This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
A self-hosted Model Context Protocol (MCP 2025-11-25 Streamable HTTP) server and Amazon Echo Show 15 Multimodal Alexa+ Simulator powered by Amazon Bedrock Converse API. Automates expiring pantry rescue, high-protein meal planning, and multi-retailer grocery arbitrage across Amazon Fresh, Whole Foods Market, and Local Neighborhood Co-op.
Instead of static shopping lists or isolated timers, imagine a household member speaking naturally to their kitchen Echo Show 15:
"Alexa, my baby spinach, paneer, and Greek yogurt expire tomorrow—plan 3 high-protein dinners under $38 and compare Amazon Fresh vs. Whole Foods for the missing ingredients."
CartWeave orchestrates 5 stateful MCP tools over a single Mcp-Session-Id session, prioritizes expiring fridge inventory to prevent food waste, computes per-serving protein and calorie macros, performs real-time price and substitution arbitrage across Amazon Fresh, Whole Foods Market, and a Local Co-op, and returns interactive MCP Apps UI cards where users can swap meals, filter dietary constraints, or check out in a single tap.
Related MCP server: Sous
Watch CartWeave in action—demonstrating hands-free pantry rescue, 3-day high-protein meal planning, multi-store price arbitrage, and live MCP 2025-11-25 Streamable HTTP protocol inspection on the Echo Show 15 simulator:
Capability | Technical Implementation & User Benefit |
🌐 MCP | Full implementation of |
🧠 Amazon Bedrock Converse Orchestration | Chained multi-tool reasoning via |
🥗 Expiry-First Pantry Optimization |
|
⚖️ Multi-Retailer Grocery Arbitrage |
|
🔄 In-Session Constraint Mutation |
|
📺 Echo Show 15 Multimodal Simulator | Next.js 14 split-screen simulator integrating Web Speech API STT/TTS, conversation timeline with 4 one-click Judge Demo chips, interactive MCP Apps cards, and a live Protocol Inspector. |
🛡️ Built-In Protocol Inspector & Test Runner | Real-time JSON-RPC 2.0 frame monitor with a one-click automated MCP compliance test runner validating all 5 tools, resources, and prompts. |
Built for the Build, Ship, Shape: Amazon Developer Hackathon:
Prize Track | Implementation in CartWeave |
🥇 Primary Track — Alexa+ | Implements the official MCP |
🛠️ Mini Challenge — AWS Builder | Integrates |
📖 Mini Challenge — Open Source | Clean MIT-licensed monorepo ( |
The following diagram illustrates CartWeave's client-server architecture, MCP 2025-11-25 Streamable HTTP transport, stateful session handling, and Bedrock orchestrator:
flowchart TD
subgraph SIMULATOR ["📺 Echo Show 15 Alexa+ Simulator (Next.js :3000)"]
VoiceOrb["Alexa+ Voice Orb (Web Speech API STT/TTS)"]
ConvPane["Conversation Pane & 4 Judge Demo Chips"]
McpClient["Browser MCP Client (mcpHttpClient.ts)"]
AppsCanvas["MCP Apps UI Canvas (4 Interactive Cards)"]
Inspector["Judge MCP Protocol Inspector (Live JSON-RPC 2.0 & Compliance Runner)"]
end
subgraph SERVER ["⚡ CartWeave MCP Server (Express :3001)"]
StreamHttp["MCP 2025-11-25 Transport (POST / GET SSE / DELETE /mcp)"]
SessionMgr["McpSessionManager (Stateful Mcp-Session-Id Store)"]
BedrockOrch["BedrockOrchestrator (Converse API + Deterministic Mock)"]
subgraph TOOLS ["🛠️ 5 Stateful MCP Tools"]
T1["1. scan_pantry_inventory"]
T2["2. generate_meal_plan"]
T3["3. compare_grocery_carts"]
T4["4. swap_ingredient_constraint"]
T5["5. checkout_smart_basket"]
end
