Food Finder

Food Finder Skill

You help the user find the best food delivery options by searching Swiggy and cross-referencing Google Maps ratings.

Prerequisites

• Authenticated Browser MCP must be available (Chrome with CDP on localhost:9222)
• Browser is logged into Google (for Maps)
• Swiggy works without login for browsing

Workflow

Step 1: Parse the Request

Extract from the user's query:

• Dish/cuisine (e.g., "momos", "pizza", "biryani", "Chinese")
• Location (ask the user if not provided)
• Party size (default: 2)
• Budget (optional)
• Veg/non-veg (default: no preference)

Step 2: Read Preferences

Read ~/.claude/skills/food-finder/references/preferences.md to check:

• Known favorite restaurants
• Rating thresholds
• Cuisine preferences
• Usual budget range

Step 3: Search Swiggy

1. Navigate to https://www.swiggy.com
2. Set delivery location:

• Click the location/address field
• Type the location
• Select the first autocomplete suggestion
• Wait for page to reload with local restaurants

1. Search for the dish/cuisine:

• Click the search bar
• Type the dish/cuisine name
• Press Enter or click search
• Wait for results to load

1. Switch to "Restaurants" tab if available (not "Dishes")
2. Collect top 8-10 results:

• Restaurant name
• Swiggy rating (out of 5)
• Delivery time (minutes)
• Price for two (₹)
• Cuisine tags
• Any offers/discounts visible

Tips for Swiggy scraping:

• Use browser_snapshot to read the page content — it's more reliable than screenshots
• Swiggy loads restaurants dynamically; scroll down if needed using browser_evaluate with window.scrollBy(0, 800)
• Look for rating in the snapshot text (usually like "4.3" near the restaurant name)
• Price for two is usually shown as "₹300 for two" or similar

Step 4: Cross-reference Google Maps

For the top 5 restaurants (by Swiggy rating):

1. Open a new tab: browser_tabs with action "new"
2. Navigate to https://www.google.com/maps/search/
3. Take a snapshot to find:

• Google Maps rating (out of 5)
• Number of reviews
• Any notable review highlights

1. Go back to the Swiggy tab

Step 5: Rank & Recommend

Create a comparison table sorted by combined score:

• Combined score = (Swiggy rating × 0.4) + (Google Maps rating × 0.4) + (review volume score × 0.2)
• Review volume score: 4.5+ if >1000 reviews, 4.0 if >500, 3.5 if >200, 3.0 otherwise

Present the table:

| # | Restaurant | Cuisine | Swiggy ⭐ | Google ⭐ (reviews) | Delivery | Price/2 | Score |
|---|-----------|---------|----------|-------------------|----------|---------|-------|
| 1 | Name      | Type    | 4.5      | 4.4 (1.2k)        | 30 min   | ₹400    | 4.48  |

Highlight the top pick with a brief reason (e.g., "Best overall rating + fast delivery").

Apply preference adjustments:

• Boost restaurants matching known favorites
• Boost restaurants above user's rating threshold
• Flag if below user's usual budget or rating preferences

Step 6: Offer Next Steps

Ask if the user wants to:

1. Open the menu on Swiggy (navigate to the restaurant page)
2. See more options (scroll for more results)
3. Try a different cuisine/location

Step 7: Update Preferences

After the interaction, run the preferences update:

python3 ~/.claude/skills/food-finder/scripts/update_prefs.py \
  --cuisine "" \
  --location "" \
  --chosen "" \
  --party-size  \
  --budget ""

This appends/updates entries in references/preferences.md.

Error Handling

• Swiggy not loading: Try refreshing. If still broken, fall back to Google Maps search only.
• Location not setting: Try typing a more specific address.
• No results: Broaden the search (e.g., "Chinese" instead of "Szechuan dumplings").
• Google Maps rate limit: Skip cross-referencing, present Swiggy results only with a note.

Output Format

Keep the final recommendation concise — a table + 1-2 line recommendation. Example:

Top picks for momos near Koramangala:

| # | Restaurant      | Swiggy | Google (reviews) | Time  | Rs/2 |
|---|----------------|--------|-----------------|-------|------|
| 1 | Khawa Karpo    | 4.5    | 4.4 (890)       | 25min | 350  |
| 2 | Momo I Am      | 4.3    | 4.2 (1.2k)      | 35min | 300  |
| 3 | WowMomos       | 4.1    | 3.9 (2.5k)      | 20min | 250  |

Top pick: Khawa Karpo — highest combined rating, reasonable delivery time.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: floomhq
• Source: floomhq/moto
• License: MIT
• Homepage: https://github.com/floomhq/moto#readme

Install and usage instructions live in the source repository linked above.