Skip to main content

Framework for multiagent systems development in Python

Project description

Python Agent DEvelopment framework (PADE)


PADE its a framework for developing, executing and mannaging multi-agent systems in distributed computing enviroments. PADE code is 100% Python and has its core in Twisted, a python package for implementing distributed applications.

PADE is also free software and licenced in terms of MIT licence. First it was developed in Federal University of Ceará (Brazil) by Electric Smart Grids Group (GREI) in Electric Engineering Department (DEE).

The researchers of Laboratory of Applied Artificial Intelligence (LAAI) of Federal University of Para (UFPA) have been contributed a lot with PADE project. We registre here our acknowledgments.

Everyone that has interest in developing PADE is welcome to download, install, test, use and send us feedback.

Scientific Paper

There is a scientific paper presenting PADE as a scientific tool for multiagent system simulation with focus in electric power systems simulation. If you have interest here is the link to access:

Python‐based multi‐agent platform for application on power grids

If you use PADE in your research work, please cite PADE as:

Melo, LS, Sampaio, RF, Leão, RPS, Barroso, GC, Bezerra, JR. Python‐based multi‐agent platform for application on power grids. Int Trans Electr Energ Syst. 2019; 29:e12012.‐7038.12012


PADE is well documented. You can access the documentation here: PADE documentation


PADE is developed in Python 3.7 and has a Twisted core.


Via Python Package Index (PyPI):

$ pip install pade

Via Github:

$ git clone
$ cd pade
$ python install

See the complete process in this video: HOW TO install PADE


Build container

$ docker-compose up -d

List containers

$ docker ps

8d7cb00972c9        pade_pade

Get inside container

$ docker exec -it <CONTAINER_ID> bash


Hello world in PADE:

from pade.misc.utility import display_message, start_loop
from pade.core.agent import Agent
from pade.acl.aid import AID
from sys import argv

class AgenteHelloWorld(Agent):
    def __init__(self, aid):
        super(AgenteHelloWorld, self).__init__(aid=aid)
        display_message(self.aid.localname, 'Hello World!')

if __name__ == '__main__':

    agents_per_process = 3
    c = 0
    agents = list()
    for i in range(agents_per_process):
        port = int(argv[1]) + c
        agent_name = 'agente_hello_{}@localhost:{}'.format(port, port)
        agente_hello = AgenteHelloWorld(AID(name=agent_name))
        c += 1000


Changes in this new version

Some changes has been added in this new version, but don't worry about that if you are using pade in your simulations, it's very easy adjust this version in old versions.

The main and bigger change in Pade is in how you launch your agents. Now when you install Pade via pip command or via install you install a pade terminal command line (cli) that launch your pade applications.

Before we start an PADE example, it's important to create the initial database file in the temporary folder of your Operating System. As the temporary folder is cleared each time that you shutdown your OS you will need to run this command in each reinitialization of your OS before executing PADE agents:

$ pade create-pade-db

If everything is well, than this message will apperar in your prompt:

[...] Creating Pade tables in selected data base.
[ok_] Tables created in selected data base

As example, if you put the hello world example code in a file with the name and you want to launch this agent just one time, you could type in your command line interface:

$ pade start-runtime 

If you want to launch this agent 3 times, than you type:

$ pade start-runtime --num 3 

If you wanto to launch the 3 agents in ports 20000, 20001 and 20002, than you just type:

$ pade start-runtime --num 3 --port 20000 

Here we have to explain some points in how Pade executes the agents.

When you type the commands --num 3 and --port 20000 you tell to Pade command line tool to execute the content of file 3 times. Each time, the file content will be executed in a new process and the attribute port will be passed as argument in this process with a unit incremment in each time. For example, in the case --num 3 and --port 2000, the arguments passed for agents are 2000, 2001 and 2002.

This arguments should be accessed in the code with sys.argv[1]. So you can execute how many agents as you want per process. In the example there is a for loop that will repeat many times as defined in agents_per_process variable. That will define the number of agents in each process. In the example, since the --num parameter is 3 and the agents_per_process variable is 3 the pade will start 9 agents in ports: 20000, 21000, 22000, 20001, 210001, 22001, 20002, 210002 and 22002.

The command line will support mode than one agent file too, for example if you have the agents in mode than one file you could start then with a command like this:

$ pade start-runtime --num 3 --port 20000

In this case the first agent receive in the sys.argv[1] the value 20000 and the second, the value 20001, and so on.

There is another way to launch the Pade agents. Is with a config file in the json format. Here it's a example of config file:

    "agent_files": [
    "port": 20000,
    "num": 2,
    "pade_ams": {
        "launch": true,
        "host": "localhost",
        "port": 8000
    "pade_web": {
        "active": true,
        "host": "localhost",
        "port": 5000
    "pade_sniffer": {
        "active": true,
        "host": "localhost",
        "port": 8001
    "session": {
        "username": "pade_user",
        "email": "",
        "password": "12345"    

To launch then, just type the command line:

pade start-runtime --config_file pade_config.json

If you need to execute simulations with a high number of agents that send and receive messages, something like 500 agents sending 5 messages per second, is recommended that you launch your pade session with a option --no_pade_sniffer because the register of this messages in database will overhead your pade execution. Than, the example could be:

$ pade start-runtime --num 3 --port 20000  --no_pade_sniffer

Another useful commands in Pade CLI are:

$ pade create-pade-db
$ pade drop-pade-db
$ pade start-web-interface

To show a complete list of pade comands in the CLI, just type pade in terminal command line.

To show the agents in action, show the video in this link: pade agents start example

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pade-2.2.4.tar.gz (39.1 kB view hashes)

Uploaded Source

Built Distribution

pade-2.2.4-py3-none-any.whl (187.6 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page