Merge pull request microsoft#428 from microsoft/macae-v3-dev-marktayl

feat: Macae v3 dev marktayl

Author: Fr4nc3

data/agent_teams/hr.json

@@ -1,7 +1,7 @@
 {
   "id": "3",
   "team_id": "team-3",
-  "name": "Retail Customer Success Team",
+  "name": "Human Resources Team",
   "status": "visible",
   "created": "",
   "created_by": "",
@@ -43,7 +43,7 @@
       "input_key": "",
       "type": "",
       "name": "ProxyAgent",
-      "deployment_name": "",
+      "deployment_name": "gpt-4.1",
       "icon": "",
       "system_message": "",
       "description": "",

data/agent_teams/new 29.txt

@@ -0,0 +1,48 @@
+Tasks:
+
+Done: Make 3 teams upload work to cosmos (HR, Marketing, Retail).  We will load this Cosmos data on deploy as default teams.
+Done: - call "/socket/{process_id}" (start_comms) to setup websocket
+Done: Make sure that a team is always selected - start with the hr.json team
+
+call init_team API for the currently loaded team on App start - 
+Spinner / team-loading should display until this call returns (user should not be able to input tasks)
+<takes about 20 seconds> - say something like "team loading" with spinner
+FE: send unload current team API and call init team for team switch - sending select_team(new team id)
+	spin while waiting for return of API
+BE: unload old team - load new team - return status
+
+BE: For Francia - implement get_plans to fill in history from cosmos
+
+BE: Create a teams container in Cosmos and move all loaded team definitions there
+
+Implement saving of plan to cosmos -> history in...
+
+================ Request submit flow ======================
+on request submission call "/create_plan" (process_request)
+This will return immediately - move to other page and display spinner -> "creating plan"
+Socket will start receiving messages ->
+Stream plan output into main window
+
+Will receive the PlanApprovalRequest message
+	Enable accept / reject UI
+Send PlanApprovalResponse message when user answers
+
+If not approved 
+	BE: plan will cancel on backend 
+	FE: - enable input again for fresh request
+	Call input_request API on backend again (just like inputing any request)
+
+If approved:
+	Display plan steps in right pane if approved
+=============================================================
+
+================== Message Streaming ========================	
+Process socket message routing to display agent output
+	See message types in src\backend\v3\models\messages.py
+	for each message from agent - process stream then rollup
+
+On FinalResultMessage
+	display final result with all agent output in rollups by agent above
+==============================================================
+
+

data/datasets/Competitor_Pricing_Analysis.csv renamed to data/datasets/competitor_pricing_analysis.csv

@@ -0,0 +1,34 @@
+Capturing the notes from auth install before deleting for docs...
+
+### Auth section:
+Requires and app registration as in azure_app_service_auth_setup.md so not deployed by default.
+
+To setup basic auth with FastMCP - bearer token - you can integrate with Azure by using it as your token provider.
+
+``` from fastmcp.server.auth import JWTVerifier```
+
+```
+auth = JWTVerifier(
+    jwks_uri="https://login.microsoftonline.com/52b39610-0746-4c25-a83d-d4f89fadedfe/discovery/v2.0/keys",
+    #issuer="https://login.microsoftonline.com/52b39610-0746-4c25-a83d-d4f89fadedfe/v2.0",
+    # This issuer is not correct in the docs. Found by decoding the token.
+    issuer="https://sts.windows.net/52b39610-0746-4c25-a83d-d4f89fadedfe/",
+    algorithm="RS256",
+    audience="api://7a95e70b-062e-4cd3-a88c-603fc70e1c73"
+)
+```
+
+Requires env vars:
+```
+export MICROSOFT_CLIENT_ID="your-client-id"
+export MICROSOFT_CLIENT_SECRET="your-client-secret"
+export MICROSOFT_TENANT="common" # Or your tenant ID
+```
+
+```mcp = FastMCP("My MCP Server", auth=auth)```
+
+For more complex and production - supports OAuth and PKCE
+
+Enabled through MCP enabled base - see lifecycle.py
+
+