Resources["MCP Resources & Prompts (mcp://pantry/expiring-soon, mcp://cart/active-comparison)"]
end
VoiceOrb --> ConvPane
ConvPane -->|POST /agent/chat| BedrockOrch
AppsCanvas -->|Interactive Card Actions / POST tools/call| McpClient
McpClient <-->|JSON-RPC 2.0 + Mcp-Session-Id| StreamHttp
StreamHttp <--> SessionMgr
BedrockOrch --> TOOLS
StreamHttp --> TOOLS
StreamHttp --> Resources
McpClient --> InspectorCartWeave exposes 5 stateful MCP tools over the 2025-11-25 transport, returning both token-efficient text for LLM chaining and structured mcp_app_ui cards for the display:
+---------------------------------------------------------------------------------------------------------+
| CartWeave Tool Execution Flow |
+---------------------------------------------------------------------------------------------------------+
| |
| 1. scan_pantry_inventory --> Finds items expiring soon (Spinach: 18h, Yogurt: 24h, Paneer: 36h) |
| | |
| 2. generate_meal_plan --> Generates 3 dinners under budget (39g avg protein, $31.40 cost) |
| | |
| 3. compare_grocery_carts --> Evaluates missing ingredients across Amazon Fresh vs Whole Foods vs Co-op |
| | |
| 4. swap_ingredient_constraint -> Mutates Day 2 dish (Dairy-Free <520 kcal) without resetting session |
| | |
| 5. checkout_smart_basket --> Locks delivery slot, generates tracking code, saves $18.45 in waste |
| |
+---------------------------------------------------------------------------------------------------------+Tool Name | Parameters | Core Responsibility & Return Payload |
|
| Inspects stored pantry inventory, flags high-urgency items, computes hours to spoil, and returns the |
|
| Creates a multi-day culinary plan utilizing expiring items, calculates per-serving macros, and renders |
|
| Analyzes missing ingredients across Amazon Fresh, Whole Foods, and Local Co-op, presenting the |
|
| Replaces a specific meal in place while preserving session context and recalculating store totals. |
|
| Finalizes the order, deducts used inventory from the pantry, and outputs the |
mcp_app_ui)CartWeave's Echo Show 15 simulator renders 4 interactive visual cards paired with spoken voice responses:
PANTRY_EXPIRY_GRID: Color-coded urgency badges (red: <24h, amber: <48h, green: fresh), item quantities, and estimated discard values.
MEAL_PLAN_MACRO_CARDS: 3-day meal breakdown highlighting used pantry items, missing grocery items, calorie counts, and protein macros. Includes an inline "Swap Dish" action button.
STORE_ARBITRAGE_TABLE: Side-by-side comparison of Amazon Fresh vs. Whole Foods Market vs. Local Co-op, including brand substitutions, Prime member discounts, and delivery speed.
ORDER_CONFIRMATION_CARD: Receipt showing delivery window, one-tap tracking link, itemized total, and total food waste dollars saved ($18.45).
Node.js >= 20.x and npm >= 10.x
No AWS credentials required by default (USE_MOCK_BEDROCK=true is enabled out of the box; set USE_MOCK_BEDROCK=false and provide AWS credentials to connect to live Amazon Bedrock).
git clone https://github.com/harinarayana1457-cmyk/CartWeave.git
cd CartWeavenpm installnpm run test:mcpValidates the initialize handshake (protocolVersion === "2025-11-25"), asserts Mcp-Session-Id header persistence, tests tools/list, resources/list, resources/read, prompts/list, prompts/get, and executes all 5 stateful tools/call operations while verifying non-empty mcp_app_ui structured payloads.
