[Issue] Quickstart: Deploy your first hosted agent - Microsoft Foundry | Microsoft Learn - errors when running app locally

[puicchan]

azd version 1.23.6 (commit 16c72e7)

Walking through Quickstart: Deploy your first hosted agent - Microsoft Foundry | Microsoft Learn

Step - Running the Agent Locally

1. Running python ./src/af-agent-with-foundry-tools/main.py locally crashed immediately with:

   AttributeError: type object 'SpanAttributes' has no attribute 'LLM_REQUEST_MODEL'

   Root Cause: Version incompatibility between pre-release packages. The quickstart's requirements.txt pinned azure-ai-agentserver-agentframework==1.0.0b12, which pulled in:

   opentelemetry-semantic-conventions-ai==0.4.14 — removed LLM_REQUEST_MODEL from SpanAttributes
   azure-ai-projects==2.0.0b4 — renamed PromptAgentDefinitionText (caused a second import error after fixing the first)
   Both broke imports in agent-framework-core==1.0.0b260107.

   Fix: Upgrade the package pin in requirements.txt:
   azure-ai-agentserver-agentframework==1.0.0b14

2. web_search_preview tool caused BadRequest — The agent code included {"type": "web_search_preview"} in the tools list, but no Bing Grounding resource was provisioned (all BING_GROUNDING_* env vars were empty). Foundry rejected the request. Fix: removed web_search_preview from the tools array.
   Fix: changed tools=[{"type": "web_search_preview"}] to tools=[] in main.py.

[puicchan]

cc: @kristenwomack @therealjohn