Skip to content

Usage Patterns

Weather Signal has three independent choices that shape a workflow:

  1. How locations are identified.
  2. Which weather surface is used.
  3. Which output transport is consumed.

Use the narrowest combination that answers the forecasting question. Start with summary for context, move to signal or threshold for row-level features, and use daily, hourly, or historical when you need raw weather variables.

Location Inputs

Mode Example Best For
Saved alias weather-signal signal london Recurring business locations
Place search weather-signal signal "London" --country GB One-off city or region lookups
Coordinates weather-signal daily "51.5074,-0.1278" Deterministic joins and known sites
CSV batch weather-signal batch signal --input locations.csv Multi-site forecasting runs

Saved aliases are the most stable option for agent workflows because the resolved location is explicit in every response and can be reviewed in config.

Weather Surfaces

Surface Command Use When
Current weather current A workflow needs the current conditions only
Forecast variables daily, hourly You need raw Open-Meteo fields
Demand features signal You need daily features that join cleanly to demand data
Forecast summary summary You need a compact overview for planning or agent context
Decision rules threshold You need days that match specific weather triggers
Historical archive historical You need backtesting or feature-store inputs

Output Transports

Transport Example Best For
CLI JSON weather-signal signal london Agents, scripts, and repeatable jobs
CLI table weather-signal daily london --table Human inspection
CLI CSV weather-signal hourly london --output csv Spreadsheets and feature stores
MCP stdio weather-signal server start --transport stdio Local MCP clients
MCP streamable HTTP weather-signal server start --transport streamable-http Local services behind an auth proxy
Profile Location Command Output
Single-site planning Saved alias summary JSON
Feature engineering Coordinates or saved alias signal JSON
Trigger-based operations Saved alias threshold JSON
Multi-site demand forecast CSV or saved places batch signal JSON
Analyst exploration Place search daily --table Table
Backtesting Coordinates historical --output csv CSV

Validation Checklist

weather-signal geocode london --country GB --count 1
weather-signal places add london "London" --country GB
weather-signal summary london --days 7
weather-signal threshold london --days 7 --rain-prob-gte 60
weather-signal cache status

For production agent loops, prefer saved aliases or coordinates, keep stdout in JSON, and route logs from stderr separately.