A Home Assistant custom integration that produces a curated list of your entities — filtered by area and domain, with the junk entities (diagnostic/config, hidden, and disabled) dropped — so you can hand a clean, actionable inventory to an LLM for automation design. By default each entity keeps all of its attributes so the LLM sees the full picture; you can switch on noise-stripping (or an allowlist) when you want a leaner payload.
It ships with a custom Lovelace card for point-and-click exports and a service/action you can call from Developer Tools, scripts, or automations.
Home Assistant already classifies most of the "junk" for you in the entity registry, so this integration leans on that instead of a brittle blocklist:
| What | How it's dropped |
|---|---|
| Diagnostic / config entities (RSSI, battery %, uptime, firmware, etc.) | entity_category is diagnostic/config → dropped by default |
| Hidden entities | hidden_by is set → dropped by default |
| Disabled entities | disabled_by is set → dropped by default |
Attributes are kept in full by default (include_all_attributes: true) so
the LLM sees everything each entity exposes — supported features, color modes,
and the like. Turn include_all_attributes off to strip a built-in noise
list (rssi, voltage, linkquality, battery, supported_features, attribution, …)
from each entity's attribute payload.
On top of that you can still pass manual overrides (exclude_entities,
exclude_attributes) and an allowlist (include_attributes, which overrides
the above and keeps only the listed attributes). Each entity auto-filter can be
turned off per-call (include_diagnostic, include_hidden, include_disabled).
Battery readings: "battery" is a
device_classon thesensordomain, not a domain of its own. Battery sensors are usually tagged diagnostic, so to include them pick thesensordomain (or leave domains empty) and turn oninclude_diagnostic.
With include_automations, include_scripts, and include_scenes on (all
default), the export also contains every automation, script, and scene with its
full config (triggers/conditions/actions/sequence/target states, mode),
current state, and the entities, devices, and areas it references. This
gives a downstream LLM both halves of the picture — the devices available and
what's already automated against them — for a much better analysis. It covers
both UI-created and YAML items.
Logic follows the area selection using containment: when you select one or
more areas, an automation/script/scene is included only if every area it
touches is within your selection. So a room-specific automation shows up for its
room, but a house-wide one (say a "smart plug offline" alert spanning six areas)
does not get pulled into a single-room export just because it also touches
that room. To get everything, leave the area filter empty. Each logic record
reports the resolved areas it touches, so the relevance is visible in the
output.
An item's areas are resolved from everything it references — the areas of its referenced entities and devices, plus any directly-referenced areas. An automation that references only unassigned entities has no resolvable area and is excluded while an area filter is active (it appears in an all-areas export).
- In HACS, open the three-dot menu → Custom repositories.
- Add
https://github.com/cobragt2000/data_exporterwith category Integration. - Search for Entity Data Exporter, install it, and restart Home Assistant.
- Go to Settings → Devices & Services → Add Integration, search for
Entity Data Exporter, and add it. (It's set up via the UI — no
configuration.yamledit needed.)
The Lovelace card is bundled and auto-registered — no manual resource entry needed.
Copy custom_components/data_exporter/ into your HA config/custom_components/
directory, restart, then add the integration from Settings → Devices &
Services → Add Integration.
Add a card to any dashboard:
type: custom:data-exporter-card
title: Entity Data Exporter # optionalPick areas and domains (both multi-selects list every area and domain in your Home Assistant; leave empty for "all"), toggle whether to include diagnostic/hidden/disabled entities and automations/scripts/scenes, choose the file format (JSON or CSV), and hit Run Export. The card reports the entity count and the file path.
Developer Tools → Actions → data_exporter.export_data, or in YAML:
action: data_exporter.export_data
data:
areas:
- living_room
- kitchen
domains:
- light
- switch
- climate
include_diagnostic: false
write_file: true
scope: downstairsTo get the data back inline (shown in Developer Tools, or usable in a script
variable) add return_response: true:
action: data_exporter.export_data
data:
domains: [light, switch]
response_variable: curatedThe response and the written file both look like:
{
"metadata": {
"timestamp": "2026-10-08T10:15:00",
"scope": "downstairs",
"entity_count": 42,
"filters": { "areas": ["living_room"], "domains": ["light"], "...": "..." }
},
"entities": [
{
"entity_id": "light.living_room_lamp",
"name": "Living Room Lamp",
"domain": "light",
"area": "Living Room",
"device": "Hue Lamp",
"device_class": null,
"state": "on",
"attributes": { "brightness": 180, "color_temp_kelvin": 2700 }
}
],
"automations": [
{
"entity_id": "automation.evening_lights",
"alias": "Evening Lights",
"id": "1700000000000",
"state": "on",
"mode": "single",
"last_triggered": "2026-10-07T23:01:00+00:00",
"referenced_entities": ["light.living_room_lamp"],
"referenced_devices": [],
"referenced_areas": ["living_room"],
"config": {
"alias": "Evening Lights",
"trigger": [{ "platform": "sun", "event": "sunset" }],
"condition": [],
"action": [{ "service": "light.turn_on", "target": { "entity_id": "light.living_room_lamp" } }],
"mode": "single"
}
}
]
}Files are written to the config directory as
ha_entities_<scope>_<timestamp>.json (or .csv when format: csv).
With format: csv, the written file is the flat entity list only — one row
per entity with columns entity_id, name, domain, area, device, device_class, state, attributes (the attributes dict is JSON-encoded into that single
column). Automations, scripts, and scenes are nested structures and stay in the
JSON export; the inline service response is always the full structured JSON
regardless of format.
| Field | Type | Default | Description |
|---|---|---|---|
areas |
list | all | Area ids or friendly names to include |
domains |
list | all | Domains to include — any HA domain (light, switch, climate, sensor, …). The card lists every domain in your HA; the action's dropdown lists the standard domains and accepts custom values too |
include_diagnostic |
bool | false |
Include diagnostic/config entities |
include_hidden |
bool | false |
Include hidden entities |
include_disabled |
bool | false |
Include disabled entities |
include_automations |
bool | true |
Include full automation configs + what they reference (area-filtered) |
include_scripts |
bool | true |
Include full script configs + what they reference (area-filtered) |
include_scenes |
bool | true |
Include scene configs + what they reference (area-filtered) |
include_all_attributes |
bool | true |
Keep every attribute per entity; set false to strip the built-in noise list |
exclude_attributes |
list | — | Attribute names to strip (applied on top of the above) |
include_attributes |
list | — | Allowlist: keep only these attributes (overrides the two above) |
exclude_entities |
list | — | Specific entity_ids to drop |
write_file |
bool | true |
Write the export file to config |
format |
string | json |
Written file format: json (full export) or csv (flat entity list) |
scope |
string | export |
Label used in filename + metadata |
- Home Assistant 2024.7.0 or newer.
MIT — see LICENSE.
