Description of problem:

Our new agentic sample https://github.com/redhat-ai-dev/llama-stack-agentic-sample should handle persistent errors with or unavailability of the github mcp server, halt polling, and provide a message saying the service is unavailable.

It currently just continues to poll.

We handle things correctly when the k8s mcp server is unavailable

rh-ee-tpetkos mfaisal2 and I have previously discussed this voice to voice and in slack.

Prerequisites (if any, like setup, operators/versions):

Steps to Reproduce

 # unless the service is unavailable, sabatoge its configuration to induced errors 

Actual results:

Expected results:

Reproducibility (Always/Intermittent/Only Once):

Build Details:

Additional info (Such as Logs, Screenshots, etc):