Food Delivery

Choose and order food with learned preferences, price comparison, and variety protection.

Safety Notice

This listing is from the official public ClawHub registry. Review SKILL.md and referenced scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "Food Delivery" with this command: npx skills add ivangdavila/food-delivery

When to Use

User wants their agent to handle the entire food ordering process — from deciding what to eat, through comparing options, to placing the actual order. Agent learns preferences over time and makes increasingly better choices.

Architecture

Memory lives in ~/food-delivery/. See memory-template.md for setup.

~/food-delivery/
├── memory.md          # Core preferences, restrictions, defaults
├── restaurants.md     # Restaurant ratings, dishes, notes
├── orders.md          # Recent orders for variety tracking
└── people.md          # Household/group member preferences

User creates these files. Templates in memory-template.md.

Quick Reference

TopicFile
Memory setupmemory-template.md
Decision frameworkdecisions.md
Ordering workflowordering.md
Common trapstraps.md

Data Storage

All data stored in ~/food-delivery/. Create on first use:

mkdir -p ~/food-delivery

Scope

This skill handles:

  • Learning cuisine and taste preferences
  • Storing restaurant ratings and dish notes
  • Comparing prices across delivery platforms
  • Finding active promotions and coupons
  • Placing orders via browser automation
  • Tracking recent orders for variety
  • Managing household member preferences
  • Coordinating group orders

User provides:

  • Delivery app credentials (stored in their browser/app)
  • Delivery address (configured in their apps)
  • Payment methods (configured in their apps)

Self-Modification

This skill NEVER modifies its own SKILL.md. All learned data stored in ~/food-delivery/ files.

Core Rules

1. Learn Preferences Explicitly

User saysStore in memory.md
"I'm vegetarian"restriction: vegetarian
"I love spicy food"preference: spice_level=high
"Allergic to shellfish"CRITICAL: shellfish (always filter)
"I don't like olives"avoid: olives
"Budget around $20"default_budget: $20
"Usually order dinner around 7pm"default_time: 19:00

2. Restriction Hierarchy

CRITICAL (allergies, medical) → ALWAYS filter, never suggest
FIRM (religious, ethical, diet) → filter unless user overrides
PREFERENCE (taste) → consider but flexible

For CRITICAL restrictions:

  • Add note to EVERY order specifying the allergy
  • Verify restaurant can accommodate
  • Never suggest "you could try it anyway"

3. The Decision Flow

When user asks to order food:

Step 1: Context

  • What time is it? (breakfast/lunch/dinner)
  • What day? (weekday functional vs weekend exploratory)
  • Any stated mood or occasion?
  • How many people?

Step 2: Filter

  • Remove anything violating CRITICAL restrictions
  • Remove recently repeated (variety protection)
  • Remove closed restaurants
  • Apply budget constraints

Step 3: Compare

  • Check same restaurant across platforms
  • Find active promos/coupons
  • Calculate total cost (food + delivery + fees)

Step 4: Present

  • Show 2-3 options maximum
  • Include reasoning for each
  • Show price comparison if relevant
  • Recommend one based on user history

Step 5: Confirm & Order

  • Get explicit confirmation
  • Place order via browser
  • Confirm order placed with ETA

4. Variety Protection

Track in orders.md:

  • Last 14 days of orders (restaurant + cuisine type)

Triggers:

  • Same restaurant 3x in 7 days → "You've ordered from [X] a lot. Want to try something similar?"
  • Same cuisine 4x in 7 days → suggest different category
  • Haven't tried category user likes in 2+ weeks → suggest it

5. Price Optimization

Before ordering:

  1. Check restaurant on all user's delivery apps
  2. Compare base prices (often differ by platform)
  3. Check for active coupons/promos
  4. Factor in delivery fees and service charges
  5. Recommend cheapest option for same food

Tell user: "Same order is $4 cheaper on [Platform] today"

6. Group Orders

When ordering for multiple people:

  1. Load ~/food-delivery/people.md for known preferences
  2. Collect any new restrictions
  3. Find intersection cuisine (works for everyone)
  4. Suggest variety restaurants (broad menus)
  5. Calculate fair split if needed

Default crowd-pleasers when no consensus:

  • Pizza (customizable)
  • Burgers (something for everyone)
  • Tacos (variety of fillings)
  • Chinese (range of dishes)
  • Indian (vegetarian options)

7. Context Adaptation

ContextBehavior
"I'm tired"Comfort food, familiar favorites
"Celebrating"Higher-end, special occasion spots
"In a hurry"Fastest delivery, simple orders
"Working lunch"Quick, not messy, productive-friendly
"Date night"Quality over speed, ambiance matters
"Hungover"Greasy comfort, hydrating, gentle
"Post-workout"Protein-heavy, healthier options
Rainy dayWarn about longer delivery times
Friday nightCan wait for quality
Sunday morningBrunch options, recovery mode

8. Proactive Suggestions

When appropriate (not spammy):

  • Notify of flash sales on favorite restaurants
  • Remind of unused loyalty points
  • Suggest reordering past successes
  • Mention new restaurants matching preferences

9. Order Execution

Via browser automation:

  1. Open user's preferred delivery app
  2. Navigate to restaurant
  3. Add items to cart
  4. Apply any coupons found
  5. Verify delivery address
  6. Confirm order total with user
  7. Place order
  8. Report confirmation and ETA

Always confirm before final checkout.

10. Problem Handling

If order has issues:

  • Missing items → help file complaint
  • Wrong items → help request refund
  • Late delivery → track and communicate
  • Quality issues → record in restaurant notes

Boundaries

Stored Locally (in ~/food-delivery/)

  • Cuisine preferences and restrictions
  • Restaurant ratings and dish notes
  • Recent order log (variety tracking)
  • Household member preferences
  • Budget defaults

User Manages (in their apps)

  • Delivery addresses
  • Payment methods
  • Account credentials

Agent Does NOT Store

  • Credit card numbers
  • Exact addresses
  • Account passwords
  • Order receipts with payment details

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

General

Fast Douyin Publish

抖音视频自动发布助手。一键上传视频到抖音,支持自动文案生成和标签优化。

Registry SourceRecently Updated
General

Skills Finder

Intelligent skill matcher that searches multiple skill marketplaces (ClawHub & Skills.sh) in real-time. Supports ANY language for user input, multi-step skil...

Registry SourceRecently Updated
General

Claw Self Improving Plus

Turn raw mistakes, corrections, discoveries, and repeated decisions into structured learnings and promotion candidates. Use when the user wants a conservativ...

Registry SourceRecently Updated