with a set of functions (plus some additional settings below), and has the capability to hand off execution to another Agent.

While it's tempting to personify an Agent as "someone who does X", it can also be used to represent a very specific workflow or step defined by a set of instructions and functions (e.g. a set of steps, a complex retrieval, single step of data transformation, etc). This allows Agents to be composed into a network of "agents", "workflows", and "tasks", all represented by the same primitive.

`Agent` Fields Copy Chapter

Instructions Copy Section

Agent instructions are directly converted into the system prompt of a conversation (as the first message). Only the instructions of the active Agent will be present at any given time (e.g. if there is an Agent handoff, the system prompt will change, but the chat history will not.)

agent = Agent(
   instructions="You are a helpful agent."
)

The instructions can either be a regular str, or a function that returns a str. The function can optionally receive a context_variables parameter, which will be populated by the context_variables passed into client.run().

def instructions(context_variables):
   user_name = context_variables["user_name"]
   return f"Help the user, {user_name}, do whatever they want."

agent = Agent(
   instructions=instructions
)
response = client.run(
   agent=agent,
   messages=[{"role":"user", "content": "Hi!"}],
   context_variables={"user_name":"John"}
)
print(response.messages[-1]["content"])

Hi John, how can I assist you today?

Functions Copy Chapter

• Swarm Agents can call python functions directly.
• Function should usually return a str (values will be attempted to be cast as a str).
• If a function returns an Agent, execution will be transferred to that Agent.
• If a function defines a context_variables parameter, it will be populated by the context_variables passed into client.run().

def greet(context_variables, language):
   user_name = context_variables["user_name"]
   greeting = "Hola" if language.lower() == "spanish" else "Hello"
   print(f"{greeting}, {user_name}!")
   return "Done"

agent = Agent(
   functions=[greet]
)

client.run(
   agent=agent,
   messages=[{"role": "user", "content": "Usa greet() por favor."}],
   context_variables={"user_name": "John"}
)

Hola, John!

• If an Agent function call has an error (missing function, wrong argument, error) an error response will be appended to the chat so the Agent can recover gracefully.
• If multiple functions are called by the Agent, they will be executed in that order.

Handoffs and Updating Context Variables Copy Section

An Agent can hand off to another Agent by returning it in a function.

sales_agent = Agent(name="Sales Agent")

def transfer_to_sales():
   return sales_agent

agent = Agent(functions=[transfer_to_sales])

response = client.run(agent, [{"role":"user", "content":"Transfer me to sales."}])
print(response.agent.name)

Sales Agent

It can also update the context_variables by returning a more complete Result object. This can also contain a value and an agent, in case you want a single function to return a value, update the agent, and update the context variables (or any subset of the three).

sales_agent = Agent(name="Sales Agent")

def talk_to_sales():
   print("Hello, World!")
   return Result(
       value="Done",
       agent=sales_agent,
       context_variables={"department": "sales"}
   )

agent = Agent(functions=[talk_to_sales])

response = client.run(
   agent=agent,
   messages=[{"role": "user", "content": "Transfer me to sales"}],
   context_variables={"user_name": "John"}
)
print(response.agent.name)
print(response.context_variables)

Sales Agent
{'department': 'sales', 'user_name': 'John'}

[!NOTE]
If an Agent calls multiple functions to hand-off to an Agent, only the last handoff function will be used.

Function Schemas Copy Section

Swarm automatically converts functions into a JSON Schema that is passed into Chat Completions tools.

• Docstrings are turned into the function description.
• Parameters without default values are set to required.
• Type hints are mapped to the parameter's type (and default to string).
• Per-parameter descriptions are not explicitly supported, but should work similarly if just added in the docstring. (In the future docstring argument parsing may be added.)

def greet(name, age: int, location: str = "New York"):
   """Greets the user. Make sure to get their name and age before calling.

   Args:
      name: Name of the user.
      age: Age of the user.
      location: Best place on earth.
   """
   print(f"Hello {name}, glad you are {age} in {location}!")

{
   "type": "function",
   "function": {
      "name": "greet",
      "description": "Greets the user. Make sure to get their name and age before calling.\n\nArgs:\n   name: Name of the user.\n   age: Age of the user.\n   location: Best place on earth.",
      "parameters": {
         "type": "object",
         "properties": {
            "name": {"type": "string"},
            "age": {"type": "integer"},
            "location": {"type": "string"}
         },
         "required": ["name", "age"]
      }
   }
}

Streaming Copy Chapter

stream = client.run(agent, messages, stream=True)
for chunk in stream:
   print(chunk)

Uses the same events as Chat Completions API streaming. See process_and_print_streaming_response in /swarm/repl/repl.py as an example.

Two new event types have been added:

• {"delim":"start"} and {"delim":"end"}, to signal each time an Agent handles a single message (response or function call). This helps identify switches between Agents.
• {"response": Response} will return a Response object at the end of a stream with the aggregated (complete) response, for convenience.

Evaluations

Evaluations are crucial to any project, and we encourage developers to bring their own eval suites to test the performance of their swarms. For reference, we have some examples for how to eval swarm in the airline, weather_agent and triage_agent quickstart examples. See the READMEs for more details.

Utils

Use the run_demo_loop to test out your swarm! This will run a REPL on your command line. Supports streaming.

from swarm.repl import run_demo_loop
...
run_demo_loop(agent, stream=True)

Core Contributors

• Ilan Bigio - ibigio
• James Hills - jhills20
• Shyamal Anadkat - shyamal-anadkat
• Charu Jaiswal - charuj
• Colin Jarvis - colin-openai
• Katia Gil Guzman - katia-openai

File: examples/airline/README.md Copy Chapter

Airline customer service

This example demonstrates a multi-agent setup for handling different customer service requests in an airline context using the Swarm framework. The agents can triage requests, handle flight modifications, cancellations, and lost baggage cases.
This example uses the helper function run_demo_loop, which allows us to create an interactive Swarm session.

