weather: The Zero-Config OpenClaw Skill That Gives Agents Real Weather Data Instantly

55,000+ downloads and 189 stars — weather by @steipete (Peter Steinberger, the same author behind the gog Google Workspace skill) is the most downloaded weather skill on ClawHub, and its secret is the most unsexy possible one: it requires absolutely nothing to set up.

No API key. No account. No rate-limit dashboard. Two curl commands and your agent has weather data.

The Problem With "Just Use a Weather API"

Most weather data skills come with friction. OpenWeatherMap, WeatherAPI, Tomorrow.io — all require account creation, API key management, and rate-limit monitoring. For a feature that should feel as simple as asking "what's the weather in Tokyo?", the setup overhead is disproportionate.

Steipete's design choice is deliberate: route everything through services that are already free and don't require authentication. The two services — wttr.in and Open-Meteo — handle the heavy lifting. The skill just teaches the agent the right curl syntax for each.

Two Services, One Skill

wttr.in (Primary)

A terminal-friendly weather service that returns human-readable output from a simple URL pattern. No API key, no JSON parsing required for basic use.

Quick one-liner:

curl -s "wttr.in/London?format=3"
# Output: London: ⛅️ +8°C

Compact multi-field format:

curl -s "wttr.in/London?format=%l:+%c+%t+%h+%w"
# Output: London: ⛅️ +8°C 71% ↙5km/h

Full 3-day forecast:

curl -s "wttr.in/London?T"

Save as PNG image:

curl -s "wttr.in/Berlin.png" -o /tmp/weather.png

Format codes reference:

URL tips:

• Spaces: use + → wttr.in/New+York
• Airport codes work: wttr.in/JFK, wttr.in/NRT
• Units: ?m for metric, ?u for USCS (Imperial)
• Forecast range: ?1 for today only, ?0 for current conditions only

Open-Meteo (Fallback / Programmatic)

When the agent needs JSON output for further processing, Open-Meteo provides structured weather data with no authentication:

# London coordinates
curl -s "https://api.open-meteo.com/v1/forecast?latitude=51.5&longitude=-0.12&current_weather=true"

Returns JSON with temperature, windspeed, winddirection, and weathercode. Useful when weather data feeds into another computation — decision logic, notifications, IoT triggers — rather than being displayed directly.

Full Open-Meteo docs: open-meteo.com/en/docs

When Each Service Fits

Real Usage Patterns

Personal assistant — daily briefing:

curl -s "wttr.in/Tokyo?format=3"
# → Tokyo: 🌤 +18°C

Travel planning — check multiple cities:

for city in London Paris Berlin Tokyo; do
  curl -s "wttr.in/${city}?format=%l:+%c+%t"
done

Automation — conditional logic on temperature:

TEMP=$(curl -s "https://api.open-meteo.com/v1/forecast?latitude=48.85&longitude=2.35&current_weather=true" | jq '.current_weather.temperature')
echo "Paris current temp: ${TEMP}°C"

Forecast before outdoor event:

# 3-day forecast for New York
curl -s "wttr.in/New+York?1"

How to Install

# Via ClawHub CLI
clawhub install steipete/weather

No additional setup. The agent can immediately use wttr.in and Open-Meteo queries after installation.

Practical Tips

Start with wttr.in format=3 for most agent use cases. It's the cleanest output — one line, location + emoji + temperature — and parses trivially if the agent needs to extract the number.

Use airport codes for travel contexts. wttr.in/LAX is more reliable than wttr.in/Los+Angeles when the user specifies a flight or airport. Supports all IATA codes.

Switch to Open-Meteo when the downstream code needs structured data. If the weather result is feeding into a decision, a comparison, or any further computation, JSON is cleaner than parsing the wttr.in text format.

For IoT or home automation scripts, use Open-Meteo's coordinates approach. Find the lat/lon once for a fixed location, hardcode it, and query on a schedule. Latency is sub-100ms, no rate limits documented for normal use.

The PNG output is surprisingly useful. curl -s "wttr.in/City.png" -o /tmp/weather.png produces a full forecast card image. Useful for dashboards, reports, or embedding forecast visuals without building a charting layer.

Considerations

wttr.in is a community service. It's run by Igor Chubin as an open-source project. It has been reliable for years, but it's not an enterprise SLA. For production systems where weather data is mission-critical, the free tier of a dedicated API (OpenWeatherMap, Tomorrow.io) may be more appropriate.

Open-Meteo has undocumented rate limits. For typical agent use — a few queries per session — there's no issue. Heavy programmatic use (thousands of queries per hour) should use their official API with a key.

Coordinate lookup is manual with Open-Meteo. Unlike wttr.in, which accepts city names and airport codes directly, Open-Meteo needs latitude and longitude. The agent needs to know or look up coordinates separately.

No historical data. Both services focus on current conditions and short-term forecasts. For historical weather queries, a different data source is required.

The Bigger Picture

Weather is a deceptively good test of how skills should be designed. The use case is universal — nearly every context-aware agent will eventually need to answer "what's the weather like?" — but the implementation doesn't need to be complex.

Most solutions to this problem reach for an API, create a dependency, and add setup friction. Steipete's approach is the opposite: find the simplest possible service that already works without authentication, write the minimal amount of curl syntax to use it, and ship.

The result is a skill that takes zero seconds to configure, works anywhere curl is available, and handles 95% of weather query needs. The remaining 5% can use Open-Meteo.

55,000 downloads suggests the OpenClaw community agreed: sometimes the best solution is the one that doesn't require a settings file.

View the skill on ClawHub: weather