Chore/remove unsed agents

.github/codeql/typescript-config.yml

@@ -1,8 +1,6 @@
 name: TypeScript CodeQL Configuration
 
 paths:
-  - packages/tools/code-evaluation/**
-  - packages/tools/web-search/**
   - packages/api/**
   - packages/ui/**
 

.github/workflows/codeql-typescript.yml

@@ -8,10 +8,6 @@ on:
       - 'packages/api/*.js'
       - 'packages/ui/*.ts'
       - 'packages/ui/*.js'
-      - 'packages/tools/code-evaluation/**/*.ts'
-      - 'packages/tools/code-evaluation/**/*.js'
-      - 'packages/tools/web-search/**/*.ts'
-      - 'packages/tools/web-search/**/*.js'
       - 'packages/tools/echo-ping/**/*.ts'
       - 'packages/tools/echo-ping/**/*.js'
       - 'src/shared/**/*.ts'
@@ -25,10 +21,6 @@ on:
       - 'packages/api/*.js'
       - 'packages/ui/*.ts'
       - 'packages/ui/*.js'
-      - 'packages/tools/code-evaluation/**/*.ts'
-      - 'packages/tools/code-evaluation/**/*.js'
-      - 'packages/tools/web-search/**/*.ts'
-      - 'packages/tools/web-search/**/*.js'
       - 'packages/tools/echo-ping/**/*.ts'
       - 'packages/tools/echo-ping/**/*.js'
       - 'src/shared/**/*.ts'
@@ -59,7 +51,7 @@ jobs:
 
       - name: Install dependencies
         run: |
-          for dir in packages/tools/code-evaluation packages/tools/web-search packages/tools/echo-ping src/api src/ui; do
+          for dir in packages/tools/echo-ping src/api src/ui; do
             if [ -f "$dir/package.json" ]; then
               cd $dir && npm install && cd -
             fi

.vscode/mcp.json

@@ -13,18 +13,6 @@
       "type": "http",
       "url": "http://localhost:5007/mcp"
     },
-    "tool-web-search": {
-      "type": "sse",
-      "url": "http://localhost:5006/sse"
-    },
-    "tool-model-inference": {
-      "type": "sse",
-      "url": "http://localhost:5005/sse"
-    },
-    "tool-code-evaluation": {
-      "type": "sse",
-      "url": "http://localhost:5004/sse"
-    },
     "tool-itinerary-planning": {
       "type": "http",
       "url": "http://localhost:5003/mcp"

AGENTS.md

@@ -29,10 +29,8 @@ All orchestrators communicate with the same MCP tool servers. Each component is
 │   │   ├── echo-ping/          # TypeScript/Node.js (testing tool)
 │   │   ├── customer-query/     # C#/.NET (customer inquiry processing)
 │   │   ├── destination-recommendation/  # Java (travel destination suggestions)
-│   │   ├── itinerary-planning/ # Python (detailed itinerary creation)
-│   │   ├── code-evaluation/    # Python (code interpreter integration)
-│   │   ├── model-inference/    # Python (local LLM with GPU support)
-│   │   └── web-search/         # TypeScript (Bing API integration)
+│   │   └── itinerary-planning/ # Python (detailed itinerary creation)
+│   │   
 │   ├── shared/                 # Common utilities and types
 │   └── docker-compose.yml      # Local development environment
 ├── azure.yaml                  # Azure Developer CLI configuration
@@ -60,9 +58,6 @@ The system implements specialized agents coordinated by orchestration layers. Al
   - **Customer Query Agent**: Extracts preferences from customer inquiries (via customer-query MCP)
   - **Destination Recommendation Agent**: Suggests destinations (via destination-recommendation MCP)
   - **Itinerary Planning Agent**: Creates detailed travel plans (via itinerary-planning MCP)
-  - **Web Search Agent**: Fetches live travel data via Bing API (via web-search MCP)
-  - **Code Evaluation Agent**: Executes custom logic (via code-evaluation MCP)
-  - **Model Inference Agent**: Runs local LLMs with ONNX/vLLM (via model-inference MCP)
   - **Echo Agent**: Testing and validation (via echo-ping MCP)
 
 **Option 2: LlamaIndex.TS Orchestration** (`packages/api/src/orchestrator/llamaindex/`)
@@ -74,9 +69,6 @@ The system implements specialized agents coordinated by orchestration layers. Al
   - **Customer Query Agent**: Extracts preferences from customer inquiries (via customer-query MCP)
   - **Destination Recommendation Agent**: Suggests destinations (via destination-recommendation MCP)
   - **Itinerary Planning Agent**: Creates detailed travel plans (via itinerary-planning MCP)
-  - **Web Search Agent**: Fetches live travel data via Bing API (via web-search MCP)
-  - **Code Evaluation Agent**: Executes custom logic (via code-evaluation MCP)
-  - **Model Inference Agent**: Runs local LLMs with ONNX/vLLM (via model-inference MCP)
   - **Echo Agent**: Testing and validation (via echo-ping MCP)
 
