Fixed broken dependencies and added option to use uv instead of pip and venv

.devcontainer/devcontainer.json

@@ -0,0 +1,42 @@
+// For format details, see https://aka.ms/devcontainer.json. For config options, see the
+// README at: https://github.com/devcontainers/templates/tree/main/src/python
+{
+	"name": "Python 3",
+	// Or use a Dockerfile or Docker Compose file. More info: https://containers.dev/guide/dockerfile
+	"image": "mcr.microsoft.com/devcontainers/python:1-3.12",
+
+	// Features to add to the dev container. More info: https://containers.dev/features.
+	"features": {
+    "ghcr.io/devcontainers/features/azure-cli:1": {},
+	"ghcr.io/devcontainers-extra/features/uv:1": {}
+  },
+
+	// Configure tool-specific properties.
+	"customizations": {
+		// Configure properties specific to VS Code.
+		"vscode": {
+			"settings": {},
+			"extensions": [
+				"streetsidesoftware.code-spell-checker"
+			]
+		}
+	},
+
+	// Use 'forwardPorts' to make a list of ports inside the container available locally.
+	// "forwardPorts": [9000],
+
+	// Use 'portsAttributes' to set default properties for specific forwarded ports. 
+	// More info: https://containers.dev/implementors/json_reference/#port-attributes
+	"portsAttributes": {
+		"9000": {
+			"label": "Hello Remote World",
+			"onAutoForward": "notify"
+		}
+	},
+
+	// Use 'postCreateCommand' to run commands after the container is created.
+	"postCreateCommand": "pip3 install -r requirements.txt"
+
+	// Uncomment to connect as root instead. More info: https://aka.ms/dev-containers-non-root.
+	// "remoteUser": "root"
+}

