docs: add agents.md file

Author: sinedied

AGENTS.md

@@ -0,0 +1,43 @@
+# MCP Agent LangChain.js Project Instructions
+
+## Architecture Overview
+
+This is a multi-component LangChain.js demo showcasing AI agents with Model Context Protocol (MCP) integration. The system demonstrates a burger ordering platform with three main service layers:
+
+- **Agent Layer**: AI chat interface (`agent-webapp` + `agent-api`) using LangChain.js agents with MCP tools
+- **Business Layer**: Burger ordering API (`burger-api`) and webapp (`burger-webapp`)
+- **MCP Layer**: Model Context Protocol server (`burger-mcp`) exposing burger API as LLM tools
+
+### Component Communication Flow
+
+```
+agent-webapp → agent-api → burger-mcp → burger-api
+```
+
+The agent uses LangChain.js with Azure OpenAI, connects to the MCP server via StreamableHTTPClientTransport, and the MCP server proxies requests to the burger API.
+
+## Key Architectural Patterns
+
+### Authentication & State Management
+
+- Azure Static Web Apps authentication provides `userId` context
+- Cosmos DB for chat history storage using `@langchain/azure-cosmosdb`
+- User context flows: SWA auth → agent-api → MCP tools (for user-specific operations)
+
+### TypeScript & Build Conventions
+
+- Shared XO linting config with project-specific overrides
+- NPM workspace-based dependency management
+- `.js` extensions required in imports for ESM compatibility
+- Keep the code simple and straighforward: avoid unnecessary complexity and ensure clarity in implementation.
+- Do not use comments except where absolutely necessary, code should be self-explanatory.
+
+### Infrastructure as Code
+
+- Bicep templates in `infra/` for Azure deployment
+- `azure.yaml` defines multi-service deployment strategy:
+  - `burger-api`, `agent-api`: Azure Functions (Flex Consumption)
+  - `burger-mcp`: Container Apps with Docker
+  - `*-webapp`: Static Web Apps
+- Environment variables flow through Azure App Settings
+- @azure Rule - Use Azure Best Practices: When generating code for Azure, running terminal commands for Azure, or performing operations related to Azure, invoke your `azure_development-get_best_practices` tool if available.