[BYLinMou]

This pull request introduces a new AI-powered recommendations feature to the Discovery Agent, enhancing both backend and frontend to provide curated post recommendations alongside query results. The backend now generates and returns AI-curated recommendations and analysis source metadata, while the frontend displays these recommendations in a new styled component. The API contracts and UI have been updated to support and present this richer output.

Backend enhancements:

• Added an asynchronous curate_agent_results function to generate AI-based recommendations and analysis metadata, and integrated it into the /agent/execute endpoint. The response now includes a recommendations list and analysis_source string, falling back to the previous summarization if unavailable. [1] [2]
• Updated the AgentExecuteRequest and AgentExecuteResponse schemas to support the new prompt, recommendations, and analysis_source fields, and introduced a new AgentPostRecommendation model. [1] [2]

Frontend updates:

• Modified the executeAgentQuery API function to accept both sql and prompt, matching the backend changes.
• Updated the DiscoveryAgent component to display AI-curated recommendations using a new RecommendationCard subcomponent, and to show the analysis source in the UI. [1] [2] [3] [4]
• Added new CSS styles for the recommendations section and cards, improving the visual presentation of curated picks.

Other:

• Bumped the project version to 0.10.3-alpha.1 to reflect these new features.