#12666: fix(weather): add precipitation cross-check guidance to prevent misleading rain reports
stale
Cluster:
Skill Enhancements and Fixes
## Problem
The weather skill's Open-Meteo section only shows how to fetch `current_weather`. When agents use `precipitation_probability_max` for rain forecasts, the API can return misleadingly high values (e.g. 65%) even when `precipitation_sum` is 0mm -- meaning no actual rain is expected.
This is especially bad during dry seasons. For example, Mexico City in February (deep dry season) returns 65% probability with 0.00mm precipitation sum. An agent naively reporting "65% chance of rain" would be completely wrong.
## Fix
Added a "Precipitation: Always Cross-Check" section to the weather SKILL.md:
- **Cross-check table**: always compare `precipitation_sum` vs `probability_max` before reporting rain
- **Rule**: if `precipitation_sum` is 0mm, ignore the probability entirely
- **Example API call** that fetches both fields together
- **Seasonal awareness** notes for dry/wet season cities where probability is especially unreliable
## Testing
Verified with Open-Meteo API for Mexico City (Feb 9, 2026):
- `precipitation_probability_max: 65`
- `precipitation_sum: 0.00`
- With the new guidance, the agent correctly ignores the phantom probability.
<!-- greptile_comment -->
<h2>Greptile Overview</h2>
<h3>Greptile Summary</h3>
This PR updates `skills/weather/SKILL.md` to add explicit guidance for agents using Open‑Meteo daily precipitation fields.
It introduces a new “Precipitation: Always Cross‑Check” section that (1) warns against reporting `precipitation_probability_max` in isolation, (2) provides an example Open‑Meteo request that fetches both `precipitation_probability_max` and `precipitation_sum`, and (3) adds a simple interpretation table plus seasonal notes (e.g., dry-season cities) to reduce misleading “chance of rain” reports when forecast precipitation totals are 0 mm.
<h3>Confidence Score: 5/5</h3>
- This PR is safe to merge with minimal risk.
- Only documentation was changed (weather skill guidance). The added section is self-contained, does not affect runtime behavior, and follows existing doc conventions in the repository.
- No files require special attention
<!-- greptile_other_comments_section -->
<sub>(4/5) You can add custom instructions or style guidelines for the agent [here](https://app.greptile.com/review/github)!</sub>
**Context used:**
- Context from `dashboard` - CLAUDE.md ([source](https://app.greptile.com/review/custom-context?memory=fd949e91-5c3a-4ab5-90a1-cbe184fd6ce8))
- Context from `dashboard` - AGENTS.md ([source](https://app.greptile.com/review/custom-context?memory=0d0c8278-ef8e-4d6c-ab21-f5527e322f13))
<!-- /greptile_comment -->
Most Similar PRs
#8873: fix: document skill loading locations in system prompt
by ytfh44 · 2026-02-04
69.6%
#13594: fix(skills): correct apple-notes SKILL.md memo -a syntax
by lailoo · 2026-02-10
67.7%
#9925: docs: warn about Python venv placement in skill directories
by kj9kj9qv4m-bit · 2026-02-05
67.7%
#5273: fix: skill frontmatter fixes + description improvements (35 skills)
by Terwox · 2026-01-31
67.5%
#9595: fix(skills): ignore .venv, __pycache__, and .openclaw to prevent FD...
by amoghacloud · 2026-02-05
66.9%
#8291: Fix: Add Python virtual environment ignore patterns to skills watcher
by vishaltandale00 · 2026-02-03
65.9%
#15115: fix: pre-load skill docs in cron sessions to prevent hallucinated syn…
by joaolcorreia · 2026-02-13
65.3%
#22306: Warn on malformed skill parsing failures in load path
by AIflow-Labs · 2026-02-21
65.0%
#19664: fix(skills): log skill YAML parsing diagnostics with skill name
by orchidsun · 2026-02-18
64.9%
#22410: docs: Add Agent Cloud Memory to community skills
by lilyjazz · 2026-02-21
64.9%