 **Option 3: Microsoft Agent Framework Orchestration** (`packages/api-python/`)
@@ -89,9 +81,6 @@ The system implements specialized agents coordinated by orchestration layers. Al
   - **CustomerQueryAgent**: Processes customer inquiries with MCP tools
   - **DestinationRecommendationAgent**: Provides destination suggestions
   - **ItineraryPlanningAgent**: Creates detailed itineraries with MCP tools
-  - **CodeEvaluationAgent**: Executes code via code-evaluation MCP server
-  - **ModelInferenceAgent**: Performs local model inference via MCP
-  - **WebSearchAgent**: Searches web using Bing API via MCP
   - **EchoAgent**: Testing and validation via echo-ping MCP
 
 #### MCP Tool Servers (Shared by All Orchestrations)
@@ -101,9 +90,6 @@ All three orchestration implementations communicate with these MCP servers:
 - **customer-query** (.NET/C#) - Port 5001 - Customer inquiry processing
 - **destination-recommendation** (Java) - Port 5002 - Travel destination suggestions
 - **itinerary-planning** (Python) - Port 5003 - Detailed itinerary creation
-- **code-evaluation** (Python) - Port 5004 - Code interpreter integration
-- **model-inference** (Python) - Port 5005 - Local LLM with GPU support
-- **web-search** (TypeScript) - Port 5006 - Bing API integration
 - **echo-ping** (TypeScript) - Port 5007 - Testing and validation
 
 ### Service Communication
@@ -234,7 +220,7 @@ For detailed comparison, see `docs/orchestration.md`.
 
 ### TypeScript/Node.js Standards
 
-**Location**: `packages/api/`, `packages/ui/`, `packages/tools/echo-ping/`, `packages/tools/web-search/`
+**Location**: `packages/api/`, `packages/ui/`, `packages/tools/echo-ping/`
 
 **Key Conventions**:
 - Use ES modules (`"type": "module"` in package.json)
@@ -529,7 +515,6 @@ chore(deps): update dependencies
    cd packages/tools/itinerary-planning && pip install .
    cd packages/tools/code-evaluation && pip install .
    cd packages/tools/model-inference && pip install .
-   cd packages/tools/web-search && npm run build
    ```
 
 2. **Test Coverage**: Run existing tests and add new tests for new functionality

docs/advanced-setup.md

@@ -203,12 +203,6 @@ The application uses environment variables to configure the services. You can se
     - `packages/tools/destination-recommendation/.env.docker`
     - `packages/tools/itinerary-planning/.env`
     - `packages/tools/itinerary-planning/.env.docker`
-    - `packages/tools/code-evaluation/.env`
-    - `packages/tools/code-evaluation/.env.docker`
-    - `packages/tools/model-inference/.env`
-    - `packages/tools/model-inference/.env.docker`
-    - `packages/tools/web-search/.env`
-    - `packages/tools/web-search/.env.docker`
     - `packages/tools/echo-ping/.env`
     - `packages/tools/echo-ping/.env.docker`
 

docs/deployment-architecture.md

@@ -41,7 +41,6 @@ graph TB
                 EP[Echo Ping]
                 CQ[Cust Query]
                 DR[Dest Rec]
-                WS[Web Search]
                 OTHER[...]
             end
         end
@@ -136,18 +135,11 @@ AZURE_OPENAI_ENDPOINT=https://your-openai.openai.azure.com/
 AZURE_OPENAI_API_KEY=your-api-key
 AZURE_OPENAI_DEPLOYMENT_NAME=gpt-4o-mini
 
-# Bing Search
-BING_SEARCH_API_KEY=your-bing-key
-BING_SEARCH_ENDPOINT=https://api.bing.microsoft.com/
-
 # MCP Server URLs (local development)
 MCP_ECHO_PING_URL=http://localhost:5007
 MCP_CUSTOMER_QUERY_URL=http://localhost:5001
 MCP_DESTINATION_RECOMMENDATION_URL=http://localhost:5002
 MCP_ITINERARY_PLANNING_URL=http://localhost:5003
-MCP_CODE_EVALUATION_URL=http://localhost:5004
-MCP_MODEL_INFERENCE_URL=http://localhost:5005
-MCP_WEB_SEARCH_URL=http://localhost:5006
 
 # Monitoring
 OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:18889
@@ -219,9 +211,6 @@ services:
   tool-customer-query:
   tool-destination-recommendation:
   tool-itinerary-planning:
-  tool-code-evaluation:
-  tool-model-inference:
-  tool-web-search:
   web-api:            # Express API server
   web-ui:             # Angular UI
 ```
@@ -243,14 +232,10 @@ MCP_ECHO_PING_URL=http://tool-echo-ping:3000
 MCP_CUSTOMER_QUERY_URL=http://tool-customer-query:8080
 MCP_DESTINATION_RECOMMENDATION_URL=http://tool-destination-recommendation:8080
 MCP_ITINERARY_PLANNING_URL=http://tool-itinerary-planning:8000
-MCP_CODE_EVALUATION_URL=http://tool-code-evaluation:5000
-MCP_MODEL_INFERENCE_URL=http://tool-model-inference:5000
-MCP_WEB_SEARCH_URL=http://tool-web-search:5000
 
 # External services (from azd provision)
 AZURE_OPENAI_ENDPOINT=https://your-openai.openai.azure.com/
 AZURE_OPENAI_API_KEY=your-api-key
-BING_SEARCH_API_KEY=your-bing-key
 
 # Docker-specific settings
 IS_LOCAL_DOCKER_ENV=true
@@ -545,11 +530,6 @@ resource apiApp 'Microsoft.App/containerApps@2023-05-01' = {
           keyVaultUrl: '${keyVault.properties.vaultUri}secrets/azure-openai-key'
           identity: managedIdentity.id
         }
-        {
-          name: 'bing-search-key'
-          keyVaultUrl: '${keyVault.properties.vaultUri}secrets/bing-search-key'
-          identity: managedIdentity.id
-        }
       ]
     }
     template: {
@@ -570,10 +550,6 @@ resource apiApp 'Microsoft.App/containerApps@2023-05-01' = {
               name: 'AZURE_OPENAI_API_KEY'
               secretRef: 'azure-openai-key'
             }
-            {
-              name: 'BING_SEARCH_API_KEY'
-              secretRef: 'bing-search-key'
-            }
             // MCP server URLs
             {
               name: 'MCP_ECHO_PING_URL'
@@ -942,7 +918,7 @@ az acr build \
   packages/api
 
 # MCP Server Images
-for server in echo-ping customer-query destination-recommendation itinerary-planning code-evaluation model-inference web-search; do
+for server in echo-ping customer-query destination-recommendation itinerary-planning; do
   echo "Building MCP server: $server"
   az acr build \
     --registry $ACR_NAME \

docs/development-guide.md

@@ -273,10 +273,7 @@ packages/tools/
 │   ├── AITravelAgent.sln       # Solution file
 │   └── Dockerfile
 ├── destination-recommendation/ # Java/Spring Boot
-├── itinerary-planning/         # Python/FastAPI
-├── code-evaluation/            # Python
-├── model-inference/            # Python with ONNX/vLLM
-└── web-search/                 # TypeScript/Node.js
+└── itinerary-planning/         # Python/FastAPI
 ```
 
 ## Development Workflow
@@ -568,9 +565,6 @@ tool-my-new-server:
 export type McpServerName =
   | "echo-ping"
   | "customer-query"
-  | "web-search"
-  | "itinerary-planning"
-  | "model-inference"
   | "code-evaluation"
   | "destination-recommendation"
   | "my-new-server";  // Add new server
@@ -997,7 +991,6 @@ test.describe('Travel Planning User Journey', () => {
 
     // Select tools
     await page.check('[data-testid="tool-echo-ping"]');
-    await page.check('[data-testid="tool-web-search"]');
 
     // Verify selection
     const newSelectedTools = await page.$$('[data-testid^="tool-"]:checked');

docs/flow-diagrams.md

@@ -69,17 +69,14 @@ sequenceDiagram
     participant U as User
     participant UI as Angular UI
     participant API as Express API
-    participant LI as LlamaIndex
+    participant LI as (LlamaIndex, Lanchain.js, or MS Agent Framework)
     participant TA as Triage Agent
     participant CQA as Customer Query Agent
     participant DRA as Destination Rec Agent
     participant IPA as Itinerary Planning Agent
-    participant WSA as Web Search Agent
     participant MCPCQ as MCP Customer Query
     participant MCPDR as MCP Destination Rec
     participant MCPIP as MCP Itinerary Plan
-    participant MCPWS as MCP Web Search
-    participant BING as Bing Search API
     participant AOAI as Azure OpenAI
 
