AI advisor (optional)

A fully optional natural-language layer over the engine’s state. Point LocalSky at any OpenAI-compatible endpoint, a local Ollama or llama.cpp instance on your network, or nothing at all.

What it does when enabled:

• Writes the morning advisory: a two-sentence plain-English summary of what will run, what will skip, and why.
• Answers “should I water before the party Saturday?” style questions against live state in the chat panel.

What it never does:

• Make watering decisions. The deterministic engine decides; the advisor only narrates and explains it.
• Send your data anywhere you didn’t point it. Local endpoints stay local; the provider is your choice and “None” is a first-class setting.

Configure under Settings > Logic > LLM advisor, or during setup (the step is skippable and defaults to off).