MCP server to manage your Aranet4 CO2 sensor. Built upon Aranet4-Python.

• Scan for nearby devices.
• Fetch new data from embedded device memory and save to a local sqlite db for tracking and later viewing. For automatic updates, see at the bottom.
• Ask questions about recent measurements or about a specific past date.
• [For MCP clients that support images] Ask data to be plotted to also have a nice visualization!
• Assisted configuration! 💫 After installation, just ask init aranet4 in your client to set up everything for the mcp server to work with your device.

An Aranet4 device already paired. The device uses BLE, so it won't be shown in regular bluetooth search. On MacOS, one way is to install LightBlue from the App Store.

1. Clone repo:

   git clone git@github.com:diegobit/aranet4-mcp-server.git`
   cd aranet4-mcp-server
   

2. Prepare environment:

   • Recommended (with uv): Nothing to do. The provided pyproject.toml handles dependencied and virtual environments.
   • Alternative (with pip): install with pip install .

3. Add to MCP client configuration:

   "aranet4": {
     "command": "{{PATH_TO_UV}}", // run `which uv`
       "args": [
         "--directory",
         "{{PATH_TO_SRC}}/aranet4-mcp-server/",
         "run",
         "src/server.py"
       ]
   }

   • Claude Desktop MacOS config file path: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Cursor MacOS config file path: ~/.cursor/mcp.json

4. Configure:

   • Recommended (AI assisted config!): start your client and ask init aranet4 to get a guided configuration.
   • Alternative (manual): edit file config.yaml. You need to provide the mac address and the device name. You can get the mac address with aranetctl --scan from Aranet4-Python (installed with this repo dependencies).

Dockerfile is available. Remember to pass env variables or update config.yaml.

Configuration and utils:

• init_aranet4_config: assisted configuration of device.
• scan_devices: scan nearby bluetooth aranet4 devices.
• get_configuration_and_db_stats: get current config.yaml and general stats from the local sqlite3 db.
• set_configuration: set values in config.yaml.

To update historical data:

• fetch_new_data: fetch new data from configured nearby aranet4 device and save to local db.

To query historical data:

• get_recent_data: get recent data from local db. Can specify how many measurements.

• get_data_by_timerange: get data in specific timerange from local db. Can specify how many measurements (careful, if the range is big and the limit is low, datapoints will be skipped).

  For both, ask to receive a plot to have it generated and displayed.

If you want your local db to always be updated, you can setup a launch agent (macOS), a systemd timer (Linux) or a cronjob that fetches data automatically every few hours.

1. Configure absolute paths in com.diegobit.aranet4-fetch.plist.
2. Install LaunchAgent:

   cp com.diegobit.aranet4-fetch.plist ~/Library/LaunchAgents/
   launchctl load ~/Library/LaunchAgents/com.diegobit.aranet4-fetch.plist

1. Configure absolute paths in systemd/aranet4-fetch.service.
2. Install and enable the timer:

   mkdir -p ~/.config/systemd/user
   cp systemd/aranet4-fetch.service systemd/aranet4-fetch.timer ~/.config/systemd/user/
   systemctl --user daemon-reload
   systemctl --user enable --now aranet4-fetch.timer

   The fetch runs once on activation and then every 3 hours, catching up on runs missed while the machine was off. To keep it running without an open login session, enable lingering with loginctl enable-linger $USER.

Alternatively, add this line to your crontab (crontab -e):

0 */3 * * * cd /home/diego/code/aranet4-mcp-server && /home/diego/.local/share/mise/shims/uv run fetch-job.py