feat: add AirQualityInput plugin with multi-connector pattern (AQICN, PMS5003, BME680)

[Wanbogang]

Overview

Adds a generic AirQualityInput plugin that supports multiple air quality sensors and APIs via a connector pattern. This allows OM1 to read real-time air quality data without changing the core plugin when switching sensors. #365

Features

• AirQualityInput plugin — reads standardized AirQualityData and converts it to human-readable text for the LLM
• Connector pattern — swap sensors via config, no code changes needed
• 3 connectors out of the box:
  • aqicn — AQICN World Air Quality Index cloud API (no hardware needed)
  • pms5003 — PMS5003/PMS7003 particulate matter sensor via Serial/UART
  • bme680 — BME680 environmental sensor (temperature, humidity, VOC) via I2C
• AQI alerts — WARNING and DANGER thresholds configurable per deployment
• Async & non-blocking — serial/I2C reads run in executor to avoid blocking event loop

Sensors Integrated

Connector	Sensor	Interface	Data
aqicn	AQICN API	HTTP	AQI, PM2.5, PM10, CO, NO2, SO2, O3, temp, humidity
pms5003	PMS5003/PMS7003	Serial/UART	PM2.5, PM10, AQI
bme680	BME680	I2C	Temperature, humidity, IAQ score

Configuration Example

# AQICN (cloud, no hardware)
connector: aqicn
connector_config:
  api_key: "your_token_here"
  latitude: -6.9667
  longitude: 110.4167

# PMS5003 (hardware)
connector: pms5003
connector_config:
  port: /dev/ttyUSB0
  location: Outdoor

# BME680 (hardware)
connector: bme680
connector_config:
  i2c_address: 0x76
  location: Indoor
  gas_baseline: 50000.0

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ All tests successful. No failed tests found.

📢 Thoughts on this report? Let us know!

[Prachi1615]

Hi @Wanbogang
Thanks for your PR.
If you have videos with working demo; please attach them here, after you resolve the conflicts.

[Wanbogang]

Hi @Prachi1615

thanks for the review! Conflicts are now resolved (merged latest main), and all 104 tests pass.

I tested the aqicn connector end-to-end against the live AQICN API (no hardware needed for this one). The screenshot below shows both the connector read and the full OM1 pipeline.

Setup: AirQualityInput plugin with connector: aqicn, using a free AQICN token, pointed at my location (Semarang, Indonesia: lat -6.9667, lon 110.4167).

What the screenshot shows:

1. Calling AqicnConnector.read() directly returns live data:
   AirQualityData(aqi=99, pm25=99, temperature=32.9, humidity=53.1, location='Semarang, Indonesia', source='aqicn')

2. Running the agent in OM1, the plugin polls the API and feeds formatted data into the LLM:

INPUT: Air Quality
Air Quality in Semarang, Indonesia: MODERATE (AQI: 99). PM2.5: 99 µg/m³. Temperature: 32.9°C, Humidity: 53.1%.

So the AQICN API → connector → AirQualityInput → LLM path works with real-time data. The AQI level classification (MODERATE) and the WARNING/DANGER threshold logic are applied in the plugin.

The pms5003 and bme680 connectors are hardware sensors (Serial/I2C) which I can't demo without the physical devices, but they share the same AirQualityConnector interface and are fully unit-tested (100% coverage).

Let me know if you'd like anything else. Thanks!