Help us improve
Share bugs, ideas, or general feedback.
From lunch-decision
Adaptive lunch decision assistant - helps you decide what to eat through intelligent questioning
npx claudepluginhub h315uk3/beig-pocHow this command is triggered — by the user, by Claude, or both
Slash command
/lunch-decision:what-should-i-eatThis command is limited to the following tools:
The summary Claude sees in its command listing — used to decide when to auto-load this command
# What Should I Eat? **Adaptive lunch decision support using Bayesian belief updating and information theory.** When you're unsure what to eat for lunch, this command helps you decide through targeted questioning. Each question maximizes expected information gain about your mood, preferences, and constraints, adapting to your answers in real-time to suggest the perfect lunch option. --- ## How It Works This assistant explores 5 dimensions of your lunch decision: - **Mood & Context** - How you're feeling and your current situation - **Taste Preferences** - What flavors and cuisines appe...
/oda-meal-planPlans weekly meals using Oda recipes based on servings, dietary preferences, and memories; suggests diverse options, adds to cart, saves to MEAL_PLAN.md, and reviews cart.
/meal-planGenerates individualized meal plans with macronutrient targets, condition-specific dietary guidelines, sample menus, and food swap suggestions for dietitian review.
/interviewActs as a thought partner, asking clarifying questions before advising on your question or situation.
/ask-me-questionsGathers task requirements through adaptive questioning on context gaps (what, who, why, how, when), with decision gates to loop until ready to execute.
/good-questionAsks adaptive, information-maximizing questions to systematically elicit and clarify software requirements until uncertainty is resolved.
/travelGenerates comprehensive travel plans with real-time weather forecasts, budget breakdowns, day-by-day itineraries, packing lists, and local tips for destinations, supporting multi-city trips, custom days, and budgets.
Share bugs, ideas, or general feedback.
Adaptive lunch decision support using Bayesian belief updating and information theory.
When you're unsure what to eat for lunch, this command helps you decide through targeted questioning. Each question maximizes expected information gain about your mood, preferences, and constraints, adapting to your answers in real-time to suggest the perfect lunch option.
This assistant explores 5 dimensions of your lunch decision:
Through 5-10 targeted questions, it narrows down what you truly want to eat right now.
When executing this command, maintain a natural, conversational tone.
Focus on Food Preferences, Not Mechanisms
Hide Technical Terminology
Present Natural Conversation
Let's figure out what you should eat for lunch! I'll ask a few questions to understand what you're in the mood for today.
Question 1: How are you feeling about lunch today?
• Want something comforting and familiar
• Feeling adventurous, want to try something new
• Need something quick and practical
• Looking for a nice social experience
Initializing session...
Dimension: mood (entropy: 2.0)
Evaluating question quality...
How are you feeling about lunch today?
export PYTHONPATH="${CLAUDE_PLUGIN_ROOT}"
python3 -m lunch_decision.cli.session init
Store the session_id from output (internal use only).
User message: "Let's figure out what you should eat for lunch! I'll ask a few questions to understand what you're in the mood for today."
Repeat until convergence (typically 5-10 questions):
export PYTHONPATH="${CLAUDE_PLUGIN_ROOT}"
python3 -m lunch_decision.cli.session next-question --session-id <SESSION_ID>
If "converged": true, skip to step 3.
Get session status:
export PYTHONPATH="${CLAUDE_PLUGIN_ROOT}"
python3 -m lunch_decision.cli.session status --session-id <SESSION_ID>
Generate a natural question based on:
Ask user with AskUserQuestion tool:
Based on the user's answer, estimate likelihood P(answer | hypothesis) for each hypothesis in the dimension.
export PYTHONPATH="${CLAUDE_PLUGIN_ROOT}"
python3 -m lunch_decision.cli.session update-with-computation \
--session-id <SESSION_ID> \
--dimension <DIMENSION> \
--question <QUESTION> \
--answer <ANSWER> \
--likelihoods '{"hyp1": 0.x, "hyp2": 0.y, ...}'
Return to step 2.1.
export PYTHONPATH="${CLAUDE_PLUGIN_ROOT}"
python3 -m lunch_decision.cli.session complete --session-id <SESSION_ID>
User message: "Great! Based on your answers, let me suggest some lunch options..."
.claude/lunch_decision/sessions/<SESSION_ID>.json
Analyze the session data:
Provide 2-3 specific, actionable lunch recommendations with reasoning:
Example:
Based on your answers, here are my top suggestions:
1. **Ramen at [Local Shop]** (Best Match)
- Matches your craving for Asian food
- Warm and comforting on a cold day
- Quick service, fits your time constraint
- Budget-friendly at ~$12
2. **Poke Bowl at [Nearby Café]**
- Fresh and healthy option
- Asian flavors you wanted
- Customizable to preferences
- About $15
3. **Thai Curry Takeout**
- Rich, satisfying flavors
- Can order ahead for speed
- Vegetarian options available
- $12-15 range
I'd go with the ramen - it hits all your key criteria and the cozy atmosphere matches your mood today!
Adjust in config/dimensions.json: