AI-powered food optimizer — set a budget, pick one search query (like Swiggy search_menu), diet filter, and demo-only ranking.
* This is a sample demo project only — no real-life data is used.
Amount
Live Food MCP has no “margin” or extra knobs — you pick an address, then search_restaurants / search_menu with a query. This demo scores static menus until OAuth is wired.
Diet
Rank results (demo scorer)
Removed for MCP honesty: spice sliders, allergen tags, ETA limits, combo/surprise modes — those are either menu-specific after get_restaurant_menu or not exposed as Food MCP inputs.
Analysis Engine Ready
Configure budget, one food query, diet, and rank. Hit EXECUTE ANALYSIS for scored picks from the demo catalog (live MCP adds address → tool calls).
Swiggy Builders Club
Built on Swiggy Food MCP
BudgetFood AI demonstrates how an agent can reason over budget, taste, and offers — then hand off to the canonical Food journey (address → search → menu → cart → coupons → place → track) using Swiggy's MCP servers. This demo uses curated sample data; production wiring uses OAuth and live tools at mcp.swiggy.com/food.
14 Food tools
Discover, cart, order, track — one endpoint.
OAuth 2.1 + PKCE
Same flow as Cursor & Claude Desktop.
Official docs
Recipes, errors, and production checklist.
How it works
Demo vs live Swiggy MCP
- • Static menus & coupons — no calls to Swiggy APIs from your browser.
- • Scoring engine ranks items by budget, taste, ETA, offers — optimizes decisions before ordering.
- • Checkout flow simulates address & payment for UX testing.
- • Agent calls
search_restaurants,update_food_cart,place_food_order, etc. - • Authenticated user session; COD v1; ₹1000 cart cap per Builders guidelines.
- • Replace
lib/providers/swiggy.tsbridge with your MCP client + OAuth tokens.
FAQ
Hackathon & MCP
Is this real Swiggy data?
No — items and restaurants are sample data for the demo. Live Swiggy MCP returns real menus, carts, and orders after OAuth.
How do I connect Cursor to Swiggy MCP?
Add the Food server URL to your MCP config and reload servers — see the Integration guide for the exact JSON and OAuth steps.
What should judges look for?
BudgetFood AI shows product thinking (budget + cravings + scoring) and clear mapping to Swiggy's documented Food tool sequence.