Building a Signal-Driven Weather Microservice

Oct 27, 2025

#fastapi #docker #automation #signal #microservices #open-meteo

Overview

weather-service is a FastAPI microservice that retrieves weather data from Open-Meteo and sends daily forecasts through the notifier gateway. It exposes two endpoints—/today and /health—for both scheduled and on-demand use. This container is part of a modular automation stack built around Signal and the assistant-core.

Features

Scheduled forecasts via cron (0 7 * * * by default)
/today?city=Orlando&state=FL returns a text summary and raw forecast data
Timezone-aware through TZ
Sends notifications through notifier-gateway
Runs inside the assistant-net Docker network

Quick Start

Add the service to your Compose stack:

weather-service:
  build: ./weather-service
  restart: unless-stopped
  env_file: [.env]
  environment:
    CRON_SCHEDULE: "0 7 * * *"
    NOTIFY_URL: "http://notifier-gateway:8787/send"
    NOTIFY_TO: "+15555551234"
    TZ: "America/New_York"
    CITY: "Orlando"
    STATE: "FL"
    LAT: 28.5383
    LON: -81.3792
  depends_on: [notifier-gateway]
  ports: ["127.0.0.1:8789:8789"]
  networks: [assistant-net]

Copy the environment file and build:

cp .env.example .env
docker compose build weather-service
docker compose up -d weather-service

API

Method	Endpoint	Description
`GET`	`/health`	Service configuration and status
`GET`	`/today?city=&state=`	Returns the current forecast

Example:

curl "http://localhost:8789/today?city=Orlando&state=FL"

Response:

{
  "city": "Orlando, Florida, United States",
  "message": "Good morning! Orlando forecast for Monday, Oct 27\nClear\nHigh 85° / Low 67° • 💧 10%",
  "raw": { ... }
}

Operation

At startup, a scheduler computes the next run time based on CRON_SCHEDULE. It triggers once immediately and then at the configured interval, sending forecasts to the notifier gateway. The /today endpoint uses the same logic for real-time access.

Integration

This service can be polled by assistant-core or other tools to generate daily summaries or respond to /weather commands in Signal. It can also feed structured data into an LLM for multi-source morning briefs.