Evaluate Prompt Refinement Options

As a User I need an in-build initial good version of the prompts to "just run" the framework, but in the long run we need a way to iterate prompts easily to get closer to the desired outcomes. Options we see currently:
- refine prompts with AI
and/or
- https://github.com/microsoft/agentrc

No files, tests, or entry points are identified in the issue. Compare the two proposed approaches—AI prompt refinement and Microsoft AgentRC—and document a recommended direction, including how prompt iteration would support the initial run and longer-term outcome improvement.

Written by the indexing model from the issue text.