Agentic RAG on Kubeflow — Multi-Index Retrieval with Kagent & MCP

Kubeflow's documentation, GitHub issues, and platform code are spread across dozens of repositories with no unified search. This project evolves kubeflow/docs-agent from a single-tool retrieval script into a multi-index Agentic RAG architecture with three MCP search tools (docs, issues, code), domain-specific ingestion pipelines (KFP), and a "Thin Context" MCP flow for IDE integration. All orchestration uses Kagent CRDs (Kubernetes-native). Deliverables: (1) Three MCP tools with dedicated Milvus collections and specialized chunking — issues at comment boundaries, code at YAML/AST boundaries; (2) Project's first test suite (71+ tests) with CI; (3) Idempotent pipeline upserts replacing the current destructive drop-and-recreate; (4) Developer IDE configs for Cursor/Claude Desktop; (5) Feedback logging for golden dataset accumulation; (6) RAGAS evaluation pipeline. Core work (PRs #140, #143) is already built and live-tested on OCI with 2,108 real chunks indexed.