Dá praia amanhã? One line the night before, picked by code and explained by Gemma 4 on my laptop

Description and use

dapraia gives users a one‑line recommendation on whether to go to the beach tomorrow, based on user preferences and open weather forecasts.

Problem

Users need a quick, reliable way to decide if tomorrow’s beach conditions will suit their outdoor activities.

Outcome

dapraia delivers a one‑line beach‑day recommendation in the evening, based on user‑defined preferences and real‑time forecast data, enabling users to plan outdoor activities efficiently.

Workflow

• 1. User writes a preference sentence in their own words (e.g., "futevôlei de manhã cedo, pouco vento, sem chuva, e maré baixa pra areia ficar firme.")
• 2. dapraia sends the sentence to Gemma 4 via Ollama to extract rules.
• 3. The tool fetches forecast data from Open‑Meteo and tide data from a model.
• 4. Plain code evaluates each hour against the extracted rules to find suitable windows.
• 5. The headline (go/no‑go, location, time window) is generated by code; Gemma writes the explanatory "why" sentence.
• 6. The user receives the one‑line recommendation in the terminal or as a push notification via ntfy.
• 7. After the outing, the user can report how it went, allowing the plan to adjust next time.

Evidence

dapraia answers that question in one line:

Source

The model reads your sentence and a JSON schema (Ollama's structured outputs) and returns rules: a field, a value, and a quote.

Source

The headline (go or not, where, from when to when) is always written by code. Gemma writes only the second half, the why, and check_why reads it first.

Source