Skip to content

ANTA as a Python Library

ANTA is a Python library that can be used in user applications. This section describes how you can leverage ANTA Python modules to help you create your own NRFU solution.


If you are unfamiliar with asyncio, refer to the Python documentation relevant to your Python version -

AntaDevice Abstract Class

A device is represented in ANTA as a instance of a subclass of the AntaDevice abstract class. There are few abstract methods that needs to be implemented by child classes:

  • The collect() coroutine is in charge of collecting outputs of AntaCommand instances.
  • The refresh() coroutine is in charge of updating attributes of the AntaDevice instance. These attributes are used by AntaInventory to filter out unreachable devices or by AntaTest to skip devices based on their hardware models.

The copy() coroutine is used to copy files to and from the device. It does not need to be implemented if tests are not using it.

AsyncEOSDevice Class

The AsyncEOSDevice class is an implementation of AntaDevice for Arista EOS. It uses the aio-eapi eAPI client and the AsyncSSH library.

  • The collect() coroutine collects AntaCommand outputs using eAPI.
  • The refresh() coroutine tries to open a TCP connection on the eAPI port and update the is_online attribute accordingly. If the TCP connection succeeds, it sends a show version command to gather the hardware model of the device and updates the established and hw_model attributes.
  • The copy() coroutine copies files to and from the device using the SCP protocol.

AntaInventory Class

The AntaInventory class is a subclass of the standard Python type dict. The keys of this dictionary are the device names, the values are AntaDevice instances.

AntaInventory provides methods to interact with the ANTA inventory:

To parse a YAML inventory file and print the devices connection status:

import asyncio

from anta.inventory import AntaInventory

async def main(inv: AntaInventory) -> None:
    Take an AntaInventory and:
    1. try to connect to every device in the inventory
    2. print a message for every device connection status
    await inv.connect_inventory()

    for device in inv.values():
        if device.established:
            print(f"Device {} is online")
            print(f"Could not connect to device {}")

if __name__ == "__main__":
    # Create the AntaInventory instance
    inventory = AntaInventory.parse(

    # Run the main coroutine
    res =
How to create your inventory file

Please visit this dedicated section for how to use inventory and catalog files.

To run an EOS commands list on the reachable devices from the inventory:

# This is needed to run the script for python < 3.10 for typing annotations
from __future__ import annotations

import asyncio
from pprint import pprint

from anta.inventory import AntaInventory
from anta.models import AntaCommand

async def main(inv: AntaInventory, commands: list[str]) -> dict[str, list[AntaCommand]]:
    Take an AntaInventory and a list of commands as string and:
    1. try to connect to every device in the inventory
    2. collect the results of the commands from each device

      a dictionary where key is the device name and the value is the list of AntaCommand ran towards the device
    await inv.connect_inventory()

    # Make a list of coroutine to run commands towards each connected device
    coros = []
    # dict to keep track of the commands per device
    result_dict = {}
    for name, device in inv.get_inventory(established_only=True).items():
        anta_commands = [AntaCommand(command=command, ofmt="json") for command in commands]
        result_dict[name] = anta_commands

    # Run the coroutines
    await asyncio.gather(*coros)

    return result_dict

if __name__ == "__main__":
    # Create the AntaInventory instance
    inventory = AntaInventory.parse(

    # Create a list of commands with json output
    commands = ["show version", "show ip bgp summary"]

    # Run the main asyncio  entry point
    res =, commands))


Use tests from ANTA

All the test classes inherit from the same abstract Base Class AntaTest. The Class definition indicates which commands are required for the test and the user should focus only on writing the test function with optional keywords argument. The instance of the class upon creation instantiates a TestResult object that can be accessed later on to check the status of the test ([unset, skipped, success, failure, error]).

Test structure

All tests are built on a class named AntaTest which provides a complete toolset for a test:

  • Object creation
  • Test definition
  • TestResult definition
  • Abstracted method to collect data

This approach means each time you create a test it will be based on this AntaTest class. Besides that, you will have to provide some elements:

  • name: Name of the test
  • description: A human readable description of your test
  • categories: a list of categories to sort test.
  • commands: a list of command to run. This list must be a list of AntaCommand which is described in the next part of this document.

Here is an example of a hardware test related to device temperature:

from __future__ import annotations

import logging
from typing import Any, Dict, List, Optional, cast

from anta.models import AntaTest, AntaCommand

class VerifyTemperature(AntaTest):
    Verifies device temparture is currently OK.