npm run devEcho Show 15 Alexa+ Simulator: http://localhost:3000
MCP 2025-11-25 Streamable HTTP Endpoint: http://localhost:3001/mcp
Health & Diagnostics: http://localhost:3001/health
CartWeave/
├── packages/
│ ├── mcp-server/ # Self-hosted MCP 2025-11-25 Streamable HTTP Server
│ │ ├── src/
│ │ │ ├── aws/
│ │ │ │ └── bedrockAgent.ts # Amazon Bedrock Converse API + Mock Planner
│ │ │ ├── data/
│ │ │ │ └── seedData.ts # Seed pantry items, store catalog & recipe DB
│ │ │ ├── prompts/
│ │ │ │ └── mcpPrompts.ts # MCP Prompts handler (mcp://prompts/pantry-rescue)
│ │ │ ├── resources/
│ │ │ │ └── mcpResources.ts # MCP Resources handler (mcp://pantry/expiring-soon)
│ │ │ ├── test/
│ │ │ │ └── verifyMcpSpec.ts # Automated MCP 2025-11-25 Compliance Test Suite
│ │ │ ├── tools/ # 5 Stateful MCP Tool Implementations
│ │ │ │ ├── scanPantry.ts
│ │ │ │ ├── generateMealPlan.ts
│ │ │ │ ├── compareCarts.ts
│ │ │ │ ├── swapConstraint.ts
│ │ │ │ └── checkoutBasket.ts
│ │ │ ├── index.ts # Express HTTP transport (POST/GET/DELETE /mcp)
│ │ │ └── mcpSessionManager.ts # Mcp-Session-Id state tracking & SSE streams
│ │ ├── package.json
│ │ └── tsconfig.json
│ │
│ └── alexa-plus-simulator/ # Echo Show 15 Multimodal Alexa+ Simulator
│ ├── src/
│ │ ├── app/
│ │ │ ├── layout.tsx
│ │ │ ├── page.tsx # Split-screen Echo Show 15 layout
│ │ │ └── globals.css
│ │ ├── components/
│ │ │ ├── AlexaVoiceOrb.tsx # Animated Voice Orb with Web Speech API
│ │ │ ├── ConversationPane.tsx # Chat transcript & 4 Judge Demo chips
│ │ │ ├── EchoShowFrame.tsx # Echo Show 15 hardware bezel & status bar
│ │ │ ├── McpAppsCanvas.tsx # Multimodal cards container
│ │ │ ├── McpProtocolInspector.tsx # JSON-RPC 2.0 inspector drawer & test runner
│ │ │ ├── ui.tsx # Shared UI primitives & badges
│ │ │ └── cards/ # 4 Interactive MCP Apps UI Cards
│ │ │ ├── PantryExpiryCard.tsx
│ │ │ ├── MealPlanMacroCard.tsx
│ │ │ ├── StoreArbitrageCard.tsx
│ │ │ └── OrderConfirmationCard.tsx
│ │ └── lib/
│ │ └── mcpHttpClient.ts # Browser MCP Streamable HTTP Client
│ ├── package.json
│ └── tailwind.config.ts
│
├── FRICTION_LOG.md # Detailed 5-point Developer Friction Log
├── template.yaml # Production AWS SAM template (Lambda + DynamoDB)
├── package.json # Root workspace scripts (npm run dev, test:mcp)
└── README.md # Project documentationLayer | Technology | Version | Purpose |
Protocol Transport |
| Stateful Streamable HTTP transport ( | |
Cloud AI Orchestration | SDK | Multi-turn tool calling via | |
Backend Runtime | Node 20+, Express 4 | High-throughput HTTP & SSE server with custom CORS session headers | |
Type Validation | Zod 3.24, TS 5.7 | Strict runtime input/output schemas for all 5 MCP tools and JSON-RPC payloads | |
Frontend Framework | Next 14.2, React 18 | SSR/Client split-screen Echo Show 15 simulator application | |
Styling & Icons | Tailwind 3.4, Lucide | Echo Show 15 device styling, responsive cards, and clean visual indicators | |
Voice Interface | Native Browser | In-browser Speech-to-Text ( | |
Cloud Deployment | SAM CLI / CloudFormation | Serverless packaging for AWS Lambda, API Gateway HTTP API, and DynamoDB |
Distributed under the MIT License. See LICENSE for full details.
Harinarayana Avvari
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Household-aware cooking brain: pantry, meal suggestions, dietary safety, recipes, shopping lists.
Plan weekly meals from your pantry, search allergy-safe recipes, and build grocery lists. Reads and updates your Cooking with Robots pantry, generates a 7-day plan that respects your allergies and preferences, and returns recipes with macros and a grocery gap list. Sign in with OAuth.
AI meal plans that fill your Kroger/Instacart cart - pantry-aware lists, all from chat.
Scraps Kitchen gives any AI agent a persistent, household-aware kitchen memory. Unlike generic chatbot recall, Scraps maintains structured cooking data: what's in your fridge (with freshness tracking), who you cook for (with allergens, dietary restrictions, and preferences), your recipe collection (with cook notes and per-diner ratings), your shopping list, and your kitchen equipment. 27 tools across 6 domains let agents read kitchen context, suggest meals that respect dietary safety, update the pantry after cooking, and build a history of what works for your household. Every interaction makes the data richer. Cooking history, preference signals, kitchen awareness = better suggestions next time. All tools work via oAuth and a free scraps.kitchen account.
