iharmony 6.0.2

Test the web app, the API behind it, the data it writes and the devices it talks to —
with one pytest-native framework, one configuration model and one set of reports.

Why iHarmony™

Quality teams usually stitch together a UI tool, an API tool, a data-diff script and a hardware harness, each with its own conventions and reports. iHarmony™ puts all of them on pytest + Playwright with a shared page-object model, YAML configuration, observability and a release quality gate — so a 5-person team and a 100-person programme work the same way.

• One workflow — UI, API, database, file and device tests are ordinary pytest tests.
• Enterprise by default — multi-project layouts, environment overlays, org → business-unit → project governance, secrets from environment variables or vaults, never from code.
• Resilient — self-healing locators, retries and flaky-test detection cut maintenance noise.
• Decision-ready — every run ends in reports for engineers, QA and leadership, plus an optional PASS / WARN / BLOCK verdict your pipeline can act on.
• Opt-in everything — install only the extras you need; features are off until enabled in YAML.

Install

pip install iharmony
playwright install chromium        # or: playwright install  (all browsers)

Requires Python 3.11+. Add extras for the capabilities you use:

Quick start

iharmony init my-qe-suite       # scaffold a suite with runnable demos
cd my-qe-suite

export IHARMONY_TEST_USERNAME=standard_user    # demo login for saucedemo.com
export IHARMONY_TEST_PASSWORD=secret_sauce

pytest tests/demo/functional/api tests/demo/functional/ui

On Windows PowerShell, set the credentials with:

$env:IHARMONY_TEST_USERNAME = "standard_user"
$env:IHARMONY_TEST_PASSWORD = "secret_sauce"

The scaffold contains configuration, page objects and demo UI, API and database tests. The data-comparison demos also need pip install "iharmony[data-comparison]".

What tests look like

Web UI — locators live in page objects; tests read like the business flow.

import pytest
from iharmony import BasePage


class LoginPage(BasePage):
    USERNAME = "#user-name"
    PASSWORD = "#password"
    SUBMIT = "#login-button"

    def login(self, user: str, password: str) -> None:
        self.fill(self.USERNAME, user)
        self.fill(self.PASSWORD, password)
        self.click(self.SUBMIT)


@pytest.mark.smoke
def test_login(page, runtime_options, login_test_data):
    login = LoginPage(page, runtime_options["resilience_policy"])
    login.navigate(runtime_options["base_url"])
    user = login_test_data["valid_user"]          # values resolved from ${ENV_VAR} placeholders
    login.login(user["username"], user["password"])

API — an HTTP client with retries, auth and chainable assertions.

from iharmony import ApiClient, ResponseValidator


def test_create_order(base_url):
    api = ApiClient(base_url=base_url, retries=3)
    response = api.post("/orders", json_body={"sku": "A-100", "qty": 2})
    ResponseValidator.assert_status_code(response, 201)
    ResponseValidator.assert_json_has_keys(response, ["id", "status"])
    ResponseValidator.assert_response_time_less_than(response, 800)

Data — compare extracts, files and tables with keys, tolerances and column mappings.

from utils.data_comparison_utils import compare_excel_files


def test_migration_matches_legacy():
    result = compare_excel_files(
        "legacy_extract.xlsx", "new_platform_extract.xlsx",
        key_columns=["employee_id"], float_tolerance=0.01, ignore_fields={"updated_at"},
    )
    assert result.passed, result.summary    # interactive HTML report at result.report_path

Devices — the same test runs against a digital twin or real hardware.

import pytest


@pytest.mark.device
@pytest.mark.requires_dut(capability="thermostat")
def test_heats_up(dut):
    dut.set_target(22)
    dut.wait_until_temperature(22, timeout=900)

pytest -m device                                   # simulated twin
pytest -m device --iharmony-dut-mode hil        # real hardware, same test

Run it your way

Every --iharmony-<name> option has the short alias --ih-<name>.

CI templates for GitHub Actions, Jenkins and Azure DevOps are included (iharmony docs ci).

AI-ready: MCP server

iHarmony ships a Model Context Protocol server so AI assistants such as Claude and GitHub Copilot can discover the framework API, list tests and page objects, read resolved configuration and test results, and get a release recommendation. Every tool is also a terminal command.

pip install "iharmony[mcp]"
iharmony-mcp serve                         # MCP server (stdio)
iharmony-mcp search-methods tolerance      # the same tools from the terminal
iharmony-mcp get-release-recommendation

Documentation

Guides are installed with the package — no network or repository access needed:

iharmony docs                  # list every topic
iharmony docs quickstart       # print one

Support

iHarmony is developed and supported by Innominds. For onboarding, enterprise support or custom integrations, contact iHarmony@innominds.com or visit innominds.com.

_{iHarmony™ is a trademark of Innominds. © Innominds. Released under the MIT License.}