Agents Copy Chapter

1. Triage Agent: Determines the type of request and transfers to the appropriate agent.
2. Flight Modification Agent: Handles requests related to flight modifications, further triaging them into:
   • Flight Cancel Agent: Manages flight cancellation requests.
   • Flight Change Agent: Manages flight change requests.
3. Lost Baggage Agent: Handles lost baggage inquiries.

Setup Copy Chapter

Once you have installed dependencies and Swarm, run the example using:

python3 main.py

Evaluations Copy Chapter

[!NOTE]
These evals are intended to be examples to demonstrate functionality, but will have to be updated and catered to your particular use case.

For this example, we run function evals, where we input a conversation, and the expected function call ('None' if no function call is expected).
The evaluation cases are stored in eval/eval_cases/ subfolder.

[
  {
    "conversation": [
      { "role": "user", "content": "My bag was not delivered!" }
    ],
    "function": "transfer_to_lost_baggage"
  },
  {
    "conversation": [
      { "role": "user", "content": "I had some turbulence on my flight" }
    ],
    "function": "None"
  }
]

The script 'function_evals.py' will run the evals. Make sure to set n to the number
of times you want to run each particular eval. To run the script from the root airline folder, execute:

cd evals
python3 function_evals.py

The results of these evaluations will be stored in evals/eval_results/

File: examples/basic/README.md Copy Chapter

Swarm basic

This folder contains basic examples demonstrating core Swarm capabilities. These examples show the simplest implementations of Swarm, with one input message, and a corresponding output. The simple_loop_no_helpers has a while loop to demonstrate how to create an interactive Swarm session.

Examples Copy Section

1. agent_handoff.py

   • Demonstrates how to transfer a conversation from one agent to another.
   • Usage: Transfers Spanish-speaking users from an English agent to a Spanish agent.

2. bare_minimum.py

   • A bare minimum example showing the basic setup of an agent.
   • Usage: Sets up an agent that responds to a simple user message.

3. context_variables.py

   • Shows how to use context variables within an agent.
   • Usage: Uses context variables to greet a user by name and print account details.

4. function_calling.py

   • Demonstrates how to define and call functions from an agent.
   • Usage: Sets up an agent that can respond with weather information for a given location.

5. simple_loop_no_helpers.py

   • An example of a simple interaction loop without using helper functions.
   • Usage: Sets up a loop where the user can continuously interact with the agent, printing the conversation.

Running the Examples Copy Chapter

To run any of the examples, use the following command:

python3 <example_name>.py

File: examples/personal_shopper/README.md Copy Chapter

Personal shopper

This Swarm is a personal shopping agent that can help with making sales and refunding orders.
This example uses the helper function run_demo_loop, which allows us to create an interactive Swarm session.
In this example, we also use a Sqlite3 database with customer information and transaction data.

Overview Copy Chapter

The personal shopper example includes three main agents to handle various customer service requests:

1. Triage Agent: Determines the type of request and transfers to the appropriate agent.
2. Refund Agent: Manages customer refunds, requiring both user ID and item ID to initiate a refund.
3. Sales Agent: Handles actions related to placing orders, requiring both user ID and product ID to complete a purchase.

Setup Copy Chapter

Once you have installed dependencies and Swarm, run the example using:

python3 main.py

File: examples/support_bot/README.md Copy Chapter

Support bot

This example is a customer service bot which includes a user interface agent and a help center agent with several tools.
This example uses the helper function run_demo_loop, which allows us to create an interactive Swarm session.

Overview Copy Chapter

The support bot consists of two main agents:

1. User Interface Agent: Handles initial user interactions and directs them to the help center agent based on their needs.
2. Help Center Agent: Provides detailed help and support using various tools and integrated with a Qdrant VectorDB for documentation retrieval.

Setup Copy Chapter

To start the support bot:

1. Ensure Docker is installed and running on your system.
2. Install the necessary additional libraries:

make install

3. Initialize docker

docker-compose up -d

4. Prepare the vector DB:

make prep

5. Run the main scripy:

make run

File: examples/triage_agent/README.md Copy Chapter

Triage agent

This example is a Swarm containing a triage agent, which takes in user inputs and chooses whether to respond directly, or triage the request
to a sales or refunds agent.

Setup Copy Chapter

To run the triage agent Swarm:

1. Run

python3 run.py

Evals Copy Chapter

[!NOTE]
These evals are intended to be examples to demonstrate functionality, but will have to be updated and catered to your particular use case.

This example uses Pytest to run eval unit tests. We have two tests in the evals.py file, one which
tests if we call the correct triage function when expected, and one which assesses if a conversation
is 'successful', as defined in our prompt in evals.py.

To run the evals, run

pytest evals.py

File: examples/weather_agent/README.md Copy Chapter

Weather agent

This example is a weather agent demonstrating function calling with a single agent. The agent has tools to get the weather of a particular city, and send an email.

Setup Copy Chapter

To run the weather agent Swarm:

1. Run

python3 run.py

Evals Copy Chapter

[!NOTE]
These evals are intended to be examples to demonstrate functionality, but will have to be updated and catered to your particular use case.

This example uses Pytest to run eval unit tests. We have two tests in the evals.py file, one which
tests if we call the get_weather function when expected, and one which assesses if we properly do NOT call the
get_weather function when we shouldn't have a tool call.

To run the evals, run

pytest evals.py