suggest-dinner

name	suggest-dinner
description	Suggest dinner recipes based on preferences, recent meals, and dietary goals. Use when the user asks "what should I eat", "suggest dinner", "recipe ideas", "what should I cook", or similar.
user-invocable	true
argument-hint	[optional: constraints like 'quick', 'fish', 'no pasta']
allowed-tools	Read, Glob, Agent

Suggest dinner ideas. See docs/food-tracking-process.md for full dietary context.

Arguments: $ARGUMENTS

Data to gather

• Read data/profile/nutrition.md for preferences and restrictions
• Read recent data/food/ entries (last 7-14 days) to avoid repetition and spot patterns
• Optionally check today's vitals/journal for energy level context

Key constraints

• Slightly lactose intolerant, chickpeas cause gas
• Prefers fish over meat, wants to reduce meat frequency
• Mediterranean and Asian cuisines preferred, but open to everything
• Aims for ~200g vegetables per meal
• Fallback meal: wrap + Italian veggies + canned tuna (don't suggest this unless they ask for something very quick)

Suggestions

Give 2-3 options at different effort levels (quick ~15min / medium ~30min / weekend ~45min+). Each should be a real, specific, cookable recipe with enough detail to actually make it — not "some kind of stir fry."

Vary the suggestions: different cuisines, different proteins, different base (rice/noodles/bread/potato). Don't repeat what was eaten in the last few days.

If $ARGUMENTS specifies constraints (like "quick" or "fish"), respect those.

Pattern recognition (as food log grows)

Over time, use the food log to identify:

• Favorites — meals that appear often, good to rotate back to
• Variety gaps — e.g. always pasta, rarely rice/grains; always salmon, never white fish
• Protein balance — are non-meat days getting enough protein?
• Vegetable variety — rotating through different types, not just the same bag of Italian veggies