Home Assistant Quality & Testing

Use this skill to align changes with Home Assistant best practices and the Integration Quality Scale expectations.

When to Use

• •Planning or reviewing changes for config flows, options, or reauth
• •Deciding how to structure polling, discovery, and entities
• •Ensuring translations, metadata, and versioning are correct
• •Setting up or updating tests and CI
• •Driving toward Bronze/Silver IQS targets (config flow coverage, >95% overall coverage, diagnostics)

Architectural Principles

• •Async-only: Never block the event loop; offload sync work via hass.async_add_executor_job.
• •Coordinator-first: Centralize device I/O in DataUpdateCoordinator; entities consume coordinator.data.
• •Single source of discovery: Rely on declared discovery (dhcp/zeroconf/ssdp) or dedicated scanner; avoid ad-hoc discovery inside polling.
• •Separation: Keep protocol/library logic outside entities; prefer a library (requirements) for raw API/UDP/HTTP handling.
• •Quality Scale focus: Bronze requires 100% config_flow coverage and connection tests; Silver pushes >95% overall coverage, strict typing, and reauth flows.

Coordinator Error Handling Patterns

from homeassistant.exceptions import ConfigEntryAuthFailed
from homeassistant.helpers.update_coordinator import UpdateFailed

async def _async_update_data(self):
    try:
        return await self.api.fetch_data()
    except AuthError as err:
        # Triggers reauth flow automatically
        raise ConfigEntryAuthFailed("Authentication failed") from err
    except RateLimitError:
        # Backoff with retry_after (seconds until retry)
        raise UpdateFailed(retry_after=60)
    except ConnectionError as err:
        raise UpdateFailed(f"Connection failed: {err}")

Coordinator Best Practices

• •Use _async_setup() for one-time initialization during first refresh
• •Set always_update=False if your data supports __eq__ comparison (avoids unnecessary entity updates)
• •Use async_contexts() to track which entities are actively listening
• •Pass config_entry to coordinator constructor for automatic linking

Config / Options / Reauth Patterns

• •UI-first: No new YAML; all setup via config_flow.py with selectors where helpful.
• •Duplicate avoidance: Abort if a device is already configured; update IP/host on existing entries when discovery reports changes.
• •Options reload: Register entry.add_update_listener to reload on options change.
• •Reauth: Trigger entry.async_start_reauth; ask only for the changed credential and update the existing entry.
• •Duplicate enforcement: Set unique_id early and call _abort_if_unique_id_configured() for user/discovery steps.

Entities & Registries

• •_attr_has_entity_name = True is MANDATORY for new integrations; entity name should be capability-only (device name is prepended by HA).
• •For the "main feature" entity of a device, set _attr_name = None to use only the device name.
• •Provide device_info with stable identifiers (prefer MAC/serial over IP); all entities for a device must share the same identifiers set.
• •Keep unique_id stable and predictable (e.g., {ble_mac}_{key}); changing it breaks history and customizations.
• •Only create entities for data that actually exists to avoid permanent unavailable noise.
• •Use _attr_* class/instance attributes pattern for cleaner code:
  python

  class MySensor(SensorEntity):
      _attr_has_entity_name = True
      _attr_device_class = SensorDeviceClass.POWER
      _attr_native_unit_of_measurement = UnitOfPower.WATT
  

EntityDescription Pattern (Recommended)

For multiple similar entities, use EntityDescription for declarative definitions:

@dataclass(kw_only=True)
class MySensorEntityDescription(SensorEntityDescription):
    value_fn: Callable[[DeviceData], StateType]
    exists_fn: Callable[[DeviceData], bool] = lambda _: True

SENSORS: tuple[MySensorEntityDescription, ...] = (
    MySensorEntityDescription(
        key="power",
        device_class=SensorDeviceClass.POWER,
        native_unit_of_measurement=UnitOfPower.WATT,
        value_fn=lambda data: data.power,
    ),
)

Entity Best Practices

• •Use constants from homeassistant.const for units:
  • •UnitOfEnergy.WATT_HOUR not "Wh"
  • •UnitOfPower.WATT not "W"
  • •UnitOfTemperature.CELSIUS not "°C"
• •Use device classes from homeassistant.components.sensor:
  • •SensorDeviceClass.BATTERY for battery level sensors
  • •SensorDeviceClass.POWER for power sensors
  • •SensorDeviceClass.ENERGY for energy sensors
  • •SensorDeviceClass.ENERGY_STORAGE for stored energy (battery capacity in Wh)
  • •SensorDeviceClass.TEMPERATURE for temperature sensors
• •Use state classes appropriately:
  • •SensorStateClass.MEASUREMENT for instantaneous values (power, temperature)
  • •SensorStateClass.TOTAL for values that can increase/decrease (net energy)
  • •SensorStateClass.TOTAL_INCREASING for cumulative counters that only increase
• •Use EntityCategory.DIAGNOSTIC for non-primary sensors (WiFi RSSI, temperatures)
• •Set entity_registry_enabled_default = False for diagnostic or rarely-used sensors
• •Use suggested_display_precision to control decimal places shown in UI
• •Use _unrecorded_attributes frozenset to exclude high-frequency attributes from recorder

