Specify exactly one of:
- 'device' — query a single device by name
- 'location' — query all devices at a given location (e.g. "Wohnzimmer")
- 'group' — query all devices in a group (e.g. "zu Hause")

When using 'location' or 'group', the response contains one entry per
device. Returns sensor names mapped to values. Set return_average=True
for time-averaged data (recommended) or False for instantaneous readings.
The response includes a _sensor_guide field with full unit and index
documentation — read it before interpreting any values.

The response includes a _config_guide field with full documentation of
all configuration keys — read it before interpreting or modifying values.

IMPORTANT — 'sensors' must be a JSON array, not a plain string.
  Correct:   sensors=["pm1","pm2_5"]
  Wrong:     sensors="pm1"

IMPORTANT — response size: air-Q records every ~2 minutes, so long ranges
produce large responses (24 h ≈ 720 readings × ~25 sensors). Always use
'sensors' and 'max_points' when querying more than 1–2 hours to stay within
response size limits. Example for a 24 h chart: sensors=["pm1","pm2_5","pm10"],
max_points=150.

Time range — specify one of:
- 'last_hours' — data from the last N hours (default: 1 hour)
- 'from_datetime' / 'to_datetime' — ISO 8601 strings
  (e.g. "2026-03-10T14:00:00" or "2026-03-10T14:00:00+01:00")
  'from_datetime' takes precedence over 'last_hours'.
  'to_datetime' defaults to now.
- 'timezone_name' — optional IANA timezone such as "Europe/Berlin".
  Naive datetimes are interpreted in this timezone. Output timestamps are
  localized into `datetime` using the same timezone.

Optional filtering:
- 'sensors' — list of sensor names to include (e.g. ["pm1", "pm2_5", "pm10"]).
  Omit to get all sensors.
- 'max_points' — downsample to at most this many evenly spaced points.

Response: column-oriented JSON with `timestamp` (Unix seconds) and localized
`datetime` columns. Compound sensor values like `[value, quality]` are split
into `<sensor>` and `<sensor>_quality`. Includes `_sensor_guide` and
`_history_guide`.

Selector:
- `device` — one specific device
- `location` — all devices at one location
- `group` — all devices in one group
- if none is specified, all configured devices are exported together

OUTPUT FORMAT:
- "csv" — one UTF-8 CSV file containing all selected devices
- "xlsx" — one Excel workbook containing all selected devices

REQUIRED:
- sensor: one sensor key to export, for example `co2`, `pm2_5`, `radon`

TIME RANGE:
- `last_hours` or `from_datetime` / `to_datetime`
- `timezone_name` controls how timestamps are rendered in the exported file

WHEN TO USE THIS TOOL: Call this whenever the user asks to see a graph,
chart, plot, or visual representation of historical sensor data.

Selector:
- `device` — one specific device
- `location` — all devices at one location
- `group` — all devices in one group
- if none is specified, all configured devices are plotted together

OUTPUT FORMAT:
- "png" (default) — one inline image containing all selected devices
- "webp" — inline image with smaller payload size
- "svg" — vector graphic as downloadable MCP resource
- "html" — self-contained interactive HTML as downloadable MCP resource

REQUIRED:
- sensor: the sensor key to visualise (one sensor per chart)

TIME RANGE:
- `last_hours` or `from_datetime` / `to_datetime`
- `timezone_name` controls how timestamps are rendered on the X axis and
  how naive input datetimes are interpreted

Common themes: 'standard', 'CO2', 'VOC', 'Humidity', 'PM2.5', 'Noise'.
Use get_possible_led_themes to see all available themes.

brightness_day/brightness_night are percentages (0-100).
fan_night_off disables the particle sensor fan at night.
wifi_night_off caches data to SD and uploads when wifi returns.
alarm_night_off disables acoustic warnings (fire/gas still trigger).