```mermaid
graph TD
%% Define Styles
classDef user fill:#f9f,stroke:#333,stroke-width:2px,color:black,font-weight:bold;
classDef ai fill:#bbf,stroke:#333,stroke-width:2px,color:black;
classDef mcp fill:#ff9,stroke:#333,stroke-width:4px,color:black,font-weight:bold;
classDef cloud fill:#9f9,stroke:#333,stroke-width:2px,color:black;
%% Nodes
User(("👤 User")):::user
AI("🤖 AI Assistant<br>(Claude / Gemini)"):::ai
MCP_Protocol{{"🔌 MCP Protocol"}}:::mcp
Server("⚙️ AWS MCP Server<br>(Local/Docker)"):::mcp
AWS_SDK("📦 AWS SDK v3")
AWS_Cloud("☁️ AWS Cloud<br>(Resources & APIs)"):::cloud
%% Flow - Request
User -->|1. 'Check my AWS costs'| AI
AI -->|2. Identifies Tool<br>'get_recent_cost'| MCP_Protocol
MCP_Protocol -->|3. Sends Tool Call| Server
Server -->|4. Executes Command| AWS_SDK
AWS_SDK -->|5. API Request| AWS_Cloud
%% Flow - Response
AWS_Cloud -.->|6. Returns Data JSON| AWS_SDK
AWS_SDK -.->|7. Returns Object| Server
Server -.->|8. Formats Result| MCP_Protocol
MCP_Protocol -.->|9. Sends Tool Result| AI
AI -.->|10. Summarizes Answer| User
%% Grouping
subgraph Local_Computer ["Your Machine / Local Environment"]
AI
MCP_Protocol
Server
AWS_SDK
end
subgraph AWS ["External"]
AWS_Cloud
end
```
### 📝 Step-by-Step Explanation
1. **User Request**: You ask a natural language question (e.g., *"What is my cloud spend?*").
2. **Tool Selection**: The AI analyzes your intent and decides it needs the `get_recent_cost` tool.
3. **MCP Protocol**: The AI sends a request via the Model Context Protocol to your local server.
4. **Execution**: The **AWS MCP Server** receives the request and translates it into a specific AWS SDK function call.
5. **API Call**: The command is sent securely to the AWS Cloud using your local credentials.
6. **Data Retrieval**: AWS returns the raw data (JSON).
7. **Response Loop**: The data flows back to the AI, which interprets the JSON and generates a human-readable summary for you.
