MealOS
Enables users to plan meals and place orders across Swiggy Food, Instamart, and Dineout directly from Claude Desktop conversations.
README
MealOS
Cross-server Swiggy meal intelligence — plan, order, and manage meals across Food, Instamart, and Dineout from a single Claude Desktop conversation.
Verified working: placed a live Swiggy order entirely through Claude Desktop chat.
What it does
MealOS is a local MCP server that gives Claude Desktop a unified meal planning brain. Instead of juggling three separate Swiggy apps, you tell Claude what you want and it:
- Plans 21 meals (breakfast/lunch/dinner × 7 days) routing each to the right Swiggy service
- Cook at home → builds an Instamart grocery cart, skipping what's already in your fridge
- Order in → finds a restaurant on Swiggy Food, builds a delivery cart, places the order
- Dine out → finds a free table reservation on Dineout for Friday/Saturday evenings
- Tracks your weekly budget split across all three modes
- Always asks for confirmation before placing any order or booking
Architecture
Claude Desktop (orchestrator)
│
└── mealos MCP server (local stdio)
├── State tools → profile, fridge, weekly plan (~/.mealos/state.json)
├── Food proxy → 14 tools via https://mcp.swiggy.com/food
├── Instamart proxy → 13 tools via https://mcp.swiggy.com/im
└── Dineout proxy → 8 tools via https://mcp.swiggy.com/dineout
Claude Desktop connects to a single local MCP server. That server proxies all 35 Swiggy tools internally using a stored OAuth token — no remote MCP config needed.
Live Order Flow (what actually happens)
You: "Order chicken biryani for dinner"
Claude → mealos: get_addresses → your Hyderabad address
Claude → mealos: search_restaurants → finds open biryani places near you
Claude → mealos: get_restaurant_menu → browses Chaitanya Food Court
Claude → mealos: search_menu → finds Chicken Dum Biryani, shows sizes
You: "Family size"
Claude → mealos: update_food_cart → adds item (menu_item_id + quantity)
Claude → mealos: get_food_cart → shows total, delivery fee, coupon applied
Claude: "Cart total ₹426. Place order?"
You: "Yes"
Claude → mealos: place_food_order → order placed ✅
→ confirmation SMS on your phone
Total time: ~2 minutes. No app switching.
Note on cart visibility: The MCP API uses a separate device session (
deviceId: im-mcp-server) from the Swiggy app. Cart won't show in the app, but placed orders appear in your order history.
MealOS State Tools
| Tool | Description |
|---|---|
mealos_get_context |
Full planning context: profile + fridge + plan + constraints. Call at session start. |
mealos_get_profile / mealos_set_profile |
Read/write preferences: budget, dietary, cook/dine days, location |
mealos_get_fridge / mealos_update_fridge |
Read/write pantry inventory — Instamart skips items already stocked |
mealos_get_plan / mealos_save_plan |
Read/write the current weekly plan |
mealos_update_meal_status |
Mark a meal as confirmed / ordered / skipped |
Swiggy Tools (proxied)
All 35 tools accessible through the single mealos server:
Food (14): get_addresses, search_restaurants, get_restaurant_menu, search_menu, update_food_cart, get_food_cart, flush_food_cart, fetch_food_coupons, apply_food_coupon, place_food_order, get_food_orders, get_food_order_details, track_food_order, report_error
Instamart (13): search_products, your_go_to_items, update_cart, get_cart, clear_cart, checkout, get_orders, get_order_details, track_order, create_address, delete_address
Dineout (8): get_saved_locations, search_restaurants_dineout, get_restaurant_details, get_available_slots, book_table, get_booking_status
Built-in Prompts
Use these from the Claude Desktop prompt picker (/ in chat):
plan_week— Generate a full weekly meal plan across all three Swiggy servicesquick_order— Place a food delivery order right nowrestock_groceries— Restock the fridge via Instamartbook_dinner— Find and book a free restaurant table for dinner
Setup
Prerequisites
- Claude Desktop
- Node.js 18+
- A Swiggy account with Builders Club API access
1. Clone and install
git clone https://github.com/naga-pavan12/mealOS.git
cd mealOS
npm install
2. Authenticate with Swiggy
npx tsx src/auth-cli.ts
This opens your browser for Swiggy OAuth login. Token saves to ~/.mealos/auth.json. Only needed once — token auto-refreshes.
3. Build and deploy the MCP server
npm run build
mkdir -p ~/.mealos
cp dist/mcp-server.mjs ~/.mealos/mcp-server.mjs
Why
~/.mealos/? Claude Desktop on macOS sandboxes spawned processes and blocks file reads from~/Downloads. The bundle must live outside Downloads.
4. Configure Claude Desktop
Edit ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"mealos": {
"command": "/opt/homebrew/bin/node",
"args": ["/Users/YOUR_USERNAME/.mealos/mcp-server.mjs"]
}
}
}
5. Restart Claude Desktop
mealos should appear as connected under Settings → Developer. You'll see 39 tools available.
6. Set up your profile (first time)
Tell Claude:
"Set up my MealOS profile. Budget ₹3000/week, non-veg, household of 1, cook on Mon/Wed/Sat, dine out Fri/Sat. I'm in [your city]."
Then add your fridge:
"Update my fridge: I have 500g chicken, 1kg rice, 4 eggs, some garlic."
State storage
All state is stored locally — no cloud sync, no external database.
~/.mealos/
├── mcp-server.mjs ← compiled bundle (from dist/)
├── state.json ← profile, fridge inventory, weekly plan
├── auth.json ← Swiggy OAuth token (chmod 0600)
└── registration.json ← OAuth client ID
Swiggy Constraints
Key limits baked into the planning rules:
| Constraint | Limit |
|---|---|
| Cart cap | ₹1000 per order (Food + Instamart) |
| Instamart minimum | ₹99 |
| Payment method | COD only (Food, v1) |
| Dineout bookings | Free only (isFree=true, bookingPrice=0) |
| Confirmation | Required before every order and booking |
| Order retry | Not idempotent — check order history before retrying |
Development
npm run build # bundle src/mcp-server.ts → dist/mcp-server.mjs
npm run typecheck # type-check without emitting
npx tsx src/auth-cli.ts # re-authenticate with Swiggy
After any code change: npm run build → cp dist/mcp-server.mjs ~/.mealos/ → restart Claude Desktop.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.