Restoring Sensor State

Use RestoreSensor (not RestoreEntity) to restore sensor state after restart:

from homeassistant.components.sensor import RestoreSensor

class MyEnergySensor(RestoreSensor):
    async def async_added_to_hass(self) -> None:
        await super().async_added_to_hass()
        if (last := await self.async_get_last_sensor_data()):
            self._attr_native_value = last.native_value

Diagnostics (Gold Quality Scale)

• •Create diagnostics.py with async_get_config_entry_diagnostics function
• •Return a dict with: entry data, device_info, coordinator_data, last_update_success, last_exception
• •Redact sensitive data before returning:
  • •IPs, MACs, SSIDs, tokens, passwords
  • •Use a TO_REDACT list pattern for consistency:
    python

    TO_REDACT = {"host", "ip", "mac", "ble_mac", "wifi_mac", "wifi_name", "SSID", "bleMac"}
    
    def _redact_dict(data: dict) -> dict:
        return {k: "**REDACTED**" if k in TO_REDACT else v for k, v in data.items()}
    

• •Test diagnostics with snapshot tests and verify redaction

Quality Scale Tracking

Integrations working toward Bronze/Silver/Gold should maintain a quality_scale.yaml:

rules:
  config_flow: done
  test_before_setup: done
  unique_config_entry: done
  diagnostics:
    status: done
    comment: Added in v0.2.0
  reauthentication-flow:
    status: exempt
    comment: Device has no authentication

Statuses: done, todo, exempt (with comment explaining why).

Manifest & Metadata

• •Pin requirements (e.g., pymarstek==x.y.z) to avoid breaking upgrades.
• •Set version, config_flow: true, iot_class, and codeowners; keep documentation and issue tracker URLs current.
• •For HACS, keep releases/tagging consistent and add hacs.json if distribution via HACS.

Translations

• •Author strings in strings.json; mirror to translations/en.json.
• •Use descriptive error keys (cannot_connect, invalid_auth, already_configured).
• •Prefer placeholders for dynamic content (e.g., {ip_address}) to keep translations flexible.

Testing & CI

• •Use pytest with pytest-homeassistant-custom-component; pin test deps in requirements_test.txt.
• •Test layout: tests/ mirrors component files (test_config_flow.py, test_init.py, test_sensor.py, etc.); put shared fixtures in tests/conftest.py (enable custom integrations).
• •Config flow: cover success + cannot_connect + invalid_auth/invalid_discovery_info + already_configured; assert unique_id and aborts.
• •Coordinator/entities: mock transport; assert happy path + errors raise UpdateFailed and surface unavailable states; gate entities on data keys.
• •Actions/commands: verify polling is paused, retries fire, verification logic works, and failures bubble.
• •Snapshot/diagnostics (Gold path): use syrupy HA extension for diagnostics/device registry dumps; redact sensitive fields.
• •CI: hassfest + lint (ruff, mypy) + pytest with coverage threshold (e.g., --cov-fail-under=95); test latest supported Python versions.

Verification After Changes (MANDATORY)

After every code change, run both checks before considering the work complete:

# 1. Type checking (strict mode)
python3 -m mypy --strict custom_components/<domain>/

# 2. Tests with coverage
pytest tests/ -q --cov=custom_components/<domain> --cov-fail-under=95

Both must pass. Fix any errors and re-run until clean.

Operational Hygiene

• •Debounce manual refreshes with coordinator.async_request_refresh().
• •Pause polling when sending control commands that reuse the same transport to avoid concurrent traffic.
• •Log warnings/errors with actionable context (host, method) but avoid noisy debug logs by default.

Quick Checklist

• • No blocking I/O on the event loop
• • Coordinator is the sole reader/writer to the device
• • Config/Options/Reauth flows implemented and reload on options change
• • Stable unique IDs + device identifiers
• • Translations updated (strings + en.json)
• • Requirements pinned; manifest fields valid; hassfest/HACS clean
• • Tests cover config flow, coordinator happy-path and failure, and entity states
• • mypy --strict passes with no errors
• • All tests pass with >95% coverage