     U->>UI: Enter travel query "Plan 7-day Japan trip"
@@ -93,7 +90,6 @@ sequenceDiagram
     LI->>CQA: Initialize Customer Query Agent  
     LI->>DRA: Initialize Destination Rec Agent
     LI->>IPA: Initialize Itinerary Planning Agent
-    LI->>WSA: Initialize Web Search Agent
 
     API->>TA: process("Plan 7-day Japan trip")
     API-->>UI: SSE: AgentSetup event
@@ -117,14 +113,6 @@ sequenceDiagram
     DRA-->>TA: return recommendations
     API-->>UI: SSE: AgentStream event (partial results)
 
-    TA->>WSA: handoff(search current travel info)
-    WSA->>MCPWS: callTool("search", "Japan travel 2024")
-    MCPWS->>BING: Search API request
-    BING-->>MCPWS: Current travel information
-    MCPWS-->>WSA: Processed search results
-    WSA-->>TA: return current info
-    API-->>UI: SSE: AgentStream event
-
     TA->>IPA: handoff(create itinerary, destinations + preferences)
     IPA->>MCPIP: callTool("plan_itinerary", full_context)
     MCPIP->>AOAI: Generate detailed itinerary
@@ -313,21 +301,12 @@ graph TD
 
     C -->|Destination| D[Destination Recommendation]
     C -->|Planning| E[Itinerary Planning]  
-    C -->|Information| F[Web Search]
-    C -->|Complex Logic| G[Code Evaluation]
-    C -->|Custom Model| H[Model Inference]
 
     D --> I[MCP Destination Server]
     E --> J[MCP Itinerary Server]
-    F --> K[MCP Web Search Server]
-    G --> L[MCP Code Eval Server]
-    H --> M[MCP Model Server]
 
     I --> N[Response Synthesis]
     J --> N
-    K --> N
-    L --> N
-    M --> N
 
     N --> O[Final Response]
 ```

docs/maf-implementation-guide.md

@@ -188,9 +188,6 @@ class Settings(BaseSettings):
     mcp_customer_query_url: str
     mcp_destination_recommendation_url: str
     mcp_itinerary_planning_url: str
-    mcp_code_evaluation_url: str
-    mcp_model_inference_url: str
-    mcp_web_search_url: str
     mcp_echo_ping_url: str
     mcp_echo_ping_access_token: Optional[str] = None
 
@@ -294,21 +291,6 @@ class ToolRegistry:
             settings.mcp_itinerary_planning_url
         )
 
-        # Code Evaluation
-        self.clients["code-evaluation"] = HTTPMCPClient(
-            settings.mcp_code_evaluation_url
-        )
-
-        # Model Inference
-        self.clients["model-inference"] = HTTPMCPClient(
-            settings.mcp_model_inference_url
-        )
-
-        # Web Search
-        self.clients["web-search"] = HTTPMCPClient(
-            settings.mcp_web_search_url
-        )
-
         # Echo Ping
         self.clients["echo-ping"] = HTTPMCPClient(
             settings.mcp_echo_ping_url,
@@ -385,9 +367,6 @@ class TriageAgent(BaseAgent):
         - CustomerQueryAgent: For understanding customer preferences
         - DestinationRecommendationAgent: For suggesting destinations
         - ItineraryPlanningAgent: For creating travel itineraries
-        - WebSearchAgent: For finding current travel information
-        - CodeEvaluationAgent: For calculations and data processing
-        - ModelInferenceAgent: For specialized AI tasks
 
         Always choose the most appropriate agent(s) for the query.
         You may coordinate multiple agents for complex queries.

docs/maf-migration-plan.md

@@ -32,10 +32,7 @@ This document outlines the migration strategy from the current LlamaIndex.TS orc
 2. CustomerQueryAgent
 3. DestinationRecommendationAgent
 4. ItineraryPlanningAgent
-5. CodeEvaluationAgent
-6. ModelInferenceAgent
-7. WebSearchAgent
-8. EchoAgent
+5. EchoAgent
 
 **Endpoints**:
 - `GET /api/health` - Health check
@@ -113,9 +110,6 @@ We will use a parallel deployment strategy to minimize risk:
 - [ ] Implement CustomerQueryAgent
 - [ ] Implement DestinationRecommendationAgent
 - [ ] Implement ItineraryPlanningAgent
-- [ ] Implement CodeEvaluationAgent
-- [ ] Implement ModelInferenceAgent
-- [ ] Implement WebSearchAgent
 - [ ] Implement EchoAgent
 - [ ] Create tool wrappers for MCP calls