    # The test name
    name = "VerifyTemperature"
    # A small description of the test, usually the first line of the class docstring
    description = "Verifies device temparture is currently OK"
    # The category of the test, usually the module name
    categories = ["hardware"]
    # The command(s) used for the test. Could be a template instead
    commands = [AntaCommand(command="show system environment temperature", ofmt="json")]

    # Decorator
    # abstract method that must be defined by the child Test class
    def test(self) -> None:
        """Run VerifyTemperature validation"""
        command_output = cast(Dict[str, Dict[Any, Any]], self.instance_commands[0].output)
        temperature_status = command_output["systemStatus"] if "systemStatus" in command_output.keys() else ""
        if temperature_status == "temperatureOk":
            self.result.is_failure(f"Device temperature is not OK, systemStatus: {temperature_status }")

When you run the test, object will automatically call its anta.models.AntaTest.collect() method to get device output for each command if no pre-collected data was given to the test. This method does a loop to call anta.inventory.models.InventoryDevice.collect() methods which is in charge of managing device connection and how to get data.

run test offline

You can also pass eos data directly to your test if you want to validate data collected in a different workflow. An example is provided below just for information:

test = VerifyTemperature(device, eos_data=test_data["eos_data"])

The test function is always the same and must be defined with the @AntaTest.anta_test decorator. This function takes at least one argument which is a anta.inventory.models.InventoryDevice object. In some cases a test would rely on some additional inputs from the user, for instance the number of expected peers or some expected numbers. All parameters must come with a default value and the test function should validate the parameters values (at this stage this is the only place where validation can be done but there are future plans to make this better).

class VerifyTemperature(AntaTest):
    def test(self) -> None:

class VerifyTransceiversManufacturers(AntaTest):
    def test(self, manufacturers: Optional[List[str]] = None) -> None:
        # validate the manufactures parameter

The test itself does not return any value, but the result is directly available from your AntaTest object and exposes a anta.result_manager.models.TestResult object with result, name of the test and optional messages:

  • name (str): Device name where the test has run.
  • test (str): Test name runs on the device.
  • categories (List[str]): List of categories the TestResult belongs to, by default the AntaTest categories.
  • description (str): TestResult description, by default the AntaTest description.
  • results (str): Result of the test. Can be one of [“unset”, “success”, “failure”, “error”, “skipped”].
  • message (str, optional): Message to report after the test if any.
  • custom_field (str, optional): Custom field to store a string for flexibility in integrating with ANTA
from anta.tests.hardware import VerifyTemperature

test = VerifyTemperature(device, eos_data=test_data["eos_data"])
assert test.result.result == "success"

Classes for commands

To make it easier to get data, ANTA defines 2 different classes to manage commands to send to devices:

AntaCommand Class

Represent a command with following information:

  • Command to run
  • Output format expected
  • eAPI version
  • Output of the command

Usage example:

from anta.models import AntaCommand

cmd1 = AntaCommand(command="show zerotouch")
cmd2 = AntaCommand(command="show running-config diffs", ofmt="text")

Command revision and version

  • Most of EOS commands return a JSON structure according to a model (some commands may not be modeled hence the necessity to use text outformat sometimes.
  • The model can change across time (adding feature, … ) and when the model is changed in a non backward-compatible way, the revision number is bumped. The initial model starts with revision 1.
  • A revision applies to a particular CLI command whereas a version is global to an eAPI call. The version is internally translated to a specific revision for each CLI command in the RPC call. The currently supported version values are 1 and latest.
  • A revision takes precedence over a version (e.g. if a command is run with version=”latest” and revision=1, the first revision of the model is returned)
  • By default, eAPI returns the first revision of each model to ensure that when upgrading, integrations with existing tools are not broken. This is done by using by default version=1 in eAPI calls.

By default, ANTA uses version="latest" in AntaCommand, but when developing tests, the revision MUST be provided when the outformat of the command is json. As explained earlier, this is to ensure that the eAPI always returns the same output model and that the test remains always valid from the day it was created. For some commands, you may also want to run them with a different revision or version.

For instance, the VerifyBFDPeersHealth test leverages the first revision of show bfd peers:

# revision 1 as later revision introduce additional nesting for type
commands = [AntaCommand(command="show bfd peers", revision=1)]

AntaTemplate Class

Because some command can require more dynamic than just a command with no parameter provided by user, ANTA supports command template: you define a template in your test class and user provide parameters when creating test object.

class RunArbitraryTemplateCommand(AntaTest):
    Run an EOS command and return result
    Based on AntaTest to build relevant output for pytest

    name = "Run aributrary EOS command"
    description = "To be used only with anta debug commands"
    template = AntaTemplate(template="show interfaces {ifd}")
    categories = ["debug"]

    def test(self) -> None:
        errdisabled_interfaces = [interface for interface, value in response["interfaceStatuses"].items() if value["linkStatus"] == "errdisabled"]

params = [{"ifd": "Ethernet2"}, {"ifd": "Ethernet49/1"}]
run_command1 = RunArbitraryTemplateCommand(device_anta, params)

In this example, test waits for interfaces to check from user setup and will only check for interfaces in params