mcp-airq

MCP server for air-Q air quality sensor devices. Enables Claude Desktop, Claude Code, and other MCP clients to directly query and configure air-Q devices on your local network.

Built on aioairq, the official async Python library for air-Q.

The same mcp-airq executable also works as a direct CLI when you pass a tool name as a subcommand.

Installation

pip install mcp-airq

Or run directly with uvx:

uvx mcp-airq

CLI Usage

Use the same command directly from the shell:

mcp-airq list-devices
mcp-airq get-air-quality --device "Living Room"
mcp-airq set-night-mode --activated --device "Bedroom"

The CLI subcommands mirror the MCP tool names. Both styles work:

mcp-airq list-devices
mcp-airq list_devices

To force MCP server mode from an interactive terminal, run:

mcp-airq serve

The CLI is pipe-friendly: successful command output goes to stdout, while tool errors go to stderr with exit code 1.

mcp-airq get-air-quality --device "Living Room" | jq '.co2'
mcp-airq get-air-quality --device "Living Room" --compact-json | jq '.co2'
mcp-airq get-air-quality --device "Living Room" --yaml | yq '.co2'

Device Configuration

Create a JSON file with your device(s), e.g. ~/.config/airq-devices.json:

[
  {"address": "192.168.4.1", "password": "your_password", "name": "air-Q Pro", "location": "Living Room", "group": "Home"},
  {"address": "192.168.4.2", "password": "your_password", "name": "air-Q Radon", "location": "Living Room", "group": "Home"},
  {"address": "office_air-q.local", "password": "other_pass", "name": "Office", "group": "Work"}
]

Each entry requires:

• address — IP address or mDNS hostname (e.g. abcde_air-q.local)
• password — Device password (default: airqsetup)
• name (optional) — Human-readable name; defaults to address
• location (optional) — Physical room/area for grouping (e.g. "Living Room")
• group (optional) — Second grouping dimension, orthogonal to location (e.g. "Home", "Work")

Then restrict access to the file (it contains passwords):

chmod 600 ~/.config/airq-devices.json

Alternatively, pass the device list inline via the AIRQ_DEVICES environment variable as a JSON string.

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "airq": {
      "command": "uvx",
      "args": ["mcp-airq"],
      "env": {
        "AIRQ_CONFIG_FILE": "/home/you/.config/airq-devices.json"
      }
    }
  }
}

Claude Code

Register the server once via the CLI:

claude mcp add airq -e AIRQ_CONFIG_FILE=~/.config/airq-devices.json -- uvx mcp-airq

This writes to ~/.claude/settings.json and is automatically picked up by the Claude Code VSCode extension as well — no separate configuration needed.

If the server fails to connect: MCP servers run in a subprocess that may not inherit your shell's PATH. Replace uvx with its full path (which uvx → e.g. /home/you/.local/bin/uvx):

claude mcp add airq -e AIRQ_CONFIG_FILE=~/.config/airq-devices.json -- /home/you/.local/bin/uvx mcp-airq

OpenAI Codex

Register the server once via the CLI:

codex mcp add airq --env AIRQ_CONFIG_FILE=~/.config/airq-devices.json -- uvx mcp-airq

This writes to ~/.codex/config.toml and is automatically picked up by the Codex VSCode extension as well.

If the server fails to connect: Use the full path to uvx (see note above).

Available Tools

Read-Only

Tool	Description
list_devices	List all configured air-Q devices (with location/group if set)
get_air_quality	Get sensor readings — by device, location, or group
get_device_info	Get device metadata (name, model, firmware version)
get_config	Get full device configuration
get_logs	Get device log entries
identify_device	Make device blink its LEDs for visual identification
get_led_theme	Get current LED visualization theme
get_possible_led_themes	List all available LED visualization themes
get_night_mode	Get current night mode configuration
get_brightness_config	Get current LED brightness configuration

Configuration

Tool	Description
set_device_name	Rename a device
set_led_theme	Change LED visualization (CO₂, VOC, Humidity, PM2.5, …)
set_night_mode	Configure night mode schedule and settings
set_brightness	Adjust LED brightness (day/night)
configure_network	Set static IP or switch to DHCP

Device Control

Tool	Description
restart_device	Restart the device (~30s downtime)
shutdown_device	Shut down the device (manual restart required)

Multi-Device Support

When multiple devices are configured, specify which device to query:

• By exact name: "air-Q Pro"
• By partial match (case-insensitive): "pro", "radon"

If only one device is configured, it is selected automatically.

Location and Group Queries

get_air_quality accepts two optional grouping parameters:

• location — query all devices in the same room (e.g. "Living Room")
• group — query all devices sharing a group tag (e.g. "Home")

Both are independent: a device can have a location, a group, both, or neither. Matching is case-insensitive and substring-based.

get_air_quality(location="Living Room")  → air-Q Pro + air-Q Radon
get_air_quality(group="Home")            → air-Q Pro + air-Q Radon + …
get_air_quality(device="air-Q Radon")   → just that one device

Exactly one of device, location, or group may be specified per call.

Example Prompts

• "How is the air quality in the living room?" — queries all devices at that location
• "What's the air quality at home?" — queries all devices in the "Home" group
• "Show me the radon level" — targets the air-Q Radon device by name
• "Show CO₂ on the LEDs"
• "Enable night mode from 10 PM to 7 AM"
• "Set brightness to 50%"
• "What's in the device log?"
• "Make the air-Q blink"

Development

git clone https://github.com/CorantGmbH/mcp-airq.git
cd mcp-airq
uv sync --frozen --extra dev
uv run pre-commit install
uv run pytest

The repository uses a project-local .venv plus uv.lock for reproducible tooling. Run all developer commands through uv run, for example:

uv run ruff check .
uv run ruff format --check .
uv run pyright
uv run pre-commit run --all-files

Release Process

1. Update version in pyproject.toml.
2. Commit and create a matching Git tag like v0.1.1.
3. Publish a GitHub Release from that tag.

The publish workflow validates that the release tag matches pyproject.toml, uploads the package to PyPI, and then publishes the same version to the MCP Registry.

License

Apache License 2.0 — see LICENSE.