[META] Support agentic search use cases

[ohltyler]

Agentic search is a growing trend of AI-based search flows leveraging agents for a variety of individual use cases, including natural language query (NLQ) understanding, query optimization, and arbitrarily-advanced RAG use cases, including multi-search flows, or leveraging MCP clients for calling different external MCP servers/tools.

This plugin should be a home for letting users leverage these new agentic search use cases as they become available as backend features.

As of Sept 2025, the scope is focused on NLQ / query understanding and optimization. Even more specifically, it allows for agents of type flow or conversational, with a QueryPlanningTooland few other optional tools, likeListIndexTool/SearchIndexTool`, if more actions are needed. This plugin should have a preset use case for supporting this.

High-level roadmap:

• (3.3) Support NLQ use-case (flow agents with a QueryPlanningTool)
• (3.4+ 3.3) Search templates integration
• (3.4+ 3.3) Support more agent types (conversational, plan-execute-reflect) and their respective configurations
• (3.4+ 3.3) Support more tools (SearchIndexTool, IndexMappingsTool, WebSearchTool, etc.)
• (3.4+) [Agentic Search] Improve export / next steps UX #804
• (3.4+) [Agentic Search] Support agent deletion #785
• {3.4+) [Agentic Search] Improve error handling #784
• (3.4+) [Agentic Search] Improve agent form field validation #776
• (3.4+) [Agentic Search] Support custom search pipelines #778
• (3.4+) [META] [Agentic Search] Improve integration with models/connectors #779
• (3.4+) [Agentic Search] Improve vector search UX #780
• (3.4+) [Agentic Search] Improve MCP UX #781
• (3.4+) [Agentic Search] Improve integration with memory / tracing #782
• (3.5+) [Agentic Search] Support index creation / ingestion #777