SETUP_UV.md

@@ -0,0 +1,216 @@
+![alt text](image-1.png)
+# Microsoft AI Agentic Workshop Setup
+
+This document describes how to setup and run your AI Agents for the workshop if you are using UV as your python manager.
+
+## Setup & Installation
+
+### 1. Clone the Repository
+
+Open VS Code terminal 
+
+```bash 
+git clone <repo_url> # from folder where you want clone to reside
+```
+### 2. Install Python dependencies  
+
+With uv, packages are managed through pyproject.toml in the subfolders. Instead of activating a virtual environment, you can use `uv run <python_filename.py>
+
+### 3. Deploy LLM model using Azure AI Foundry
+
+1. Login to ai.azure.com. Create account if you don't already have access to an account.
+2. Create project, use new hub is none exists. This will setup a hub, project container, AI services, Storage account and Key Vault
+3. Use API Key, Azure OpenAI Service endpoint and Project connection string and add to .env file (next step)
+4. On project page, go to Models + endpoints -> Deploy model -> Deploy base model -> gpt-4.1
+5. Select deployment type (Standard, Global Standard etc.) and region if desired
+6. Customize deployment details to reduce tokens per minute to 10K, disable dynamic quote 
+
+### 4. Set up your environment variables and select the agent to run 
+
+Rename `.env.sample` to `.env` and fill in all required fields:  
+
+```bash  
+############################################  
+#  Azure OpenAI – chat model configuration #  
+############################################  
+# Replace with your model-deployment endpoint in Azure AI Foundry  
+AZURE_OPENAI_ENDPOINT="https://YOUR-OPENAI-SERVICE-ENDPOINT.openai.azure.com"  
+
+# Replace with your Foundry project’s API key  
+AZURE_OPENAI_API_KEY="YOUR-OPENAI-API-KEY"  
+
+# Connection-string that identifies your Foundry project / workspace. Only needed if you're using Azure Agent Service
+AZURE_AI_AGENT_PROJECT_CONNECTION_STRING="YOUR-OPENAI-PROJECT-CONNECTION-STRING"  
+
+# Model deployment & API version  
+AZURE_OPENAI_CHAT_DEPLOYMENT="gpt-4.1"  
+AZURE_AI_AGENT_MODEL_DEPLOYMENT_NAME="gpt-4.1"  
+AZURE_OPENAI_API_VERSION="2025-01-01-preview"  
+OPENAI_MODEL_NAME="gpt-4.1-2025-04-14"  #only applicable for Autogen
+
+############################################  
+#     Local URLs for backend & MCP server  #  
+############################################  
+BACKEND_URL="http://localhost:7000"  
+MCP_SERVER_URI="http://localhost:8000/mcp"  
+
+############################################  
+
+############################################  
+#         Agent module to be executed      #  
+############################################  
+# AGENT_MODULE="agents.autogen.multi_agent.reflection_agent"
+# AGENT_MODULE="agents.autogen.single_agent.loop_agent"
+# AGENT_MODULE="agents.autogen.multi_agent.collaborative_multi_agent_round_robin"
+# AGENT_MODULE="agents.autogen.multi_agent.collaborative_multi_agent_selector_group"
+# AGENT_MODULE="agents.autogen.multi_agent.handoff_multi_agent_domain"
+# AGENT_MODULE="agents.semantic_kernel.multi_agent.collaborative_multi_agent"
+# AGENT_MODULE="agents.semantic_kernel.multi_agent.a2a.collaborative_multi_agent"
+AGENT_MODULE="agents.autogen.single_agent.loop_agent"  
+
+# -----------------------------------------------------------  
+# If you are experimenting with Logistics-A2A, uncomment:  
+# LOGISTIC_MCP_SERVER_URI="http://localhost:8100/sse"  
+# LOGISTICS_A2A_URL="http://localhost:9100"  
+# -----------------------------------------------------------  
+
+
+#############################################################  
+#          Cosmos DB – state persistence settings           #  
+#############################################################  
+# Endpoint for your Cosmos DB account (SQL API)  
+COSMOSDB_ENDPOINT="https://YOUR-COSMOS-ACCOUNT.documents.azure.com:443/"  
+
+# ---------  Choose ONE authentication method  --------------  
+# (1) Account key  
+#COSMOSDB_KEY="YOUR-COSMOS-ACCOUNT-KEY"  
+
+# (2) Azure AD service-principal (preferred in production)  
+#AAD_CLIENT_ID="00000000-0000-0000-0000-000000000000"  
+#AAD_CLIENT_SECRET="YOUR-AAD-CLIENT-SECRET"  
+#AAD_TENANT_ID="11111111-1111-1111-1111-111111111111"  
+# -----------------------------------------------------------  
+
+# Logical (application) tenant for data isolation  
+# Leave as "default" unless you partition data by customer / org  
+DATA_TENANT_ID="default"  
+
+# Database & container names (created automatically if not present)  
+COSMOSDB_DB_NAME="ai_state_db"  
+COSMOSDB_CONTAINER_NAME="state_store"  
+```
+
+**Note:**    
+#### Choosing a State Store  
+
+- **Do nothing** ➜ the workshop uses an in-memory Python `dict` (fast, but data is lost when the process exits).  
+- **Fill in the Cosmos variables** ➜ the app automatically switches to an Azure Cosmos DB container with a hierarchical partition-key (`/tenant_id + /id`) so chat history survives restarts and scales across instances.  
+
+> If neither `COSMOSDB_KEY` nor the AAD credential set is provided, the code silently falls back to the in-memory store.  
+> **Important:**    
+> If you choose Cosmos DB and use Azure AD service-principal authentication, you must grant the service principal a custom role for data plane (read/write) access in Cosmos DB.    
+> See: [Grant data plane access using custom roles in Azure Cosmos DB](https://learn.microsoft.com/en-us/azure/cosmos-db/nosql/how-to-grant-data-plane-access?tabs=custom-definition%2Ccsharp&pivots=azure-interface-cli)  
+>  
+> Without this role, the application will not be able to access or persist chat history in Cosmos DB using Azure AD authentication.  
+
+#### Make sure your Azure resources are configured to use the correct model deployment names, endpoints, and API versions.  
+
+---  
+
+### 5. Run MCP Server 
+
+Navigate to ```agentic_ai/backend_services``` folder, and in terminal window with virtual environment activated, run MCP server
+
+```bash
+uv run mcp_service.py  
+# Keep this terminal open; open another terminal for the next step.  
+
+```
+
+### 6. Run application  
+Navigate to ```agentic_ai/applications```
+
+The common backend application runs the agent selected in the .env file and connects to the frontend UI. 
+
+### Option 1: Run Both Backend and Frontend Together  
+
+```bash  
+bash run_application.sh  
+```
+This script will start the FastAPI backend (`backend.py`) and the Streamlit frontend (`frontend.py`) simultaneously.  
+
+- The backend will listen on [http://localhost:7000](http://localhost:7000).  
+- The Streamlit user interface will open (usually at [http://localhost:8501](http://localhost:8501)).
+
+### Option 2: Run Backend and Frontend Separately
+
+### 1. Start the FastAPI Backend  
+
+```bash  
+uv run backend.py  
+# Keep this terminal open; open another terminal for the frontend.  
+```
+The backend will be available at `http://localhost:7000/chat`.  
+
+### 2. Start the Streamlit Frontend  
+
+```bash  
+uv run streamlit run frontend.py  
+```
+Navigate to the address Streamlit provides (typically http://localhost:8501) to use the chat interface.  
+Streamlit should popup a chat window for the Agent in a new Edge tab. 
+
+If you successfully completed all the steps, setup is complete and your agent should be running now !
+
+## How It Works  
+
+1. **Web UI (Streamlit):**    
+   Users input messages and interact with the assistant. A unique session ID is generated for each chat session.  
+
+2. **Backend (FastAPI):**    
+   Receives user prompts, manages the session and in-memory chat history, and retrieves or creates an agent according to the environment setting.  
+
+3. **Agent (specified by AGENT_MODULE):**    
+   Processes the input using Azure OpenAI and optional MCP tools. The agent may operate in single, multi-agent, or collaborative modes, depending on configuration.  
+
+4. **Chat History:**    
+   Conversation history is stored per session and can be displayed in the frontend or reset as needed.  
+
+---  
+
+## FastAPI Endpoints  
+
+- `POST /chat`    
+  Send a JSON payload with `{ "session_id": ..., "prompt": ... }`. Returns the assistant’s response.  
+
+- `POST /reset_session`    
+  Send a payload `{ "session_id": ... }` to clear the conversation history for that session.  
+
+- `GET /history/{session_id}`    
+  Fetches all previous messages for a given session.  
+
+---  
+
+## Notes & Best Practices  
+
+- The current session store uses an in-memory Python dictionary; for production deployments, substitute this with a persistent store such as Redis or a database.  
+- Ensure secrets in your `.env` file (like API keys) are never committed to version control.  
+- The MCP server and Azure endpoint URLs must be accessible from the backend.  
+- To experiment with different agent behaviors, adjust the `AGENT_MODULE` in `.env`.  
+
+---  
+
+## Credits  
+
+- **Microsoft Azure OpenAI Service**  
+- **MCP Project**  
+- **AutoGen**  
+
+
+---    
+## Acknowledgments  
+
+- Microsoft Azure OpenAI Service    
+- MCP Project    
+- AutoGen
+- SDP CSA & SE Team - James Nguyen, Anil Dwarkanath, Nicole Serafino, Claire Rehfuss, Patrick O'Malley, Kirby Repko, Heena Ugale, Aditya Agrawal    

agentic_ai/applications/.env.sample

@@ -21,7 +21,7 @@ OPENAI_MODEL_NAME="gpt-4.1-2025-04-14"  #this is need for Autogen setting only.
 BACKEND_URL="http://localhost:7000"  
 # If MCP is fronted by APIM, point to the APIM gateway SSE endpoint  
 MCP_SERVER_URI="https://YOUR-APIM-GATEWAY.azure-api.net/mcp/sse"  
-# For local (no APIM) dev, use: http://localhost:8000/sse  
+# For local (no APIM) dev, use: http://localhost:8000/mcp  
 
 ############################################  
 #              APIM / AAD settings         #  

agentic_ai/applications/.python-version

@@ -0,0 +1 @@
+3.12

agentic_ai/applications/pyproject.toml

@@ -0,0 +1,31 @@
+[project]
+name = "applications"
+version = "0.1.0"
+description = "Add your description here"
+readme = "README.md"
+requires-python = ">=3.12"
+dependencies = [
+    "a2a-sdk>=0.3.7",
+    "autogen-agentchat==0.7.1",
+    "autogen-ext[mcp]==0.7.1",
+    "azure-cosmos==4.9.0",
+    "chainlit==1.3.2",
+    "faker==26.0.0",
+    "fastapi==0.115.12",
+    "fastmcp==2.7.1",
+    "flasgger==0.9.7.1",
+    "flask==3.0.3",
+    "httpx==0.28.1",
+    "mcp==1.12.2",
+    "msal==1.31.0",
+    "openai==1.77.0",
+    "pydantic==2.11.4",
+    "pyjwt[crypto]==2.9.0",
+    "python-dotenv==1.1.0",
+    "requests==2.32.3",
+    "semantic-kernel==1.35.0",
+    "streamlit==1.45.0",
+    "tenacity==8.5.0",
+    "uvicorn>=0.25.0",
+    "websockets>=15.0.1",
+]

agentic_ai/applications/run_applications_uv.sh

@@ -0,0 +1,40 @@
+#!/bin/bash
+# Determine project root
+PROJECT_ROOT=$(git rev-parse --show-toplevel 2>/dev/null || dirname "$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)")
+
+# Set paths relative to project root
+MCP_DIR="$PROJECT_ROOT/mcp"
+APP_DIR="$PROJECT_ROOT/agentic_ai/applications"
+OUTPUT_LOG="$PROJECT_ROOT/logs/output.log"
+ERROR_LOG="$PROJECT_ROOT/logs/error.log"
+
+mkdir -p "$PROJECT_ROOT/logs"
+
+# Function to add timestamps and service prefixes
+log_with_prefix() {
+    local service_name="$1"
+    awk -v service="$service_name" '{print "[" strftime("%Y-%m-%d %H:%M:%S") "] [" service "] " $0; fflush()}'
+}
+
+# Start MCP server
+echo "Starting MCP server..."
+cd "$MCP_DIR"
+uv run mcp_service.py 2>&1 | log_with_prefix "MCP" >> $OUTPUT_LOG &
+MCP_PID=$!
+
+sleep 5
+
+# Start backend
+echo "Starting backend..."
+cd "$APP_DIR"
+uv run backend.py 2>&1 | log_with_prefix "BACKEND" >> $OUTPUT_LOG &
+BACKEND_PID=$!
+
+sleep 5
+uv run streamlit run frontend.py 2>&1 | log_with_prefix "FRONTEND" >> $OUTPUT_LOG &
+FRONTEND_PID=$!
+
+echo "MCP Server PID: $MCP_PID"
+echo "Backend PID: $BACKEND_PID"
+echo "Frontend PID: $FRONTEND_PID"
+echo "Project root: $PROJECT_ROOT"